The problem was an optimization I put in to triangulate quads.
It was wrong if the quad, after projecting onto a 2d plane, was
not convex. Handling quads the same as other faces fixes the bug.
Unfortunately, this will slow down Exact Boolean when the input has
many quads (the usual case, of course).
Will attempt to fix that with a later change, but for now, this
at least restores correctness.
Now, the mask layers are copied and later a cleanup is done in order to verify all mask layer exist in destination object. If the layer mask does not exist, it's removed from the list.
This is related to T89234.
Skip updating normals & tessellation for contiguous geometry regions
for operations such as translate & uniform scale.
This means when all geometry is selected, no updates are needed
as the relative locations of vertices aren't being modified.
Performance:
As this is skipping a multi-threaded operation,
larger improvements are noticeable on systems with fewer cores.
- ~1.15x to ~1.3x overall gain for 32 cores.
- ~1.7x to ~2.2x overall gain for 1 core (limited using `-t 1` argument).
Details:
- Rotate & non-uniform scale only skip tessellation.
- Proportional editing and axis-mirror have special handling
ensure geometry is properly grouped before considering
a face part of a single group that can be skipped.
- Loose vertices always need their normals to be recalculated
since they're calculated based on the location.
- Non-affine transform operations such as shrink-fatten & bend,
don't take advantage of this optimization.
- Snap projection also disables the optimization.
editor dataset region always showed 0. This was caused by a conditional
statement that needed a domain to be set, which is not the case for
Instances component type.
Reviewer: Hans Goudey (Hoogly Boogly)
Differential Revision: https://developer.blender.org/D11710
This patch enables sample filtering when scaling preview images in File
Browser, improving the result a bit. Reduces blockiness and other
artifacts when enlarging the images.
see D11706 for details and examples.
Differential Revision: https://developer.blender.org/D11706
Reviewed by Julian Eisel
`t->values` does not necessarily represent a final value of the
transformation, as each mode treats this value differently.
So, unfortunately, we cannot have a generic offset solution for modal
transform operations. Offset needs to be handled by each mode.
Note: Currently only, `Move`, `Rotate` and `Resize` support this.
When baking some data, we create a new Mesh with edits and modifiers applied.
However, in some cases (e.g. when there is no modifier), the returned Mesh is
actually referencing the original one and its data layers. When autosmooth is
enabled we also split the Mesh. However, since the new Mesh is referencing the
original one, although `BKE_mesh_split_faces` is creating new vertices and edges,
the reallocation of the custom data layers is preempted because of the
reference, so adding the new vertices and edges overwrites valid data
To fix this we duplicate referenced layers before splitting the faces.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D11703
User can specify filtering options inside line art modifier,
like inverting selection and including face mark region border.
Reviewed By: Sebastian Parborg (zeddb)
Differential Revision: https://developer.blender.org/D11307
This patch adds a left aligned sidebar to the spreadsheet editor. This
Sidebar can be used to navigate the geometry component types and
attribute domains. It also provides a quick overview of domain sizes.
It replaces the two dropdowns in the regions header.
Next step will be to add the domain cycling shortcut
using the CTRL + mouse wheel.
Reviewer: Dalai Felinto (dfelinto), Julian Eisel (Severin),
Hans Goudey (HooglyBoogly).
Differential Revision: https://developer.blender.org/D11046
This patch includes: Floating edge type support,
Special chaining option for floating edge,
Chaining option for reducing jagged edges when floating
edges are involved.
Reviewed By: Sebastian Parborg (zeddb)
Differential Revision: https://developer.blender.org/D11306
* Mark either all or no class methods with override
* Don't use zero sized array since it has a different size in C and C++.
Using a little more memory here is not significant.
* Don't use deprecated mechanism to mark private GSet members in clang
just like we don't for MSVC, it warns even for simple zero initialization.
In-progress Safari download files/packages are now recognized as bundles
and therefore not treated as directories.
Differential Revision: https://developer.blender.org/D11613
This resolves a bottleneck where every update while transforming
copied the entire mesh data-block, which isn't needed as the edit-mesh
is the source of the data being edited.
Testing shows a significant overall speedup when transforming:
- ~1.5x with a subdivided cube 1.5 million vertices.
- ~3.0x with the spring mesh (edit-mode with modifiers disabled,
duplicated 10x to drop performance).
Reviewed By: sergey
Ref D11337
Avoid computationally expensive copying operations
when only some settings have been modified.
This is done by adding support for updating parameters
without tagging for copy-on-write.
Currently only mesh data blocks are supported,
other data-blocks can be added individually.
This prepares for changing values such as edit-mesh auto-smooth angle
in edit-mode without duplicating all mesh-data.
The benefit will only be seen when the user interface no longer tags
all ID's for copy on write updates.
ID_RECALC_GEOMETRY_ALL_MODES has been added to support situations
where non edit-mode geometry is modified in edit-mode.
While this isn't something user are likely to do,
Python scripts may change the underlying mesh.
Reviewed By: sergey
Ref D11377
Very stupid typo in override apply code on GP modifiers (typical
copy/paste mistake from original modifiers code).
@jbakker this should be back-ported to 2.93LTS.
This patch enables bound box check when loading geometry
into line art. Works with overscan as well. Will discard
object if its bbox completely lies in one side of the
clipping space frustum.
Reviewed by: Sebastian Parborg (zeddb)
Differential Revision: https://developer.blender.org/D11545
For scenes that have a lot of edges, this could potentially
save some time generating individual strokes that are outside
camera frustum.
Reviewed By: Sebastian Parborg (zeddb)
Differential Revision: https://developer.blender.org/D11525
682a74e090 renamed Bendy Bones properties and replaced existing float
properties with float-vector properties. This updates the property names used
by the action baking operator (`NLA_OT_bake`).
Iteration was incorrectly calling the same object. Had it called every
window it would still be unnecessary as WM_DISPLAYCHANGE is sent to all
broadcast receiving windows.
Previously the smooth shading of the voxel remesher was controlled by a
mesh property. With this change, the output will try to match the
current shading of the object. This only takes into consideration the
shading mode of the first polygon of the model, but it is probably what
most users expect as it works as intended with the shade smooth/flat
object mode options.
Reviewed By: JulienKaspar, JacquesLucke
Differential Revision: https://developer.blender.org/D11626
SCULPT_nearest_vertex_get expects a distance, not a distance squared.
This should make symmetry work as expected, but it still can fail if the
mesh topology is not completely symmetrical.
Reviewed By: JacquesLucke
Maniphest Tasks: T89221
Differential Revision: https://developer.blender.org/D11642
Adds full frame implementation to this node operations.
No functional changes.
1.2x faster than tiled fallback.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11638
Adds full frame implementation to Image node operations.
Mostly refactored into buffer utility methods for reuse in other
operations.
No functional changes.
1.8x faster than tiled fallback.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11559
When use smooth tool in a cyclic stroke, the smoothing algorithm was not using the adjacent points if these are the end or start of stroke.
Now, the smooth uses the other extreme of the stroke to compute smoothness.
This change will ensure at least one 'local' collection can host the new
'local' override of all objects (indirectly) overridden by this
operation, such that no new override of object ends up in master
collection (which can become extremely messy in production files).
In practice, it means often at least one of the linked collection owning
those objects also has to be overridden.
NOTE: This only affect cases where root overridden linked object has
some dependencies outside of its own root linked collection. While this
situation should be avoided, it cannot always be, so we try to support
it as best as we can.
This is again in the fuzzy area of how embedded IDs are handled
respectively by partial undo code and depsgraph... Should not be
necessary currently, but better be safe and explicit, and also tag
those embeded IDs from re-used owner ID.
In certain CMake configurations it was possible
that OCIO gave linker errors due to it thinking
it was using the shared library rather than the
static library we ship.
Would crash when adding or removing a collection directly to the master
collection of a scene.
Consequence of change to handle depsgraph-controlled evaluation of some
IDs (like excluded collections which do not get evaluated and do not get
a COW anymore). See rBcf4258673755 and D10907.
Note that this mostly demonstrates once again how weak and flacky our
handling of embedded IDs still remains, with some part of the code
handling them as independent IDs, some as fully local/private data, some
as a mix of both... and lots and lots of custom handling code and corner
cases that are a bottomless pit of issues.
Also quiet incredible that this was not reported already, luckily this
original change did not make it to 2.93 release.
The check on the transformation introduced in rBb313525c1bd0 to fix
T88515 would sometimes prevent to update the light if the Blender object
changed. To fix this, reverse the order in which the checks happen so
that we do not shortcuit the object change check.
Note that this commit uses a second LogRef (`blo.readfile.undo`) for undo
specific meassages. Allows to use `--log "*undo*"` cli option to match
all undo reporting.
Also did some minor tweaks to some reports on the way.
Those were missing proper (explicit) object cache clear, and DEG
tagging.
Note that this was most likely not an common issue in practice (Collection
object cache clearing recursively goes into all parents, so master
collection would only miss it in case they had no child collections at
all, and tagging of those happens almost always at other steps on
remapping). But better to be explicit and consistent here in any case.
Fix T88808.
Caused by {rB5f2c5e5bb8c15bf0d6679351e3482f9c38c00935}
object type for `TEXT object` was missing in following check
that's why `Set Origin` option was lost from object context menu.
Reviewed By: lichtwerk
Maniphest Tasks: T88808
Differential Revision: https://developer.blender.org/D11495
This reverts commit rB3a48147b8ab92, and fixes the issues with linking
etc.
Change compared to previous buggy commit (rBf8d219dfd4c31) is that
new `BlendFileReadReports` reports are now passed to the lowest level
function generating the `FileData` (`filedata_new()`), which ensures
(and asserts) that all code using it does have a valid non-NULL pointer
to a `BlendFileReadReport` data.
Sorry for the noise, it's always when you think a change is trivial and
do not test it well enough that you end up doing those kind of
mistakes...
Fix by reverting the part of ec30cf0b74
that assigned `but->editval` in `ui_numedit_begin_set_values`.
Causing access freed memory when using tab to switch
to a numeric input and then leaving the textbox by clicking outside.
This was because `ui_numedit_begin_set_values` shouldn't need to set
`but->editval` and overwrite the pointer.
This would set a pointer that had previously been freed,
causing a `NULL` check to fail later on.
Ref D11679
When cut a stroke of 1 point, the clean up done to avoid keep 1 point strokes removes the memory, but the pointer to the first stroke was not set to NULL. As this pointer is invalid now, any use of this produces a segment fault because the pointer is corrupted..
Sphinx has rather loose dependency requirements which can cause issues if we aren't careful.
As a solution they recommend that you pin sphinx dependency versions
Currently the emboss is only fading on left side of the widget,
resulting in the emboss extending vertically on the right side
and ending abruptly. This patch fixes this by also fading the
emboss on the right side and making it symmetric.
Differential Revision: https://developer.blender.org/D10810
This fixes T88346.
The code is also more readable by making a better distinction between
the texts used for Distances, "Proportional Size" and "AutoIK-Len".
And the text used to translate the "Proportional Size" is reused.
This patch fixes many minor spelling mistakes, all in comments or
console output. Mostly contractions like can't, won't, don't, its/it's,
etc.
Differential Revision: https://developer.blender.org/D11663
Reviewed by Harley Acheson
This problem has surprisingly been there for quite a few months.
For point clouds all attributes were handled the same, even "position",
which should be transformed when combining source points into the
destination.
Previously the code assumed that curve instances had no attributes.
This is true when the data came from curve objects, which don't support
attributes currently, but it isn't necessarily true when retrieving curves
from evaluated geometry sets.
Make the virtual functions protected and simpler, so that the logic is
better contained in the base class's implementation. Also introduce a
`copy_without_attributes` method to be used for realizing instances.
Add direct user feedback (as a warning report) to user when recursive
resync of overrides was needed.
And some timing (as CLOG logs) about main readfile process steps.
This is essentially adding a new BlendFileReadReport structure that wraps
BKE_reports list, and adds some extra info (some timing, some info about
overrides and (recursive) resync, etc.).
Allows to decompose a given amount of seconds into a random set of
days/hours/minutes/seconds/milliseconds values.
Also add matching test.
Differential Revision: https://developer.blender.org/D11581
Delay depsgraph visibility update tagging until it is known that
graph relations are up to date, and until it is known that the graph
is actually needed to be evaluated.
Differential Revision: https://developer.blender.org/D11660
Currently, the OptiX BVH build options are selected based on whether
we are in background mode (final renders) or not (viewport renders).
In background mode, the BVH is built for fast path tracing and low
memory footprint, while in viewport, it is built for fast updates.
However, on platforms without OpenGL support, the background flag is
always set to true and prevents using fast BVH builds in the viewport.
Now, the BVH options derive from the Scene BVH settings:
* if BVH is static, a fast to trace BVH is built
* if BVH is dynamic, a fast to update BVH is built
Reviewed By: #cycles, brecht
Differential Revision: https://developer.blender.org/D11154
Apply the same fix for T32214 (edge-select failing) to bones
which also failed when their end-points were outside of the view.
- Add V3D_PROJ_TEST_CLIP_CONTENT support for edit & pose bone iterator
and use for selection operators.
- Remove unnecessarily complicated checks with pose-mode lasso tagging.
- Correct error in pose-mode LassoSelectUserData.is_changed
(currently harmless as it's not read back).
`src` and `dst` are perfectly clear, and avoid repeating unecessary
characters when writing the variables many times, allowing more space
for everything else.
When CMake detects and incompatible Python version
it errors out with an error saying at-least python 3.9
is required, but doesn't mention the version it detected.
This makes troubleshooting the problem harder than it
needs to be.
This diff changes the error message to include the python
version CMake detected.
Differential Revision: https://developer.blender.org/D11666
Reviewed By: Ray Molenkamp
This adds preliminary VS 2022 support, since
there currently is no CMake version that
supports the VS2022 IDE only ninja support
was tested.
IDE support should work without any additional
changes as soon as an updated CMake becomes
available.
As VS2022 appears to keep binary compatibility
with earlier MSVC versions, the current SVN
libraries will work for this version.
This commit optimizes the node for the case where it works on many
splines by allowing it to generate mesh data from their combinations
in parallel. By itself, this made the node around twice as fast in my
test file with a result of 20 million vertices, around 600ms instead of
1.2s before.
That isn't actually a very good result; it reveals another bottleneck,
a single threaded loop over all face corners in the mesh normal
calculation code. As a simple change that might improve performance
in some situations, this commit moves normal calculation out of this
node, so at least the work isn't wasted if the mesh is changed later
on in the node tree anyway.
The depth cache (located in `RegionView3D::depths`) is used for quick
and simple occlusion testing in:
- particle selection,
- "Draw Curve" operator and
- "Interactive Light Track to Cursor" operator,
However, keeping a texture buffer in cache is not a recommended practice.
For displays with high resolution like 8k this represents something
around 132MB.
Also, currently, each call to `ED_view3d_depth_override` invalidates
the depth cache. So that depth is never reused in multiple calls from
an operator (this was not the case in blender 2.79).
This commit allows to create a depth cache and release it in the same
operator. Thus, the buffer is kept in cache for a short time, freeing
up space.
No functional changes.
This patch uses threading to flush selection from verts to edges and
from edges to faces. The solution is lockless and should scale well on
modern CPU architectures.
Master:{F10185359}
Patch:{F10185361}
End user performance went from 3.9 to 4.6 FPS (Stanford Dragon) but that
was measured on a Intel Core i7-6700 CPU and AMD RX480 Gpu. The more
cores the better improvements you get.
Reviewed By: mano-wii
Differential Revision: https://developer.blender.org/D11644
This resolves a long standing bug in edge selection
(picking, circle, box & lasso).
Now when one of the edges vertices fails to project into screen space,
the edge is clipped by the viewport to calculate an on-screen location
that can be used instead.
This isn't default as it may be important for the on the screen location
not to be clipped by the viewport.
Adds two new output modes to the CDT api which detect and remove
holes. A hole is a face from which a ray shot to the outside
intersects an even number of constraint edges, except we don't
count constraint edges in the same connected region of faces,
where a region is connected via non-constraint edges.
These modes are useful for filling in outlines meant to represent
text characters and the like.
Original patch was from Erik Abrahamsson, modified by me to work
with the "valid Bmesh" output type too. I also added tests
for the new modes.
Caused by improper testing on my part, assuming a helper function
existed in the industry compatible keymap file, and also assuming it
also used the N and T keys for the left and right side-regions.
A different data structure / implementation is used for curves in the
node tree currently. Converting from the DNA `Curve` structure to this
wasn't slow, but it's nice to decrease overhead. In a test of 14000
splines with 128000 points, this halves the runtime from about 5ms
to about 2.5ms.
The spreadsheet filter tried to apply the mesh selection filter on non-
mesh geometry components. Add a check for the component type,
and also refactor the function to be more easily readable.
The normals were broken because the normal calculation mode wasn't set.
This patch adds a default normal mode so all code creating a spline does
not necessarily have to set it manually. In the future there should be a
way to change this value in the node tree.
- Multi-thread BKE_mesh_recalc_looptri.
- Add BKE_mesh_recalc_looptri_with_normals,
this skips having to calculate normals for ngons.
Exact performance depends on number of faces, size of ngons and
available CPU cores.
For high poly meshes the isolated improvement to BKE_mesh_recalc_looptri
in my tests was between 6.7x .. 25.0x, with the largest gains seen in
meshes containing ngons with many sides.
The overall speedup for high poly meshes containing quads and triangles
is only ~20% although ngon heavy meshes can be much faster.
Remove `seq->tmp` usage from transform code. It was used to tag strips
that need to be "shuffled". Pass these strips in `SeqCollection`
instead.
Reviewed By: sergey, mano-wii
Differential Revision: https://developer.blender.org/D11631
This patch adds support for filtering rows based on rules and values.
Filters will work for any attribute data source, they are a property
of the spreadsheet rather than of the attribute system. The properties
displayed in the row filter can depend on data type of the currently
visible column with that name. If the name is no longer visible, the
row filter filter is grayed out, but it will remember the value until
a column with its name is visible again.
Note: The comments in `screen.c` combined with tagging the sidebar
for redraw after the main region point to a lack of understanding
or technical debt, that is a point to improve in the future.
**Future Improvements**
* T89272: A search menu for visible columns when adding a new filter.
* T89273: Possibly a "Range" operation.
Differential Revision: https://developer.blender.org/D10959
This code was disabled in 2.8 and all other associated code/comments
have been removed/cleared. These rna properties have been replaced with
`seq_prev_type`
Removal of unused local variable. Calculation of underline thickness
no longer needed with change to text output of underscore character.
Introduced in aee04d4960
Differential Revision: https://developer.blender.org/D11641
Currently B-Bone scaling can only be controlled via their
properties, thus requiring up to 8 drivers per joint between
B-Bones to transfer scaling factors from the handle bone.
A Scale Easing option is added to multiply the easing value
by the Y scale channels to synchronize them - this produces a
natural scaling effect where both the shape of the curve and
the scale is affected.
In addition, four toggles are added for each handle, which
multiply each of the X, Y, Z and Ease values by the matching
Local Scale channel of the handle bone, thus replacing trivial
drivers. The Scale Easing option has no effect on this process
since it's easy to just enable both Length and Ease buttons.
Differential Revision: https://developer.blender.org/D9870
Implement actual behavior for the B-Bone Y Scale channels added
to DNA and UI in the previous commit in addition to the existing
X and Z Scale inputs.
The two length scale inputs control the ratio between the lengths
of the start and end segments of the bone: although for convenience
two inputs are provided, the whole chain is still uniformly scaled
to fit the curve.
Differential Revision: https://developer.blender.org/D9870
In addition to the base bone transformation itself, B-Bones have
controls that affect transformation of its segments. For rotation
the features are quite complete, allowing to both reorient the
Bezier handles via properties, and to control them using custom
handle bones. However for scaling there are two deficiencies.
First, there are only X and Y scale factors (actually X and Z),
while lengthwise all segments have the same scaling. The ease
option merely affects the shape of the curve, and does not cause
actual scaling.
Second, scaling can only be controlled via properties, thus
requiring up to 6 drivers per joint between B-Bones to transfer
scaling factors from the handle bone. This is very inefficient.
Finally, the Z channels are confusingly called Y.
This commit adds a B-Bone Y Scale channel and extra B-Bone flag
fields to DNA with appropriate versioning (including for F-Curves
and drivers) in preparation to addressing these limitations.
Functionality is not changed, so the new fields are not used
until the following commits.
Differential Revision: https://developer.blender.org/D9870
When drawing mnemonic underlines for hotkeys, use text output of
underscore character instead of direct drawing a line. Otherwise these
are not visible in dialog buttons.
Introduced in 0fcc063fd9
Differential Revision: https://developer.blender.org/D11641
Reviewed by Campbell Barton
Using an ampersand here is more semantically correct. A slash indicates "or" while an ampersand indicates "and".
An ampersand here is best because the view type shows both the Sequencer and the Preview.
When having multiple materials in a mesh the triangles are sorted based
on material index. This sorting is done single threaded, but needs two
loops over the data. One to count the bucket size and the second one to
add the triangles to the right position in the buckets.
This patch will do the counting in a multithreaded approach that would
speed up the cache creation. It has been measured that this part is the
most blocking part of the cache creation.
Reviewed By: mano-wii
Differential Revision: https://developer.blender.org/D11615
During selections the total selection is refreshed at the end. This
process was done single threaded. This patch will do a parallel iter
approach.
Master: 0.043612s Threaded 0.017964s.
Master: {F10179586}
This patch: {F10179587}
Reviewed By: mano-wii
Differential Revision: https://developer.blender.org/D11622
`ED_fileselect_get_asset_params` would only return actual data pointer
when file browser is in ASSET mode.
Calling that whole section only makes sense if filebrowser is in asset
mode anyway.
Regression introduced in rBf6c5af3d4753 I think.
@Severin committing this fix now as this is a fairly critical bug for
the studio, feel free to revert and do proper fix if this one is not the
best solution.
This patch ensures that selection mode flushing updates total selection
counts internally. This reduces recounting when we are sure that the
input total selection counts were up to date.
For example for circle selection the total selection counts were
correct. But during flushing the selection could have been changed and
therefore the selection was always recounted.
This increased the performance on selected system from 6.90 FPS to 8.25
FPS during circle selection operations.
Before: {F10179981}
After: {F10179982}
Reviewed By: mano-wii
Differential Revision: https://developer.blender.org/D11647
When an object, whose mesh gets loaded from Alembic, gets animated in
Blender and the Alembic CacheFile datablock also gets animated, editing
keyframes causes both datablock to be re-copied for evaluation. This
caused a threading issue and a double-free of some memory. This is fixed
by expanding the scope of the spin lock in
`BKE_cachefile_reader_free()`.
Contributed by @erik85 in D11400. The idea from this patch was placed in
a more generic context: A new FOR macro has been added that loops
over the neighbors of a cell within a given radius.
This matches BMesh which also has tessellation in it's own file.
Using a separate file helps with organization when
extracting code into smaller functions.
The //Raycast// node intersects rays from one geometry onto another.
It computes hit points on the target mesh and returns normals, distances
and any surface attribute specified by the user.
A ray starts on each point of the input //Geometry//. Rays continue
in the //Ray Direction// until they either hit the //Target Geometry//
or reach the //Ray Length// limit. If the target is hit, the value of the
//Is Hit// attribute in the output mesh will be true. //Hit Position//,
//Hit Normal//, //Hit Distance// and //Hit Index// are the properties of the
target mesh at the intersection point. In addition, a //Target Attribute//
can be specified that is interpolated at the hit point and the result
stored in //Hit Attribute//.
Docs: D11620
Reviewed By: HooglyBoogly
Differential Revision: https://developer.blender.org/D11619
This node creates splines with more control points in between the
existing control points. The point is to give the splines more
definition for further tweaking like randomization with white noise,
instead of deforming a resampled poly spline with a noise texture.
For poly splines and NURBS, the node simply interpolates new values
between the existing control points. However, for Bezier splines,
the result follows the existing evaluated shape of the curve, changing
the handle positions and handle types to make that possible.
The number of "cuts" can be controlled by an integer input, or an
attribute can be used. Both spline and point domain attributes are
supported, so the number of cuts can vary using the value from the
point at the start of each segment.
Dynamic curve attributes are interpolated to the result with linear
interpolation.
Differential Revision: https://developer.blender.org/D11421
The minimum twist mode is important because it allows creating
normals without sudden changes in direction. The disadvantage
of minimum twist normals is that the normals depend on all control
points. So changing one control point can change the normals
everywhere. The computed normals do not match the existing
code exactly, although they do match quite well on non-cyclic and
on some cyclic curves. I also noticed that the existing implementation
has some fairly simple failure cases that I haven't found in the new
implementation so far.
Differential Revision: https://developer.blender.org/D11621
This makes the parts where a node is locked more explicit. Also, now the thread
is isolated when the node is locked. This prevents some kinds of deadlocks
(which haven't happened in practice yet).
- BKE_mesh_copy_parameters_for_eval to be used for evaluated meshes only
as it doesn't handle ID user-counts.
- BKE_mesh_copy_parameters is a general function for copying parameters
between meshes.
Refactor function `freeSeqData` so it is readable.
One strip can have multiple transform operations defined. To prevent
processing strip multiple times, build `SeqCollection` and use
sequencer iterator instead of iterating `TransData` directly.
No functional changes.
Differential Revision: https://developer.blender.org/D11618
Reduce complexity of sequencer transform code by removing recursivity.
This is possible by treating meta strips (mostly) as any other strip and
containing all transform code within SEQ_ functions.
Unfortunately internally meta strips still require special treatment,
but all complexity from code all over transform code seems to be
possible to contain within one function.
Functional change:
Previously adjusting handle of single image strip moved animation.
Now animation is not moved, which is behavior for all other strips.
Reviewed By: sergey, mano-wii
Differential Revision: https://developer.blender.org/D11493
Starts scrolling when dragging a node or node link and going outside the current window.
Largely copied from the VIEW2D_OT_edge_pan operator.
Edge panning operator customdata and supporting functions now in
UI_view2d.h, so they could be used by operators in other editor
libraries. The VIEW2D_OT_edge_pan operator also uses this customdata and
shared functions now. Operators properties can be used to configure
edge panning margins and speed for each use case, rather than using
hardcoded values.
The speed function for edge panning has been tweaked somewhat:
* "Speed per pixel" has been replaced with a "speed ramp" distance.
This is more intuitive and also creates an upper bound for the speed,
which can otherwise become extreme with large cursor distance.
* "Max speed" is reached at the end of the speed ramp.
* Padding the region inside and outside is applied as before, but both
values are operator properties now.
Node transform operator also supports edge panning. This requires
an offset for changes in the view2d rect, otherwise nodes are "stuck"
to the original view.
Transform operator had cursor wrapping categorically enabled, but this
gets quite confusing with the edge scrolling mechanism. A new TransInfo
option T_NO_CURSOR_WRAP has been introduced to disable this behavior.
The double negative is a bit annoying, but want to avoid affecting the
existing transform modes, so by default it should still set the
OP_IS_MODAL_GRAB_CURSOR flag (which then sets the WM_CURSOR_WRAP_XY
flag during modal execution).
Reviewed By: HooglyBoogly, JacquesLucke
Differential Revision: https://developer.blender.org/D11073
This allows line art to run only once for each modifier stacks,
with an option to toggle a specific line art modifier should
use cache or re-do their own calculations.
Reviewed By: Sebastian Parborg (zeddb), Hans Goudey (HooglyBoogly)
Differential Revision: https://developer.blender.org/D11291
This makes it easier to use task isolation in c++ code.
Previously, one either had to check `WITH_TBB` (possibly indirectly
through `WITH_OPENVDB`) or one had to use the C function which
is less convenient.
This namespace groups threading related functions/classes. This avoids
adding more threading related stuff to the blender namespace. Also it
makes naming a bit easier, e.g. the c++ version of BLI_task_isolate could
become blender::threading::isolate_task or something similar.
Differential Revision: https://developer.blender.org/D11624
Similar to what we do for constraints and modifiers, except that
currently adding or editing shaderfx in liboverride objects is
completely unsuported.
Fix T88974.
Used to detect if a shaderfx is purely local, or comes from linked data,
in case of a liboverride.
Not actually used yet since we do not currently support adding
shaderfx's to overrides, but will be in the future, and matches
constraints and modifiers code.
Mainly:
* Make `ED_operator_object_active_editable_ex` properly report poll
messages on failure.
* Add `ED_operator_object_active_local_editable_posemode_exclusive` for
bone constraints requiring pure local Object (non-override one).
* General cleanup and adding more poll messages on failures.
On selecting a multi-layer image with a combined pass, a "Combined"
socket is created and default combined pass socket "Image" is
disabled by setting `SOCK_UNAVAIL` flag. When converting into
operations, `ImageNode` converts alpha socket on finding any socket with
a combined pass without checking the flag.
Since commit rB93e2491ee724 an assertion fails when mapping sockets
twice because now map `add_new` is used.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11566
Adds full frame implementation to Value node operation.
No functional changes.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11594
This adds support for importing UV sets which are defined per vertex,
instead of per face corners. Such UV sets can be generated when the
mesh is split according to UV islands, or when there is only one UV
island, in which cases only a single UV value can be stored per
vertex since vertices will never be on a seam.
- Include all objects in pose mode.
- Show the number of objects in pose mode.
- Show the number of objects in edit mode for all types of objects
(not just meshes).
This adds support for importing UV sets which are defined per vertex,
instead of per face corners. Such UV sets can be generated when the
mesh is split according to UV islands, or when there is only one UV
island, in which cases only a single UV value can be stored per
vertex since vertices will never be on a seam.
Reviewed By: sybren
Differential Revision: https://developer.blender.org/D11584
Implementation of T86970. This node takes a geometry input with
multiple components and outputs them by component type. Meshes,
Curves, and Point Clouds support combining multiple input instances,
while volumes will only output the first volume component input until
suitable instance realization for multiple volumes is finished.
When direct geometry instancing is implemented it will be possible to
avoid realizing instances in this node.
Differential Revision: https://developer.blender.org/D11577
Creating a new full screen area had it's area initialized as empty,
updating the screen then set the area to a 3D view (as a fallback),
before the actual area type was set.
This made setting the intended space-type run the 3D views exit callback
on a 3D view without a View3D struct allocated, which the exit callback
needed to account for.
Resolve by calling ED_screen_change after the area type has been set.
This patch improves the 3DView statistics overlay to show LOCAL stats
while in local view. This means the stats can vary between 3DViews and
the statusbar when views are in local view, but this gives a much more
accurate count of the objects, and their components, that you are
directly working with rather than just scene values.
Differential Revision: https://developer.blender.org/D8883
Reviewed by Campbell Barton
Versioning code for converting strip offset property doesn't work, when
property was animated and disabled or when crop was used.
When offset property is animated and offset is enabled, animation is
converted to be used with new transform design. When offset is disabled,
animation is left untouched. New transform design doesn't have option
to disable offset, and therefore old unconverted animation is used
instead of converted static value.
Remove animation from propery if it was unused.
Another issue was that both X and Y offset animation was being corrected
by factor caluclated for X channel.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11370
Use lookup string callback function for `sequences_all` RNA property
`rna_SequenceEditor_sequences_all_lookup_string` using a GHash for faster lookups.
When names are changed or strips are added/removed the lookup is tagged invalid.
The next time the lookup is used it will rebuild it.
Reviewed By: sergey, jbakker
Differential Revision: https://developer.blender.org/D11544
Just like the way we often have a choice between an attribute input and
a single float, this adds the ability to choose between attribute and
integer input sockets, useful for D11421.
There is no particular reason these two functions shouldn't be used
outside of the bezier spline implementation since they don't do anything
particularly controversial.
Create maps that specify which batches have vbo or ibo as a reference
and use these maps to discard batches along with buffers.
Differential Revision: https://developer.blender.org/D11588
Adds `source/blender/blendloader/intern/versioning_common.cc` and
`versioning_common.h` for version independent versioning functions.
I only placed `do_versions_add_region_if_not_found()` in there for now.
`blo_do_version_old_trackto_to_constraints()` could also be added, but
that's so old, I prefer keeping that in `versioning_legacy.c`.
Widget drawing code already supported drawing right-aligned, grayed out
shortcut strings. This patch generalizes things a bit so this can also
be used to draw other hints in the same way. There have been a few
instances in the past where this would've been useful, D11046 being the
latest one.
Note that besides some manual regression testing, I didn't check if this
works yet, as there is no code actually using it (other than the
shortcuts). Can be checked as part of further development for D11046.
A possible further improvement would be providing a way to define how
clipping should be done. E.g. sometimes the right-aligned text should be
clipped first (because it's just a hint), in other cases it should be
left untouched (like current code explicitly does it for shortcuts).
Removes the `UI_BUT_HAS_SHORTCUT` flag, which isn't needed anymore.
After looking into task isolation issues with Sergey, we couldn't find the
reason behind the deadlocks that we are getting in T87938 and a Sprite Fright
file involving motion blur renders.
There is no apparent place where we adding or waiting on tasks in a task group
from different isolation regions, which is what is known to cause problems. Yet
it still hangs. Either we do not understand some limitation of TBB isolation,
or there is a bug in TBB, but we could not figure it out.
Instead the idea is to use isolation only where we know we need it: when
holding a mutex lock and then doing some multithreaded operation within that
locked region. Three places where we do this now:
* Generated images
* Cached BVH tree building
* OpenVDB lazy grid loading
Compared to the more automatic approach previously used, there is the downside
that it is easy to miss places where we need isolation. Yet doing it more
automatically is also causing unexpected issue and bugs that we found no
solution for, so this seems better.
Patch implemented by Sergey and me.
Differential Revision: https://developer.blender.org/D11603
When using multiple materials in a single mesh the most time is spend in
counting the offsets of each material for the sorting.
This patch moves the counting of the offsets to render mesh data and
caches it as long as the geometry doesn't change.
This patch doesn't include multithreading of this code.
Reviewed By: mano-wii
Differential Revision: https://developer.blender.org/D11612
Clear the runtime data structs instead of individual members,
this simplifies adding new runtime members as there are at least
two places they would need to be cleared.
Resolves error in D8883.
This commit lets the join geometry node transfer dynamic attributes
to the result, the same way that point cloud and mesh attributes are
joined. The implementation is different though, because of an
optimization implemented for curves to avoid copying splines.
The result attribute is added with the highest priority domain (points
over splines), and the highest complexity data type. If one curve had
the attribute on the spline domain but not others, the point domain
values will be used.
Generally this is a bit lower level than I would have liked this code
to be, but should be efficient, and it's really not too complicated.
Differential Revision: https://developer.blender.org/D11491
Instead of building a set and then determining the final domain and
type for every attribute separately in the loop, construct a map with
the necessary data in the first place. This is simpler and should be
slightly more efficient.
Split from D11491
This is an implementation of T88722. It accepts a curve object and
for each spline, reverses the order of the points and all attributes.
This is more of a foundational node to support other nodes in the
future (like curve deform)
Selection takes spline domain attributes to determine which splines
are selected. If no selection is present all splines are reversed.
Differential Revision: https://developer.blender.org/D11538
This node implements the second option of T87429, creating points
along the input splines with the necessary evaluated information
for instancing: `tangent`, `normal`, and `rotation` attributes.
All generic curve point and spline attributes are copied to the
result points as well.
The "Count" and "Length" methods are just like the current options
in the resample node, but the output is points instead of a curve.
The "Evaluated" method uses the points you see on the curve directly,
and therefore should be the fastest.
The rotation data is retrieved from a transform matrix built with the
same method that the curve to mesh node uses. The radius attribute is
divided by 10 so the points don't look absurdly huge in the viewport.
In the future that could be an option.
For the implementation, one thing that could use an improvement
is the amount of temporary allocations while resampling to evaluated
points before the final points. I expect that reusing a buffer for
each thread would give a nice improvement.
Differential Revision: https://developer.blender.org/D11539
This patch allows Windows users to specify that their current blender
installation should be used to create thumbnails and be associated
with ".blend" files. This is done from Preferences / System. The only
way to do this currently is from the command-line and this is sometimes
inconvenient.
Differential Revision: https://developer.blender.org/D10887
Reviewed by Brecht Van Lommel
If the last decoded frame had the same timestamp as the GOP current
packet, then we would skip over this frame when fast forwarding and we
would seek until the end of the file.
This would could only be triggered reliably in single threaded mode.
Reviewed By: Richard Antalik
Differential Revision: http://developer.blender.org/D11601
The warning would appear when using the `ENUM_OPERATORS()` macro inside
of an `extern "C"` block.
Didn't cause a warning in master currently, but in the
`asset-browser-poselib` branch.
After macro expansion, there would be C++ code in code with C linkage
(`extern "C"`). So make sure the expanded C++ code always uses C++
linkage.
The syntax used is totally C++ compliant: the C++ standard requires that
in such nested linkage specifications, the innermost one determines the
linking language (e.g. see
https://timsong-cpp.github.io/cppwp/n4659/dcl.link#4).
On the Windows platform there will be some errors printed to the
console if the user's thumbnail cache folder doesn't already exist.
While creating those folders there is an attempt to do so multiple
times and so we get errors when trying to create when exists. This
is caused by paths that have hard-coded forward slashes, which causes
our path processing routines to not work correctly. This patch defines
those paths using platform-varying separator characters.
Differential Revision: https://developer.blender.org/D11505
Reviewed by Brecht Van Lommel
Support calculating face normals when tessellating. When this is done
before updating vertex normals it gives ~20% performance improvement.
Now vertex normal calculation only needs to perform a single pass on the
mesh vertices when called after tessellation.
Extended versions of normal & looptri update functions have been added:
- BM_mesh_calc_tessellation_ex
- BM_mesh_normals_update_ex
Most callers don't need to be aware of this detail by using:
- BKE_editmesh_looptri_and_normals_calc
- BKE_editmesh_looptri_and_normals_calc_with_partial
- EDBM_update also takes advantage of this,
where calling EDBM_update with calc_looptri & calc_normals
enabled uses the faster normal updating logic.
Bypass stored edge-vectors for ~16% performance gains.
While this increases unit-length edge-vector calculations by around ~4x
the overhead of a parallel loop over all edges makes it worthwhile.
Note that caching edge-vectors per-vertex performs better and may be
worth investigating further, although in my tests this increases code
complexity with barley measurable benefits over not using cache at all.
Details about performance and possible optimizations are noted in
bm_vert_calc_normals_impl.
Rename function EDBM_update_generic to EDBM_update, use a parameters
argument for better readability.
Also add calc_normals argument, which will have benefits when
calculating normals and tessellation together is optimized.
For some custom rendering engines it's advantageous not to write the image files to disk.
An example would be a network rendering engine which does it's own image writing.
This feature is only supported when bl_use_postprocess is also disabled, since render
engines can't influence the saving behavior of the sequencer or compositor.
Differential Revision: https://developer.blender.org/D11512
Better to avoid explicit vectors components direct manipulation when a
generic operation for whole vector exists, if nothing else it avoids
potential mistakes in indices.
Those were caused by various tools used on degenerate geometry, see
T79775.
Note that fixes are as low-level as possible, to ensure they cover as
much as possible of unreported issues too.
We still probably have many more of those hidden in BLI_math though.
This commit does two things:
* Disallows creating more than one link from one socket to a multi socket input.
* Properly count links if there happen to be more than one link between the same sockets.
The new link counting should also be more efficient asymptotically.
Differential Revision: https://developer.blender.org/D11570
- The argument with named with an `r_` prefix when it was in fact
also read from.
- The argument passed in had to be 'psys->clmd->hairdata',
if it was not - the function would not worked.
When moving a linked collection, we seem to only receive a depsgraph update
for an empty object so the Blender synchronization cannot discriminate it
and tag the object(s) (light or geometry) for an update through
id_map.set_recalc.
This missing transform update only affects lights since we do not check
manually if the transformations were modified like we do for objects.
To fix this, add a check to see if the transformation is different provided
that a light was already created.
Reviewed By: brecht
Maniphest Tasks: T88515
Differential Revision: https://developer.blender.org/D11574
This patch improves the positioning of child windows when on monitors
that are arranged vertically (any above any other). When calculating a
window position in Ghost coordinates from GL coordinates we were using
monitor height, which can give incorrect values when desktop is taller
than any single monitor. So use desktop height instead.
See D10637 for more details and examples.
Differential Revision: https://developer.blender.org/D10637
Reviewed by Brecht Van Lommel
This patch makes the "Render" window a top-level window, not a child of
the main window, which was the case in blender versions prior to 2.93.
This means it is no longer "on top", nor is the icon grouped on the
taskbar in the same way, but you can Alt-Tab between it and the main
window. This change only affects the Windows platform as the other
platforms behave this way.
See D11576 for links to negative feedback that prompts this change.
Differential Revision: https://developer.blender.org/D11576
Reviewed by Brecht Van Lommel
This moves the flash on mode transfer effect option from the overlays to
an operator property of the mode transfer operator.
- This effect is intended to show the target object when no overlays or
a minimal set of overlays is enabled. Making it part of the whole set of
overlays invalidates this use case.
- The effect is not intended to be configurable per viewport, it should
be a global option.
The effect is still implemented using the overlay engine (instead of a
draw modal callback) due to performance and drawing artifacts. Having it
implemented as an overlay with runtime timer data in the objects makes
also possible to run multiple animations at the same time without any
visual glitches.
Reviewed By: campbellbarton, JulienKaspar
Differential Revision: https://developer.blender.org/D11519
Note: Linking in this case as in link vs. append. Easily confused with linking
a data-block to multiple usages (e.g. single material used by multiple
objects).
Adds a drop-down to the Asset Browser header to choose between Link and Append.
This is probably gonna be a temporary place, T54642 shows where this could be
placed eventually.
Linking support is crucial for usage of the asset browser in production
environments. It just wasn't enabled yet because a) the asset project currently
focuses on single user, not production assets, and b) because there were many
unkowns still for the workflow that have big impact on production use as well.
With the recently held asset workshop I'm more confident with enabling linking,
as design ideas relevant to production use were confirmed.
Differential Revision: https://developer.blender.org/D11536
Reviewed by: Bastien Montagne
The operator run when dropping objects would duplicate the dropped object and
place that in the scene, even though that was just appended. Addressed by
making the duplication optional for the operator. If the duplication is not
requested, the object is just added to the scene (if needed), repositioned
based on the drop location and selected (deselecting other objects).
This makes the operator work as expected when using it to drop assets.
Reviewed as part of https://developer.blender.org/D11536.
Reviewed by: Bastien Montagne
Fix segmentation fault that can occur when reordering animation
channels.
Under some specific conditions, the list "act->curves" is empty in the
"join_groups_action_temp" function. In particular, this happens when a
scene contains an action that has not been pushed down, and with no
keyframe in it.
Reviewed By: sybren
Differential Revision: https://developer.blender.org/D11569
While this preprocessing does take some time upfront,
it avoids longer lookup later on, especially as nodes get
more sockets.
It's probably possible to make this more efficient in some cases
but this is good enough for now.
Scanline processor did its own heurestic what didn't scale well when
having a multiple cores. In stead of using our own code this patch will
leave it to TBB to determine how to split the scanlines over the
available threads.
Performance of the IMB_transform before this change was 0.002123s, with
this change 0.001601s. This change increases performance in other areas
as well including color management conversions.
Reviewed By: zeddb
Differential Revision: https://developer.blender.org/D11578
When using large sequences including audio the drawing of the audio on
top of the strip takes a lot of time. This effects the playback
performance heavily.
During the animation playback performance there was a solution for this
by only drawing the playhead overlay. This was reverted for the sequence
editor as it didn't update the color strips when they were animated.
This patch checks if there are animated color strips if so the full
screen is redrawn, otherwise only the playhead is redrawn.
Reviewed By: ISS
Differential Revision: https://developer.blender.org/D11580
One drawback to trying to predict the number of threads that will be
used in the `task_graph` is that we are only sure of the number when the
threads are running.
Using `BLI_task_parallel_range` allows the driver to
choose the best thread distribution through `parallel_reduce`.
The benefit is most evident on hardware with fewer cores.
This is the result on an 4-core laptop:
||before:|after:
|---|---|---|
|large_mesh_editing:|Average: 5.203638 FPS|Average: 5.398925 FPS
||rdata 15ms iter 43ms (frame 193ms)|rdata 14ms iter 36ms (frame 187ms)
Differential Revision: https://developer.blender.org/D11558
This is an adaptation of {D11488}.
A disadvantage of manually setting the iter ranges per thread is that
we don't know how many threads are running in the background and so we
don't know how to best distribute the ranges.
To solve this limitation we can use `parallel_reduce` and thus let the
driver choose the best distribution of ranges among the threads.
This proved to be especially beneficial for computers with few cores.
**Benchmarking:**
Here's the result on an 4-core laptop:
||master:|PATCH:
|---|---|---|
|large_mesh_editing:|Average: 5.203638 FPS|Average: 5.398925 FPS
||rdata 15ms iter 43ms (frame 193ms)|rdata 14ms iter 36ms (frame 187ms)
Here's the result on an 8-core PC:
||master:|PATCH:
|---|---|---|
|large_mesh_editing:|Average: 15.267482 FPS|Average: 15.906881 FPS
||rdata 9ms iter 28ms (frame 65ms)|rdata 9ms iter 25ms (frame 63ms)
|large_mesh_editing_ledge: |Average: 15.145966 FPS|Average: 15.520474 FPS
||rdata 9ms iter 29ms (frame 65ms)|rdata 9ms iter 25ms (frame 64ms)
|looptris_test:|Average: 4.001917 FPS|Average: 4.061105 FPS
||rdata 12ms iter 90ms (frame 236ms)|rdata 12ms iter 87ms (frame 230ms)
|subdiv_mesh_cage_and_final:|Average: 1.917769 FPS|Average: 1.971790 FPS
||rdata 7ms iter 37ms (frame 261ms)|rdata 7ms iter 31ms (frame 258ms)
||rdata 7ms iter 38ms (frame 252ms)|rdata 7ms iter 33ms (frame 249ms)
|subdiv_mesh_final_only:|Average: 6.387240 FPS|Average: 6.591251 FPS
||rdata 3ms iter 25ms (frame 151ms)|rdata 3ms iter 16ms (frame 145ms)
|subdiv_mesh_final_only_ledge:|Average: 6.247393 FPS|Average: 6.596024 FPS
||rdata 3ms iter 26ms (frame 158ms)|rdata 3ms iter 16ms (frame 148ms)
**Notes:**
- The improvement can only be noticed if all extracts are multithreaded.
- This patch touches different areas of the code, so it can be split into another patch if the idea is accepted.
These screenshots show how threads behave in a quadcore:
Master:
{F10164664}
Patch:
{F10164666}
Differential Revision: https://developer.blender.org/D11558
ffmpeg_generic_seek_workaround did work properly and our start pts
calculation was wrong.
Reviewed By: Richard Antalik
Differential Revision: http://developer.blender.org/D11562
Because of the added sanity checks in rB14508ef100c9 (D11492), seeking
in proxies would not work correctly any more. This is because it wasn't
working as intended before, but in most cases this wouldn't be
noticeable. However now when the sanity checks are tripped it is very
noticeable that something is wrong
The indexer tried to use dts values for time stamps when we used pts in
our decode functions to get the time positions. This would make it
start in the wrong GOP frames when searching. Now that we enforce no
crossing of GOP frames when decoding after seek, this would lead to
issues.
Now we correctly use pts (or dts if pts is not available) and thus we
don't have any seeking issues because of time stamp format missmatch.
Reviewed By: Richard Antalik
Differential Revision: http://developer.blender.org/D11561
When sampling ImBuf can be a char or a float buffer. Current sampling
functions added overhead by checking which kind of buffer was passed
every pixel that was sampled. When performing image processing this
check can be removed outside the inner loop adding 5% of performance
increase in the `IMB_transform` operator.
Remap {key Alt I} from `anim.delete_key_v3d` to `anim.delete_key`. This
makes it keyframe removal symmetrical with keyframe insertion ({key I}).
Both the default keymap {key Alt I} and the Industry Compatible keymap
{key Alt S} have been updated.
Reviewed By: sybren, #animation_rigging
Maniphest Tasks: T88068
Differential Revision: https://developer.blender.org/D11528
Inside the sequencer the cropping and transform of images/buffers were
implemented locally. This reduced the optimizations that a compiler
could do and added confusing code styles. This patch adds
`IMB_transform` to reduce the confusion and increases compiler
optimizations as more code can be inlined and we can keep track of
indices inside the inner loop.
This increases end-user performance by 30% when playing back aa video
in VSE.
Reviewed By: ISS, zeddb
Differential Revision: https://developer.blender.org/D11549
This patch changes occurrences of percentage to factor.
There are some usages of percentage left in there on purpose.
They are distinguished as follows:
- factor is 0-1 float
- percentage is 0-100 int
Ref D11361
Reviewed by: sybren, campbellbarton
Prepare node for conversion to Geometry Nodes.
There should be no functional changes.
Reviewed By: HooglyBoogly
Differential Revision: https://developer.blender.org/D11506
When using non-default system separator in filename path, code would end up
with an absolute path mixing regular and alternative separator,
confusing the rest of the path manipulations later on.
So this commit add proper replacements of alternative separators, and
path normalization.
Splitting out thread safe iteration logic means regular iteration
isn't checking for the thread-safe pointer each step.
This gives a small but measurable overall performance gain of 2-3%
when redrawing a high-poly mesh.
Ref D11564
Reviewed By: mont29
This change was prompted by D6408 which moves thumbnail extraction into
a shared function that happens use these endian defines but only links
blenlib.
There is no need for these defines to be associated with globals
so move into their own header.
Allows to define properties which will have proper units displayed
in the interface. The internal storage is expected to be seconds
(which matches how other times are stored in Blender).
Is not immediately used in Blender, but is required for the upcoming
feature in Cycles X (D11526)
The naming does not sound very exciting, but can't think of anything
better either.
For test it probably easiest to define FloatProperty with subdtype
of TIME_ABSOLUTE.
Differential Revision: https://developer.blender.org/D11532
Previously, a vertex from destination mesh would always be added to all
transferred vgroup (with a 0.0 weight), even if none of its matching
sources in source mesh belonged to the matching source vgroups.
Now a destination vertex is only added to a given destination vgroup if
at least one of its source vertices belong to the matching source
vgroup.
Issue found and initial investigation by @pls in D11524, thanks!
Doxygen by default leaves out any functions inside
#ifdef blocks that it thinks are disabled.
This change adds a DOXYGEN symbol, so you can
still get the functions included in the
documentation even if the #ifdef would
have normally excluded them.
before
#if defined(_WIN32)
after
#if defined(_WIN32) || defined(DOXYGEN)
Patch provided by Campbell Barton on chat.
TBBmalloc_proxy already takes care of any allocations
being done from MSVC compiled code, some of the dependencies
like GMP cannot be build with MSVC and we have to use
mingw to build them. mingw however links against the older
msvcrt.dll for its allocation needs, which TBBMallocProxy
does not hook.
GMP has an option to supply your own allocation functions
so we can still manually redirect them to TBBMalloc.
In a test-file with a boolean geometry node, this patch
uses 32s effective CPU time compared to 52s before.
Differential Revision: https://developer.blender.org/D11435
Reviewed by Campbell Barton, Ray Molenkamp
Remove an unnecessary call to pose_slide_mouse_update_percentage
That call was overriding the typed value
Reviewed By: #animation_rigging, Sybren A. Stüvel
Differential Revision: https://developer.blender.org/D11395
Ref D11395
To draw the overlay stats in columns the strings must be measured to
find the longest one. In some circumstances this measurement can be
incorrect. We draw the text with a specific size yet do not explicitly
set the size before calling BLF_size. This patch properly sets the size
so that the measurement will match what will be used for output.
See T88799 for examples of measurement failure.
Differential Revision: https://developer.blender.org/D11504
Reviewed by Hans Goudey
First, expand on the interpolation to evaluated points with a templated
helper function, and a function that takes a GSPan. Next, add a set of
functions to `Spline` for interpolating at arbitrary intervals between
the evaluated points. The code for doing that isn't that complicated
anyway, but it's nice to avoid repeating, and it might make it easier
to unroll the special cases for the first and last points if we require
the index factors to be sorted.
This commit adds a node to output the convex hull of any input geometry
as a mesh, which is an enclosing geometry around a set of points.
All geometry types are supported, besides volumes.
The code supports operating on instances to avoid copying all input
geometry before the operation. The implementation uses the same backend
as the operation in edit mode, but uses Mesh directly instead of BMesh.
Attribute transfer is not supported currently, but would be a point of
improvement for the future if it can work in a predictable way on
different geometry input types.
Differential Revision: https://developer.blender.org/D10925
Note that these changes are limited simple cases as these kinds of
changes could allow for errors when refactoring code when the known
state is not so obvious.
Use BM_iter_parallel for face tessellation, this gives around
6.5x speedup for BM_mesh_calc_tessellation on high poly meshes in my
tests, although exact speedup depends on available cores.
Support thread local storage for BLI_task_parallel_mempool,
as well as support for the reduce and free callbacks.
mempool_iter_threadsafe_* functions have been moved into a private
header thats only shared between task_iterator.c and BLI_mempool.c
so the TLS can be made part of the iterator array without having to
rely on passing in struct offsets.
Add test task.MempoolIterTLS that ensures reduce and free
are working as expected.
Reviewed By: mont29
Ref D11548
Also modified existing utility functions to take
care of the new surface normal interpolation and so on.
Reviewed By: Antonio Vazquez (antoniov)
Differential Revision: https://developer.blender.org/D11543
While this should not be allowed in general, there are some cases
(library overrides at least) where supporting this is mandatory.
Further more, comment stating that this could crash is from 2011, could
not reproduce any issue with current code. Commit comment was referring
to undo/redo, but in use cases here those should not affect things.
Note that in general, such relatively high-level checks should be
handled by high-level, close to user code (like in ED area e.g.), not in
low-level BKE code anyway.
This patch adds a specific extraction method when the mesh has only
one material. This method is multi-threaded.
There is a trade-off in this patch as the ibo isn't compressed (it adds
restart indexes for hidden faces). So it depends if threading is faster
than the additional GPU buffer upload.
# Subdivided cube
I used a cube subdivided 7 times, modifiers applied. that gives around 400000 faces.
The test is selecting some vertices and move them. During this test the next buffers are updated on each frame:
* vbo.pos_nor
* vbo.lnor
* vbo.edit_data
* ibo.tris
* ibo.points
System info:
|platform| Linux-5.11.0-7614-generic-x86_64-with-glibc2.33|
| renderer| AMD SIENNA_CICHLID (DRM 3.40.0, 5.11.0-7614-generic, LLVM 11.0.1)|
|vendor| AMD|
|version| 4.6 (Core Profile) Mesa 21.0.1|
|cpu| Intel(R) Core(TM) i7-6700 CPU @ 3.40GHz|
|compiler| gcc version 10.3.0|
Timing have been measured using DEBUG_TIME in `draw_cache_extract_mesh`.
master: `rdata 8ms iter 45ms (frame 153ms)`
this patch `rdata 6ms iter 36ms (frame 132ms)`
Reviewed By: mano-wii
Maniphest Tasks: T88352
Differential Revision: https://developer.blender.org/D11290
This commit adds a flag to disable displaying some socket labels which
just redundant eye sores. We still want to have a label, because that
information can potentially be accessed elsewhere in the UI.
The flag is used in a few geometry nodes.
Differential Revision: https://developer.blender.org/D11540
This is an optimization, but the difference is still not that
significant as some extractions are still done in single thread.
**Benchmarking**
||before:|after:
|---|---|---|
|large_mesh_editing:|Average: 14.246502 FPS|Average: 15.438118 FPS
||rdata 9ms iter 31ms (frame 69ms)|rdata 9ms iter 27ms (frame 65ms)
|large_mesh_editing_ledge: |Average: 14.913622 FPS|Average: 15.856538 FPS
||rdata 9ms iter 30ms (frame 67ms)|rdata 9ms iter 26ms (frame 63ms)
|looptris_test:|Average: 3.970774 FPS|Average: 4.095200 FPS
||rdata 11ms iter 90ms (frame 235ms)|rdata 12ms iter 87ms (frame 229ms)
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11467
A single threaded task with thread data over 8192 bytes would leak.
While this didn't happen in practice, it could cause issues in the
future.
The free call for `task_parallel_iterator_do` wasn't running
if callbacks weren't set, also not an issue in practice but avoids
potential problems in the future too.
Changes introduced in commit rBe9f2f17e8518
can create different render results when there is
a Math or Mix operation after TextureOperation
on tiled execution model.
This is due to WriteBufferOperation forcing a single pixel
resolution when these operations use a preferred
resolution of 0 to check if their inputs have resolution.
Fixing this behaviour creates different renders too.
This patch keeps previous tiled implementation and
adds the new implementation only for full frame execution.
Reviewed By: Jeroen Bakker (jbakker)
Differential Revision: https://developer.blender.org/D11546
In order to reduce stack size this patch converts full frame
recursive methods into iterative.
- No functional changes.
- No performance changes.
- Memory peak may slightly vary depending on the tree because
now breadth-first traversal is used instead of depth-first.
Tests in D11113 have same results except for test1 memory peak:
360MBs instead of 329.50MBs.
Reviewed By: Jeroen Bakker (jbakker)
Differential Revision: https://developer.blender.org/D11515
CMake builder and install deps changes, precompiled libraries are still to be
committed.
Ref T88438, T88434
Differential Revision: https://developer.blender.org/D11486
During transforming an image, a matrix multiplication per pixel was done.
The matrix in itself is always linear so it could be replaced by two additions.
During testing in debug builds playing back a movie went from 20fps to
300 fps.
Reviewed By: zeddb
Differential Revision: https://developer.blender.org/D11533
Talked with Bastien and we ended up looking into this. Issue is that the
dupliation through drag & drop should also be considered a
"sub-process", like Shift+D duplicating does. Added a comment explaining
why this is needed.
Current index builder is designed to be used in a single thread.
This makes all index buffer extractions single threaded.
This patch adds a thread safe solution enabling multithreaded
building of index buffers.
To reduce locking the solution would provide a task/thread local
index buffer builder (called sub builder).
When a thread is finished this thread local index buffer builder
can be joined with the initial index buffer builder.
`GPU_indexbuf_subbuilder_init`: Initialized a sub builder. The
index list is shared between the parent and sub buffer, but the
counters are localized. Ensuring that updating counters would
not need any locking.
`GPU_indexbuf_subbuilder_finish`: merge the information of the
sub builder back to the parent builder. Needs to be invoked outside
the worker thread, or when sure that all worker threads have been
finished. Internal the function is not thread safe.
For testing purposes the extract_points extractor has been migrated to
the new API. Herefore changes to the mesh extractor were needed.
* When creating tasks, the task number of current task is stored in
ExtractTaskData including the total number of tasks.
* Adding two functions in `MeshExtract`.
** `task_init` will initialize the task specific userdata.
** `task_finish` should merge back the task specific userdata back.
* adding task_id parameter to the iteration functions so they can
access the correct task data without any need for locking.
There is no noticeable change in end user performance.
Reviewed By: mano-wii
Differential Revision: https://developer.blender.org/D11499
This node creates a boolean face attribute that is "true" for
every face that has the given material.
Differential Revision: https://developer.blender.org/D11324
Selection of an FCurve with box/circle select now selects the entire
curve and all its keys:
- Box selecting a curve selects all the keyframes of the curve.
- Ctrl + box selecting of the curve deselects all the keyframes of the
curve.
- Shift + box selecting of the curve extends the keyframe selection,
adding all the keyframes of the curves that were just selected to the
selection.
- In all cases, if the selection area contains a key, nothing is
performed on the curves themselves (the action only impacts the
selected keys).
Reviewed By: sybren, #animation_rigging
Differential Revision: https://developer.blender.org/D11181
Some of the primitive nodes can return null in an error condition.
This is confusing mixed with adding a maderial slot in calling
functions. This is the second crash caused by that confusion. It's
simpler to add the slot right when allocating the mesh, and it will
lend itself better to copy & paste coding in the future.
Differential Revision: https://developer.blender.org/D11530
Under some circumstances using task isolation can cause deadlocks.
Previously, our task pool implementation would run all tasks in an
isolated region. Now using task isolation is optional and can be
turned on/off for individual task pools.
Task pools that spawn new tasks recursively should never enable
task isolation. There is a new check that finds these cases at runtime.
Right now this check is disabled, so that this commit is a pure refactor.
It will be enabled in an upcoming commit.
This fixes T88598.
Differential Revision: https://developer.blender.org/D11415
Simplify vertex normal calculation by moving the main normal
accumulation function to operate on vertices instead of faces.
Using faces had the down side that it needed to zero, accumulate and
normalize the vertex normals in 3 separate passes, accumulating also
needed a spin-lock for thread since the face would write it's normal
to all of it's vertices which could be shared with other faces.
Now a single loop over vertices is performed without locking.
This gives 5-6% speedup calculating all normals.
This also simplifies partial updates, fixing a problem where
all connected faces were being read from when calculating normals.
While this could have been resolved separately,
it's simpler to operate on vertices directly.
Move BMesh conversion and all loading code into worker.
Reviewed By: Sebastian Parborg (zeddb)
Differential Revision: https://developer.blender.org/D11288
ChildWindowFromPoint retrieves the child of the provided window at a
point. In this case it always returns 0 because HWND_DESKTOP is flag
defined as 0, which is never a valid window handle and is not intended
for use in place of a window handle.
Forwarding of mousewheel events was added in adb08def61, and later
modified to the current unworking state in e9645806f5. Sending mouse
wheel events to the window under the cursor is a system preference and
therefore should not be overridden by Blender, therefore the noop code
has been removed.
Added a new api function to stich multires grids
on specific faces in a mesh,
subdiv_ccg_average_faces_boundaries_and_corners,
and changed multires normal calc to use it.
VTune profiling showed that this was a major
performance hit once you get above 10,000 or so
base mesh faces and/or have a high number of
subdivision levels.
Here's a video comparing the difference. Note the
bpy.app_debug switch is not in the final commit.
{F10145323}
And the .blend file:
{F10145346}
Reviewed By: Sergey Sharybin (sergey)
Differential Revision:
https://developer.blender.org/D11334
Added a new api function to stich multires grids
on specific faces in a mesh,
subdiv_ccg_average_faces_boundaries_and_corners,
and changed multires normal calc to use it.
VTune profiling showed that this was a major
performance hit once you get above 10,000 or so
base mesh faces and/or have a high number of
subdivision levels.
Here's a video comparing the difference. Note the
bpy.app_debug switch is not in the final commit.
{F10145323}
And the .blend file:
{F10145346}
Reviewed By: Sergey Sharybin (sergey)
Differential Revision:
https://developer.blender.org/D11334
Also use Curve as an argument instead of Object, since the object was
only used to retrieve the curve, and the calling code is already working
with curve data.
Some of the comments referenced code that was no longer there, or even
defines that were removed. Other comments were more confusing and
vague than helpful. Also adjust formatting in a few cases.
This patch improves the positioning of the little mnemonic underlines
shown under some hotkey letters in menus, especially when using custom
fonts.
see D11521 for details and examples.
Differential Revision: https://developer.blender.org/D11521
Reviewed by Campbell Barton
This revert went too far, only one line (the minimal version of FFMPEG
for `install_deps.sh` script`) actually needed to be reverted...
Sorry for the noise.
No functional changes as logic elsewhere already ensured this.
This just makes it obvious to anyone reading over the code that
these arguments are keyword only.
FOV was expanded to cover the shifting range,
rather than to precisely cut at the image border. Now fixed.
Reviewed By: Sebastian Parborg (zeddb)
Differential Revision: https://developer.blender.org/D11523
Moving the bounds code to the builder can be useful
for future optimizations like building multithreaded.
Reviewed By: fclem, jbakker
Differential Revision: https://developer.blender.org/D11455
Small bug that's causing edge count to be incorrect in
final culled list, just being offset exactly 1 entry.
Reviewed By: Sebastian Parborg (zeddb)
Differential Revision: https://developer.blender.org/D11513
This field was used for extend feature to get handle position of
metastrip children. Since D9972 extend feature works only on meta
strip itself, not it's children.
So `SEQ_transform_get_left_handle_frame()` second argument is always
false and can be removed.
Another instance of `seq->tmp usage` is hack to distinguish strips to be
shuffled, which is not covered by this patch.
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D10321
Use bounding box tests quickly tell that two components cannot
have a containment relation between each other. This change
cut about 0.6s off a test with 25 big icospheres.
Add a call to orthogonalize the matrix before processing for the
same reasons as D8915, and an early exit in case no limits are
enabled for a bit of extra efficiency.
Since the constraint goes through Euler decomposition, it would
in fact remove shear even before this change, but the resulting
rotation won't make much sense.
This change allows using the constraint without any enabled limits
purely for the purpose of efficiently removing shear.
Differential Revision: https://developer.blender.org/D9626
Since Limit Rotation is based on Euler decomposition, it should allow
specifying the order to use for the same reasons as Copy Rotation does,
namely, if the bone uses Quaternion rotation for its animation channels,
there is no way to choose the order for the constraint.
Ref D9626
Fixed the logic for seeking in ffmpeg video files.
The main fix is that we now apply a small offset in ffmpeg_get_seek_pos
to make sure we don't get the frame in front of the seek position when
seeking backward.
The rest of the changes is general cleanup and untangling code.
Reviewed By: Richard Antalik
Differential Revision: http://developer.blender.org/D11492
Due to the way we ship the CRT on windows TBB's
malloc proxy was unable to attach it self to
the memory management functions on windows 10.
This change moves ucrtbase.dll out of the blender.crt
folder and back into the main blender folder to side
step some undesirable behaviour on win10 making TBB
once more able to attach it self.
Having this work again, should give a speed
boost in memory allocation heavy workloads
such as mantaflow.
For details on how this only failed on Win10
see T88813
This patch exposes functionality for performing partial mesh updates
for normal calculation and face tessellation while transforming a mesh.
The partial update data only needs to be generated once,
afterwards the cached connectivity information can be reused
(with the exception of changing proportional editing radius).
Currently this is only used for transform, in the future it could be
used for other operators as well as the transform panel.
The best-case overall speedup while transforming geometry is about
1.45x since the time to update a small number of normals and faces is
negligible.
For an additional speedup partial face tessellation is multi-threaded,
this gives ~15x speedup on my system (timing tessellation alone).
Exact results depend on the number of CPU cores available.
Ref D11494
Reviewed By: mano-wii
Cycles, Eevee, OSL, Geo, Attribute
This operator provides consistency with the standard math node. Allows users to use a single node instead of two nodes for this common operation.
Reviewed By: HooglyBoogly, brecht
Differential Revision: https://developer.blender.org/D10808
rB847579b42250 updated the TBB build script
which had some unintended consequences for
windows as the directory layout slightly
changed.
This change adjusts the builder to the new
structure, there are no version/functional
changes.
When changing to another texture paint slot, the texture displayed in
the viewport should change accordingly (as well as the image displayed
in the Image Editor).
The procedure to find the texture to display in the viewport
(BKE_texpaint_slot_material_find_node) could fail
though because it assumed iterating nodes would always happen in the
same order (it was index based). This is not the case though, nodes can
get sorted differently based on selection (see ED_node_sort).
Now check the actual image being referenced in the paint slot for
comparison.
ref T88788 (probably enough to call this a fix, the other issue(s)
mentioned in the report are more likely a feature request)
Reviewed By: mano-wii
Maniphest Tasks: T88788
Differential Revision: https://developer.blender.org/D11496
The (tracking) camera presets have not been updated in the last 7 or
more years, so they are very outdated. I found it pointless to have a
few specific camera models in the list and instead add the most commonly
used sensor sizes/film sizes. This way the list is shorter, easier to
maintain/becomes later outdated, and is more user friendly for most people
who don't own any of the specific cameras. I added the Crop Factor to the
Beginning of the name, so it gets sortet in the correct order and presets
are easier to find based on the size.
Reviewed By: #render_cycles, #motion_tracking, brecht, sergey
Differential Revision: https://developer.blender.org/D10739
These were only showing in the Properties Editor, but there is no reason
to have the panels be different in the sidebar (they should not show in
the top bar though).
agreed upon by both @anoniov and @mendio
ref T88787
So far, linked IDs were not properly sorted at all, only the ones
explicitely linked from WM code would be, but any indirectly linked
data-blocks would end up in some random order in their lists.
While not ideal, this is not a huge issue in itself, but it had bad
side-effects, e.g. causing (recursive) resync of overrides to happen in
random order, leading to mismatches between name indices of newly-generated
override IDs and the one existings e.g.
And in general, it is much better to be consistent here.
Note that the file sub-version is bumped for this commit, since some
sorting (the directly linked IDs which we keep a reference to) should
never need to be re-done after relevant doversion process.
This commit adds a node that outputs the total length of all
evalauted curve splines in a geometry set as a float value.
Differential Revision: https://developer.blender.org/D11459
This commit implements support for deleting curve data in the geometry
delete node. Spline domain and point domain attributes are supported.
Differential Revision: https://developer.blender.org/D11464
This implements T87633
This overlay renders a flash animation on the target object when
transfering the mode to it using the mode transfer operator.
This provides visual feedback when switching between objects without
extra overlays that affect the general color and lighting in the scene.
Differences with the design task:
- This uses just a fade out animation instead of a fade in/out animation.
The code is ready for fade in/out, but as the rest of the overlays
(face sets, masks...) change instantly without animation, having a fade
in/out effect gives the impression that the object flashes twice (once
for the face sets, twice for the peak alpha of the flash animation).
- The rendering uses a flat color without fresnel for now, but this can
be improved in the future to make it look more like the shader in the
prototype.
- Not enabled by default (can be enabled in the overlays panel), maybe
the defaults can change for 3.0 to disable fade inactive and enable this
instead.
Reviewed By: jbakker, JulienKaspar
Differential Revision: https://developer.blender.org/D11055
This operator is needed in some cases to update image preview.
In workspaces with smaller timelines this is limiting, because users
need to first check that mouse cursor is in correct place, then press
CTRL+R shortcut.
The Wayland server will not update hidden surfaces. This will block the
main event loop and thus also block updates to visible windows in a multi-
window setup.
Keyboard click-drag events now use the "Drag Threshold".
This resolves a problem where keyboard click drag events
used a much smaller threshold when using a tablet.
In some cases e.g. only objects would actually need resync, so
collections on the override character would not be resynced, and if some
objects were sharing relationships with others those could be
lost/destroyed.
max unneccessarily
Since rB298d5eb66916 [which was needed to update buttons with custom
property range functions correctly], using tab would always clamp
(hardmin/hardmax) properties which were using FLT_MAX / INT_MAX as range
in their property definitions.
The clamping of rB298d5eb66916 was copied over from rB9b7f44ceb56c
[where it was used for the softmin/softmax], and while the re-evaluation
of hardmin/hardmax is needed for custom property range functions, the
clamping should actually not take place.
There are many properties using FLT_MAX / INT_MAX etc. and while it
probably would be good to update these with ranges that make more sense
-- not using FLT_MAX / INT_MAX would not have done the clamping here --
there should not be an arbitrary limit to these and they should stay as
they are.
Maniphest Tasks: T88762
Differential Revision: https://developer.blender.org/D11473
Ensure 'virtual' linked override IDs generated by the recursive resync
process are tagged as indirectly linked data.
This is needed to avoid the 'missing data' messages on those virtual
data-blocks after saving and reloading.
`BKE_main_collections_parent_relations_rebuild`,
`BKE_collection_parent_relations_rebuild` anf their internal
dependencies had two issues fixed by this commit:
* Main one was that a same collection could be processed several times,
sometimes even in an infinite loop (in some rare corner cases), by
`collection_parents_rebuild_recursive`.
* More exotic, code here would not ensure that the collections it was
processing were actually in Main (or a master one from a scene in
Main), which became an issue with some advanced ID management
processes involving partially out-of-main remapping, like liboverride
resync.
Use SEQ_time_strip_intersects_frame function to test if strip intersects with frame.
Note: There are cases where this function should not be used. For example splitting
strips require at least 1 frame "inside" strip. Another example is drawing, where
playhead technically doesn't intersect strip, but it is rendered, because current
frame has "duration" or "thickness" of 1 frame.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11320
Remove unused flag `SEQ_DUPE_ANIM` and code used by this flag.
Remove flag `SEQ_DUPE_CONTEXT` and refactor code, to split operator
logic from duplication code.
Reduce indentation level in for loop.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11318
Images with 4:2:2 and 4:4:4 chroma subsampling were blurred when
`SWS_FAST_BILINEAR` interpolation is set for `anim->img_convert_ctx`.
Use `SWS_BILINEAR` interpolation for all movies, as performance is
not impacted by this change.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11457
Changes in rBce649c73446e, affected established proxy codec preset.
Presets were not working and all presets were similar to `veryfast`.
Tunes are now working too, so `fastdecode` tune can be used. I have
measured little improvement, but I tested this only on 2 machines and
I have been informed that `fastdecode` tune does influence decoding
performance for some users.
Change preset from `slow` to `veryfast` and add tune `fastdecode`
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11454
While encoder parameters for lossless encoding are set correctly,
output is not lossless due to pixel format being set to
`AV_PIX_FMT_YUV420P` which is inherently lossy due to chroma subsampling.
This was reported in T61569 and was merged to T57397, but there were
2 bugs - one for encoding and one for decoding.
Set pixel format to `AV_PIX_FMT_YUV444P` when rendering lossless H264
files. This format isn't available in `codec->pix_fmts[0]` and it looks,
that it has to be hard-coded.
Reviewed By: sergey
Differential Revision: D11458
This patch is from erik85, who says:
This patch makes populate_plane inside polymesh_from_trimesh_with_dissolve run in parallel.
On a test file with a boolean between two subdivided cubes (~6 million verts) this gives a 10% speed increase (49.5s to 45s) on my 6 core CPU.
Also there is an optimization of other_tri_if_manifold to skip the contains-call and get the pointer directly.
This reduces CPU time for find_patches from 5s to 2.2s on the same test file.
When there are many components (separate pieces of connected mesh),
a part of the algorithm to determine component containment was slow.
Using a float version of finding the nearest point on a triangle
as a prefilter sped this up enormously. A case of 25 icospheres
subdivided twice goes 11 seconds faster on my Macbook pro with this
change.
This just moves a couple of files in `space_node` to C++ and fixes
related errors.
The goal is to be able to use C++ data structures to simplify the code.
Differential Revision: https://developer.blender.org/D11451
The mesh to curve node generated an empty curve because no edges were
selected. This commit changes that node to not add a curve in that case.
This also changes the curve to mesh node to not add a material when
no mesh was created. Even though we don't expect null curves or meshes
in this case, the change is harmless.
EEVEE uses hashing to sync aov names and types with the gpu.
For the type a hashed value was overridden making `decalA`
and `decalB` choose the same hash. This patches fixes this
by removing the most significant bit.
EEVEE uses hashing to sync aov names and types with the gpu. For the type a hashed value was overridden making `decalA` and `decalB` choose the same hash. This patches fixes this by removing the most significant bit.
Sometimes is required to reset the thickness or the opacity of the strokes. Actually this was done using a modifier, but this operators solves this.
Reviewed By: mendio, filedescriptor
Maniphest Tasks: T87427
Differential Revision: https://developer.blender.org/D11453
Often you need to copy a spline to do an operation, but don't want
to manually copy over all of the settings. I've already forgotten to
do it once anyway. These functions copy a spline's "settings" into a
new spline, but not the data. I tried to avoid duplicating any copying
so this is easier to extend in the future.
Differential Revision: https://developer.blender.org/D11463
This is due to cam->obmat precision issue,
where it affects view vector precision.
Reviewed by Sebastian Parborg (zeddb)
Differential Revision: https://developer.blender.org/D11461
Often it would be beneficial to avoid the virtual array implementation
in `geometry_component_curve.cc` that flattens an attribute for every
spline and instead read an attribute separately for every input spline.
This commit implements functions to do that.
The downside is some code duplication-- we now have two places handling
this conversion. However, we can head in this general direction for the
attribute API anyway and support accessing attributes in smaller
contiguous chunks where necessary.
No functional changes in this commit.
Differential Revision: https://developer.blender.org/D11456
Use the D-key to access the view menu instead of the Tilda key,
which isn't accessible on some international layouts.
To resolve the conflicts the following changes have been made.
- `D` (motion) opens the view menu.
- `D` (mouse-button) uses grease-pencil (as it does currently).
- `Tilda` is used to for "Object Mode Transfer" instead of the View menu.
See T88092 for details.
Reviewed By: Severin
Ref D11189
Add a doc-string explaining the purpose of each call back and
how they should be used.
Also add a currently unused callback 'POST_FAIL' that is to be used in
cases the action fails - giving script authors, a guarantee that a
call to `pre` will always have a matching `post/post_fail` call.
- D11422: adds a callback that can use 'post_fail'.
- T88696: proposed these conventions.
This node is similar to the mask modifier, but it deletes the elements
of the geometry corresponding to the selection, which is retrieved as
a boolean attribute. The node currently supports both mesh and point
cloud data. For meshes, which elements are deleted depends on the
domain of the input selection attribute, just like how behavior depends
on the selection mode in mesh edit mode.
In the future this node will support curve data, and ideally volume
data in some way.
Differential Revision: https://developer.blender.org/D10748
Metadata panel was visible in each category. In other editors, this
panel is usually placed in category with other source media properties.
In sequencer, metadata is transfered over while compositing and relation
to particular strip is lost, therefore separate category for metadata
seems to be best option. Since Metadata panel is alone in this category,
it will be open by default.
This commit skips the eager recalculation of mesh normals in the
transform node. Often another deformation or topology-altering
operation will happen after the transform node, which means the
recalculation was redundant anyway.
In one of my test cases this made the node more than 14x faster.
Though depending on the situation the cost of updating the normals
may just be shifted elsewhere.
This was reported for the special case of mapping with "Strand /
Particle" coords, but was not working with other coordinates either.
Dont see a reason for not supporting Size influence textures for these
kinds of particles (and since these types of particles have an "age"
like all others as well, even the "Strand / Particle" coords are
supported here as well)
Maniphest Tasks: T88715
Differential Revision: https://developer.blender.org/D11449
This fixes T88455 by adding an empty material slot to newly
generated meshes. This allows the object to overwrite the
"default" material without any extra nodes. Technically,
all polygons reference the material index 0 already, so it
makes sense to add a material slot for this material index.
Differential Revision: https://developer.blender.org/D11439
Removing scripts that were placed in the source tree that would drive
the old buildbot. With the new buildbot in place these scripts aren't
being used anymore.
The buildbit is currently driven by
`build_files/config/pipeline_config.json`. In the near future make
update would also use this config.
Overview of the new buildbot: https://wiki.blender.org/wiki/Infrastructure/BuildBot
Work done by James Monteath.
`bvhtree_from_mesh_edges_create_tree` can actually leave the BVHTree
NULL (e.g. if no edges are present).
Now dont allocate `BVHTreeFromMesh` on the `SurfaceModifierData` at all
in case the tree would be NULL anyways.
Places like `get_effector_data` check for `SurfaceModifierData`-
>`BVHTreeFromMesh` and dont try to stuff like getting a closest point on
surface, which would crash as soon as BVHNodes would need to be accessed
(from the NULL BVHTree).
Maniphest Tasks: T88658
Differential Revision: https://developer.blender.org/D11443
This commit fixes two different issues:
* In some cases, when an object was added to a sub-collection and used
into a different subcollection, and the root common collection would
not need to be resynced, it would end up creating multiple overrides
of the new object. This was affecting both normal and recursive
resync.
* In recurisve resync case, the barrier code to define what is part or
not of a override group/hierarchy was wrong.
Note that the current solution for the first issue is sub-optimal (it
goes back to the root of the override hierarchy and resync the whole
thing), a better solution is TODO for now.
This function would considere that there was a name conflict even in
case existing ID would be a linked one.
This is only a (symbolic) perforance improvement and logical fix, since
`BKE_id_new_name_validate` would not do that mistake anyway.
This is mandatory for liboverride resync, since this feature may imply
we have to create linked overrides in libraries, and there may be
several copies of those.
This is also a first step to a more general support of IDmanagement-editing
library data.
Note that this commit should have absolutely no effect on current code,
as the only function allowed to check unique names for linked IDs
currently is `BKE_libblock_management_main_add`, which is unused.
This commit also adds some basic testing for `BKE_id_new_name_validate`.
This patch adds the base code needed to make the full-frame system work for both current tiled/per-pixel implementation of operations and full-frame.
Two execution models:
- Tiled: Current implementation. Renders execution groups in tiles from outputs to input. Not all operations are buffered. Runs the tiled/per-pixel implementation.
- FullFrame: All operations are buffered. Fully renders operations from inputs to outputs. Runs full-frame implementation of operations if available otherwise the current tiled/per-pixel. Creates output buffers on first read and free them as soon as all its readers have finished, reducing peak memory usage of complex/long trees. Operations are multi-threaded but do not run in parallel as Tiled (will be done in another patch).
This should allow us to convert operations to full-frame in small steps with the system already working and solve the problem of high memory usage.
FullFrame breaking changes respect Tiled system, mainly:
- Translate, Rotate, Scale, and Transform take effect immediately instead of next buffered operation.
- Any sampling is always done over inputs instead of last buffered operation.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11113
draw_cache_extract_mesh for task scheduling. Will be refactored to draw_cache_extract_mesh_scheduling later on after migrating to CPP.
draw_cache_extract_mesh_render_data extraction of mesh render data from edit mesh/mesh into a more generic structure.
draw_cache_extract_mesh_extractors containing all the extractors. This will be split up further into a single file per extractor.
This was kept since these blocks are easier to follow.
Remove as the overall result wasn't so readable
(especially with nested ifdef's).
Replace disabled code with comment on the indices used for quads/tris.
2 sided faces aren't supported and will cause problems in many areas
of Blender's code.
Removing (implied) support for faces with fewer than 3 sides
means the total number of triangles is known ahead of time.
This simplifies adding support for multi-threading and partial updates
to an existing tessellation - as the face and loop indices can be used
to access the range of triangles associated with a face.
Also correct outdated comments.
When projecting into screen space Z value isn't always needed.
Add 2D projection functions, renaming them to avoid accidents
happening again.
- Add GPU_matrix_project_2fv
- Add ED_view3d_project_v2
- Rename ED_view3d_project to ED_view3d_project_v3
- Use the 2D versions of these functions when the Z value isn't used.
This is from patch D11432 from Erik Abrahamsson. He found that
in some mpq3 functions called frequently from loops, passing in
buffers for termporary mpq3 values can save substantial time.
On my machine, his example in that patch went from 9.48s to 7.50s
for the boolean part of the calculation. On his machine, a running
time went from 17s to 10.3s.
The fseek() function on Windows only accepts a 32-bit long offset
argument. Because of this we have our own version, BLI_fseek(), which
will use 64-bit _fseeki64() on Windows. This patch just replaces some
fseek() calls with BLI_fseek().
Differential Revision: https://developer.blender.org/D11430
Reviewed by Brecht Van Lommel
Now FPS is displayed in the video source for videos to provide easy
access.
Reviewed By: Richard Antalik
Differential Revision: http://developer.blender.org/D11441
bf_bmesh historically always build with the /WX flag
on windows making all warnings errors, somewhere along
the way this has broken for msbuild, ninja still exhibits
the expected behaviour.
The flags are still passed to the target, and I've validated
they are there when the add_library call fires, but they
somehow never make it to the generated msbuild project files.
I suspect this is a cmake bug but I'm seemingly unable
to extract a repro case to file a bug upstream.
Setting the same options target_compile_options seems to work,
I'm not happy about the unexplained nature of the breakage
but this will have to do for now.
The same code existed in 2.82 and earlier so this should be safe. Removing the
custom implementation of shutil.copytree in f34d5d9 did not correctly add back
the option to copy symlinks.
This patch replaces / redoes the entire MeshExtractors system.
Although they were useful and facilitated the addition of new buffers, they made it difficult to control the threads and added a lot of threading overhead.
Part of the problem was in traversing the same loop type in different threads. The concurrent access of the BMesh Elements slowed the reading.
This patch simplifies the use of threads by merging all the old callbacks from the extracts into a single series of iteration functions.
The type of extraction can be chosen using flags.
This optimized the process by around 34%.
Initial idea and implementation By @mano-wii.
Fine-tuning, cleanup by @atmind.
MASTER:
large_mesh_editing:
- rdata 9ms iter 50ms (frame 155ms)
- Average: 6.462874 FPS
PATCH:
large_mesh_editing:
- rdata 9ms iter 34ms (frame 136ms)
- Average: 7.379491 FPS
Differential Revision: https://developer.blender.org/D11425
For 2.93 we bumped the minimum windows requirement
to windows 8.1, but did not do any clean-up of any
win 8/8.1 API usage we dynamically accessed though
LoadLibrary/GetProcAddress.
This patch bumps _WIN32_WINNT to 0x0603 (win 8.1)
and cleans up any API use that was accessed in a
more convoluted way than necessary
Differential Revision: https://developer.blender.org/D11331
Reviewed by: harley, nicholas_rishel
Use the appropriate notifier, listeners were already doing the rest
properly.
Maniphest Tasks: T88569
Differential Revision: https://developer.blender.org/D11436
Cause is that initializing the cryptomatte session would reset the
current frame of an image sequence. The solution is to always use the
scene current frame so it resets to the correct frame.
This was a todo that wasn't solved after it landed in master.
Needs to be backported to 2.93.
When compositor node tree has a texture node, TextureOperation vector inputs has always {0, 0} resolution instead of having same resolution as TextureOperation which is the expected behaviour for resolutions propagation.
Current TextureOperation determineResolution implementation doesn't determine inputs resolution, breaking propagation of preferred resolution and that's the reason why they are always 0. Setting scene resolution always would mean it is its own resolution and could make sense, but setting it only when preferred resolution is 0, breaks preferred resolution logic affecting other operations as explained in D10972. In any case scene resolution is already the default preferred resolution on viewer and compositor nodes.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11381
The issue was two fold. We didn't properly:
1. Initialize the codec default values which would lead to VLC
complaining because of garbage/wrong codec settings.
2.Calculate the time base for the video. FFmpeg would happily accept
this but VLC seems to assume the time base value is at least somewhat
correct and couldn't properly display the frames as the internal time
base was huge. We are talking about 90k ticks (tbn) for one second of
video!
This patch initializes all codecs to use their default values and fixes
the time base calculation so it follows the guidelines from ffmpeg.
Reviewed By: Sergey, Richard Antalik
Differential Revision: http://developer.blender.org/D11426
Code checking for potential collection loop dependencies can be called
in cases where we cannot guarantee that there is no NULL pointers, so we
need to check those. Was already done for objects.
NOTE: doubled-checked by @jbakker, thanks.
Code checking for potential collection loop dependencies can be called
in cases where we cannot guarantee that there is no NULL pointers, so we
need to check those. Was already done for objects.
Reuse loose geometry during selection (and other operations) from
previous calculation. Loose geometry stays the same, but was
recalculated to determine the size of GPU buffers. This patch would
reuse the previous loose geometry when geometry wasn't changed.
Although not the main bottleneck during selection it is measurable.
Master.
`rdata 46ms iter 55ms (frame 410ms)`
This patch.
`rdata 5ms iter 52ms (frame 342ms)`
Reviewed By: mano-wii
Differential Revision: https://developer.blender.org/D11339
This patch from Erik Abrahamsson uses a parallel_for to speed up
the case where the input is not manifold and the "hole_tolerant"
option is set.
In a test case on a 24 core (48 thread) machine, this sped up a
the boolean part on an object with 221k triangles from 12.06s to 0.46s.
When using ``Path`` alignment, if the stroke has one point the texture rotates randomly when move the viewport. This was because with one point is impossible to calculate a path.
Now, if the stroke has only one point, the texture for this stroke is aligned to Object.
Before the FFmpeg commit: github.com/FFmpeg/FFmpeg/commit/1c0885334dda9ee8652e60c586fa2e3674056586
FFmpeg would use deprecated variables to calculate the video fps.
We don't use these deprecated variables anymore, so ensure that the
duration is correct in ffmpeg versions without this fix.
Reviewed By: Sergey, Richard Antalik
Differential Revision: http://developer.blender.org/D11417
We didn't initialize the scaled proxy frame properly.
This would lead to issues in ffmpeg 4.4 as they are more strict that the API is properly used.
Now we initialize the size and format of the frame.
Based on the task T88006, there are a few simple changes
to make to improve the switch node:
- Change the label to "False" / "True" for clarity
- Change default to geometry, as it's the basic data container in
geometry nodes.
- Change node class to `NODE_CLASS_CONVERTOR`, which was an oversight
in the original patch.
I will add the new socket types (material and texture) in a separate commit.
Thanks to @EitanSomething for the original patch.
Differential Revision: https://developer.blender.org/D11165
This was caused by unsafe sqrt calls.
Fixes T86578 white artifacts in EEVEE
Reviewed By: brecht, dfelinto
Differential Revision: https://developer.blender.org/D11428
While we could make this node work for boolean inputs in the future,
currently it's really just designed to compare "float-like" inputs.
Many comparison modes don't even make sense for boolean inputs.
Therefore, the simplest fix for this bug is just to disable the
boolean attribute input modes for this node.
Differential Revision: https://developer.blender.org/D11427
When using ``Path`` alignment, if the stroke has one point the texture rotates randomly when move the viewport. This was because with one point is impossible to calculate a path.
Now, if the stroke has only one point, the texture for this stroke is aligned to Object.
This node creates poly curve splines from mesh edges. A selection
attribute input allows only using some of the edges from the mesh.
The node builds cyclic splines from branchless groups of edges where
possible, but when there is a three-way intersection, the spline stops.
The node also transfers all attributes from the mesh to the resulting
control points. In the future we could add a way to limit that to a
subset of the attributes to improve performance.
The algorithm is from Animation Nodes, written by @OmarSquircleArt.
I added the ability to use a selection, attribute transferring, and
used different variable names, etc, but other than that the algorithm
is the same.
Differential Revision: https://developer.blender.org/D11265
Not entirely sure why this was not an issue for 16.9
but TBB includes the Windows.h header which by default
will define min and max macro's
These collide with the stl versions in <algorithm>
This patch requests Windows.h not to define the
problematic macro's, resolving the conflict.
Caused by {rB66923031e6f2}.
Code would process unselected sequences and skip selected, needs to be
the other way around.
Maniphest Tasks: T88635
Differential Revision: https://developer.blender.org/D11424
When using the operator `ui.copy_data_path_button(full_path=True)` ({key
ctrl shift Alt C} on hover) the copied path does not consider the
library origin. That means that when there is a name clash the data path
is not accurate and refers to the local item instead.
This patch adds the library (if the ID is linked) of the returned string
from RNA_path_full_ID_py.
bpy.data.objects["Cube", "//library.blend"] instead of
bpy.data.objects["Cube"]
note: parsing this happens in
pyrna_prop_collection_subscript_str_lib_pair_ptr
Maniphest Tasks: T88499
Differential Revision: https://developer.blender.org/D11412
A utility that supports passing in actions as command line arguments for
writing reproducible interactions, benchmarking, profiling and testing.
Unlike regular scripts this is able to control model operators usefully.
Typical ways of controlling Blender using this utility are via
operator id's, menu search and explicit events.
Others methods can be added as needed.
See the doc-string for example usage.
This patch will use compute shaders to create the VBO for hair.
The previous implementation uses transform feedback.
Timings before: between 0.000069s and 0.000362s.
Timings after: between 0.000032s and 0.000092s.
Speedup isn't noticeable by end-users. The patch is used to test
the new compute shader pipeline and integrate it with the draw
manager. Allowing EEVEE, Workbench and other draw engines to
use compute shaders with the introduction of `DRW_shgroup_call_compute`
and `DRW_shgroup_vertex_buffer`.
Future improvements are possible by generating the index buffer
of hair directly on the GPU.
NOTE: that compute shaders aren't supported by Apple and still use
the transform feedback workaround.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D11057
Added missing topbar in VSE.
Also added the Stabilizer options to Topbar for all modes.
Reviewed By: mendio, pepeland
Maniphest Tasks: T86465
Differential Revision: https://developer.blender.org/D11347
We need to re-evaluate what needs to be resynced after each step of
processing overrides from a given 'indirect level' of libraries.
Otherwise, recusrive overrides (overrides of linked overrides) won't
work.
Note that this should not change too much in practice currently, since
there are other issues with recursive overrides yet.
Also, checks (CLOG errors) added show that some ID (node trees) seem to
be detected as needing resynced even after beig just resynced, this
needs further investigation still. Could be though that it is due to
limit currently set on nodetrees, those are always complicated
snowflakes to deal with...
Oversight in {rB470f17f21c06}.
Hiding was only done for the first mesh, then the operator finished (in
case of UV_SYNC_SELECTION).
Now just continue to the next.
Maniphest Tasks: T88625
Differential Revision: https://developer.blender.org/D11413
Both before and after can have artifacts with some normal maps, but this seems to give
worse artifacts on average which are not worth the minor performance increase.
This reverts commit 5c4d24e1fd.
Ref T88368, D10084
This allows choosing material and texture sockets for the group input
node in the modifier. Note that currently grease pencil materials are
displayed in the list, even though grease pencil data is not supported
yet by geometry nodes. That is more complicated to fix in this case,
since we use IDProperties to store the dynamic exposed inputs.
Differential Revision: https://developer.blender.org/D11393
blender-laucher.c was not an ideal name for this file
since it's not directly clear it is windows only.
This change renames it to blender_launcher_win32.c
to be more in line with other win32 specific files
we have.
This patch adds relatively small changes to the curve draw
cache implementation in order to draw the curve data in the
viewport. The dependency graph iterator is also modified
so that it iterates over the curve geometry component, which
is presented to users as `Curve` data with a pointer to the
`CurveEval`
The idea with the spline data type in geometry nodes is that
curve data itself is only the control points, and any evaluated
data with faces is a mesh. That is mostly expected elsewhere in
Blender anyway. This means it's only necessary to implement
wire edge drawing of `CurveEval` data.
Adding a `CurveEval` pointer to `Curve` is in line with changes
I'd like to make in the future like using `CurveEval` in more places
such as edit mode.
An alternate solution involves converting the curve wire data
to a mesh, however, that requires copying all of the data, and
since avoiding it is rather simple and is in-line with future plans
anyway, I think doing it this way is better.
Differential Revision: https://developer.blender.org/D11351
This patch removes unnecessary calls to `BKE_main_id_tag_all` where the
same job is done by `BKE_main_id_clear_newpoins` on the following line.
Reviewed By: campbellbarton, mont29
Ref D11379
Use array instead of ListBase for line art
bounding area linked triangles and edges.
Reviewed By: Sebastian Parborg (zeddb)
Differential Revision: https://developer.blender.org/D11302
This is not supposed to happen, but better be safe than sorry, and
assume it is beyond unlikely that someone would use chains of over 10k
linked libraries.
Allows to centralize storage and modification checks in a single place,
avoiding duplication in the synchronization code.
Ideally we would somehow be able to more granularly modify Cycles side
objects. Leaving this for a future decision, because it might be better
to implement it as a graph on the sync side.
Makes it more explicit they operate on shading/light.
Gives room to move more viewport related settings into this class and
cover with specific or generic modification checks.
Such pattern should only be used when it is really needed. Otherwise
just stick to a more regular design, without worrying who is the user
of the class. Otherwise it will be annoying to subclass or unit test.
No need to state that it is a viewport display pass, since the method
is within viewport parameters it is implied that parameters do belong
to the viewport.
Brings this code closer to the Cycles-X branch.
Very stupid mistake in libraries indirect-level building code, was not
skipping 'loop-back' ID pointers.
Note that we also need some level of checks for the case where there
would be an actual dependency loop between libraries, this is not
supposed to be possible, but better be safe than sorry. Will add in next
commit.
An arbitrary size offsets was used in float_array_to_string,
simplify the loop, use exact size limits.
Also rename variables so it's clear which array the length apply to.
Event though in practice this wasn't causing problems as the fixed size
buffers are generally large enough not to truncate text.
Using the result from `snprint` or `BLI_snprintf` to step over a fixed
size buffer allows for buffer overruns as the returned value is the size
needed to copy the entire string, not the number of bytes copied.
Building strings using this convention with multiple calls:
ofs += BLI_snprintf(str + ofs, str_len_max - ofs);
.. caused the size argument to become negative,
wrapping it to a large value when cast to the unsigned argument.
A deadlock could happen under certain circumstances when
geometry nodes is used on multiple objects.
Once T88598 is resolved, multi-threading can be enabled again.
Differential Revision: https://developer.blender.org/D11405
While the advantage isn't large,
it's simpler to skip the intermediate link.
Also remove unused next and previous struct members
from MeshUndoStep_Elem.
When editing more than 1 object at a time, complete copies of each mesh
were being stored. Now the most recent undo-data for each mesh is used
(when available).
This commit adds interpolation from the point domain to the spline
domain and the other way around. Before this, spline domain attributes
were basically useless, but now they are quite helpful as a way to use
a shared value in a contiguous group of points.
I implementented a special virtual array for the spline to points
conversion, so that conversion should be close to the ideal performance
level, but there are a few ways we could optimize the point to spline
conversion in the future:
- Use a function virtual array to mix the point values for each spline
on demand.
- Implement a special case for when the input virtual array is one of
the virtual arrays from the spline point attributes. In other words,
decrease curve attribute access overhead.
Differential Revision: https://developer.blender.org/D11376
This patch fixes a long-standing complaint from users:
the console window shortly flashing when they start
blender.
This is done by adding a new executable called
blender-launcher.exe which starts blender.exe while
hiding the console.
Any command line parameters given to blender-launcher
will be passed on to blender.exe so it'll be a drop
in replacement.
Starting blender.exe on its own will still function as
a proper console app so no changes required here for
users that use blender for batch processing.
Notable changes:
Registering blender (-R switch) will now register
blender-launcher as the preferred executable.
This patch updates the installer and updates the
shortcuts to start blender-launcher.exe rather
than blender.exe
Differential Revision: https://developer.blender.org/D11094
Reviewed by: brecht, harley
This reverts commit 8f9599d17e.
Mac seems to have an error with this change.
```
ERROR: /Users/blender/git/blender-vdev/blender.git/source/blender/draw/intern/draw_hair.c:115:44: error: use of undeclared identifier 'shader_src'
ERROR: /Users/blender/git/blender-vdev/blender.git/source/blender/draw/intern/draw_hair.c:123:13: error: use of undeclared identifier 'shader_src'
ERROR: make[2]: *** [source/blender/draw/CMakeFiles/bf_draw.dir/intern/draw_hair.c.o] Error 1
ERROR: make[1]: *** [source/blender/draw/CMakeFiles/bf_draw.dir/all] Error 2
ERROR: make: *** [all] Error 2
```
Both before and after can have artifacts with some normal maps, but this seems to give
worse artifacts on average which are not worth the minor performance increase.
This reverts commit 21bc1a99ba.
Ref T88368, D10084
The skin modifier was moving vertices without updating normals for the
connected faces, this happened when smoothing and welding vertices.
Reviewed By: mont29
Ref D11397
Recursive resync means also resyncing overrides that are linked from
other library files into current working file.
Note that this allows to get 'working' files even when their
dependencies are out of sync. However, since linked data is never
written/saved, this has to be re-done every time the working file is
loaded, until said dependencies are updated properly.
NOTE: This is still missing the 'report' side of things, which is part
of a larger task to enhance reports regarding both linking, and
liboverrides (see T88393).
----------
Technical notes:
Implementing this proved to be slightly more challenging than expected,
mainly because one of the key aspects of the feature was never done in
Blender before: manipulating, re-creating linked data.
This ended up moving the whole resync code to use temp IDs out of bmain,
which is better in the long run anyway (and more aligned with what we
generally want to do when manipulating temp ID data). It should also
give a marginal improvement in performances for regular resync.
This commit also had to carefully 'sort' libraries by level of indirect
usage, as we want to resync first the libraries that are the least directly
used, i.e. libraries that are most used by other libraries.
Even though shepkeys are not strictly speaking an embedded data, they
share quiet a few points with those, and from liboverride perspective
they are embedded, so...
While indeally we should only skip refcounting when relevant tag is set,
doing this in remapping code is too risky for now.
Related to previous commit and T88555.
While this is still very fuzzy in current code, this old behavior makes
it close to impossible to efficiently use out-of-main temp data, as it
implies that we'd need to update refcounts everytime we add something
back into BMain (an 'un-refcount' ID usages when removing from BMain).
Now that we have two separate flags/tags for those two different things,
let's not merge them anymore.
Note that this is somewhat on-going process, still needs more checks and
cleanup. Related to T88555.
This patch will use compute shaders to create the VBO for hair.
The previous implementation uses tranform feedback.
Timings master (transform feedback with GPU_USAGE_STATIC between 0.000069s and 0.000362s
Timings transform feedback with GPU_USAGE_DEVICE_ONLY. between 0.000057s and 0.000122s
Timings compute shader between 0.000032 and 0.000092s
Future improvements:
* Generate hair Index buffer using compute shaders: currently done single threaded on CPU, easy to add as compute shader.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D11057
With the compute pipeline calculation can be offloaded to the GPU.
This patch only adds the framework for compute. So no changes for users at
this moment.
NOTE: As this is an OpenGL4.3 feature it must always have a fallback.
Use `GPU_compute_shader_support` to check if compute pipeline can be used.
Check `gpu_shader_compute*` test cases for usage.
This patch also adds support for shader storage buffer objects and device only
vertex/index buffers.
An alternative that had been discussed was adding this to the `GPUBatch`, this
was eventually not chosen as it would lead to more code when used as part of a
shading group. The idea is that we add an `eDRWCommandType` in the near
future.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D10913
This patch adds an option in the Layers > Relations panel called "Disable Masks in Render".
When checked, no masks on this layer are included in the render.
Example:
| {F10087680} | {F10087681} |
See T88202 for why this is needed.
Reviewed By: antoniov
Maniphest Tasks: T88202
Differential Revision: https://developer.blender.org/D11234
Currently when rendering the view layer of a grease pencil layer that has
a mask layer attached, the mask layer would show in the rendered image.
This is inconsistent with the default behaviour with no mask on the
grease pencil layer, because it would only render what's on that
particular layer and not anything from any other layer.
This patch makes the masks invisible in the render.
Note: This might seem like not the best solution, but because masks are
just regular grease pencil layers, it's tricky to pass this edge-case to the
drawing code. The way it is handled right now is the best I could come
up with, without making changes that could affect something else.
Reviewed By: antoniov
Maniphest Tasks: T88202
Differential Revision: https://developer.blender.org/D11403
This issue was that `BKE_object_eval_uber_data` was not called for
the text object, because its geometry was not dependent upon
and its `is_directly_visible` tag was `false`. The crash happens in
rendering code, because the evaluated data is missing.
This not only affects text objects, but all object types that have a
geometry component that geometry nodes does not support yet.
The solution is to just add the missing dependencies.
Differential Revision: https://developer.blender.org/D11385
This operator is a common request of animators to convert the transformation (inluding modifiers) of one grease pencil object, into a new object, generating strokes.
Reviewed By: pepeland
Maniphest Tasks: T87424
Differential Revision: https://developer.blender.org/D11014
Calling BKE_nodetree_attribute_hint_add from multiple threads still
was not safe before..
One issue was that context_map embedded its values directly. So
when context_map grows, all NodeUIStorage would move as well.
I could patch around that by using std::unique_ptr in a few places,
but that just becomes too complex for now.
Instead I simplified the locking a bit by adding just locking a mutex
in NodeTreeUIStorage all the time while an attribute hint is added.
Differential Revision: https://developer.blender.org/D11399
When an `AttributeSet` is tagged as modified, which happens after the addition or
removal of an `Attribute` from the set, during the following GeometryManager device
update, we update and repack the kernel data for all attribute types. However, if we
only add or remove a `float` attribute, `float2` or `float3` attributes should not
be repacked for efficiency.
This patch adds some mechanisms to detect which attribute types are modified from
the AttributeSet.
Firstly, this adds an `AttrKernelDataType` to map the data type of the Attribute to
the one used in the kernel as there is no one to one match between the two since e.g.
`Transform` or `float4` data are stored as `float3s` in the kernel.
Then, this replaces the `AttributeSet.modified` boolean with a set of flags to detect
which types have been modified. There is no specific flag type (e.g.
`enum ModifiedType`), rather the flags used derive simply from the
`AttrKernelDataType` enumeration, to keep things synchronized.
The logic to remove an `Attribute` from the `AttributeSet` and tag the latter as modified
is centralized in a new `AttributeSet.remove` method taking an iterator as input.
Lastly, as some attributes like standard normals are not stored in the various
kernel attribute arrays (`DeviceScene::attribute_*`), the modified flags are only
set if the associated standard corresponds to an attribute which will be stored
in the kernel's attribute arrays. This makes it so adding or removing such attributes
does not trigger an unnecessary update of other type-related attributes.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D11373
Sharing data between duplicated IDs should be restricted to depsgraph
(CoW) cases, not all NO_MAIN ones...
While this was probably not an issue currently, we aim at using more and
more out-of-main IDs for temp data processing.
NOTE: Somewhat related to T88555, and similar issue as the one fixed in
rBdfb963c70df5.
Objects modified by geometry nodes modifiers were not caught as being
"dynamic".
Now add this modifier type to the list of modifiers making them "dynamic"
in the eyes of mantaflow.
(noticed by @sebbas in chat)
Maniphest Tasks: T88531
Differential Revision: https://developer.blender.org/D11389
(regression)
Code was actually checking for shapekeys, but these were not detected
properly (some effects like shape keys are added as virtual modifiers
before the user created modifiers)
Now go over virtual modifiers as well.
Maniphest Tasks: T88566
Differential Revision: https://developer.blender.org/D11388
Caused by {rB919558854d62}.
Same fix as in {rBdc8a43c8755a} -- let RNA enum item callbacks check
for NULL context.
The NULL context is used to extract items for document generation.
Maniphest Tasks: T88251
Differential Revision: https://developer.blender.org/D11391
Bump FFmpeg version to 4.4 to fix a problem where it would write the
wrong frame rate. Their old API was deprecated and Blender moved to the
new one in rB8d6264ea12bfac0912c7249f00af2ac8e3409ed1. The new one
produced files with the wrong frame rate, which was fixed in FFmpeg 4.4.
Manifest Task: T88568
Reviewed By: LazyDodo, zeddb
Differential Revision: https://developer.blender.org/D11392
This option is default off when creating line art objects
because line art seldom use lighting and the normal data
would be all over the place anyway.
Reviewed By: Antonio Vazquez (antoniov)
Differential Revision: https://developer.blender.org/D11372
This clarifies the data structures for storing edges
for different calculation stages.
Reviewed By: Sebastian Parborg (zeddb)
Differential Revision: https://developer.blender.org/D11386
Colors are often thought of as being 4 values that make up that can make any color.
But that is of course too limited. In C we didn’t spend time to annotate what we meant
when using colors.
Recently `BLI_color.hh` was made to facilitate color structures in CPP. CPP has possibilities to
enforce annotating structures during compilation and can adds conversions between them using
function overloading and explicit constructors.
The storage structs can hold 4 channels (r, g, b and a).
Usage:
Convert a theme byte color to a linearrgb premultiplied.
```
ColorTheme4b theme_color;
ColorSceneLinear4f<eAlpha::Premultiplied> linearrgb_color =
BLI_color_convert_to_scene_linear(theme_color).premultiply_alpha();
```
The API is structured to make most use of inlining. Most notable are space
conversions done via `BLI_color_convert_to*` functions.
- Conversions between spaces (theme <=> scene linear) should always be done by
invoking the `BLI_color_convert_to*` methods.
- Encoding colors (compressing to store colors inside a less precision storage)
should be done by invoking the `encode` and `decode` methods.
- Changing alpha association should be done by invoking `premultiply_alpha` or
`unpremultiply_alpha` methods.
# Encoding.
Color encoding is used to store colors with less precision as in using `uint8_t` in
stead of `float`. This encoding is supported for `eSpace::SceneLinear`.
To make this clear to the developer the `eSpace::SceneLinearByteEncoded`
space is added.
# Precision
Colors can be stored using `uint8_t` or `float` colors. The conversion
between the two precisions are available as methods. (`to_4b` and
`to_4f`).
# Alpha conversion
Alpha conversion is only supported in SceneLinear space.
Extending:
- This file can be extended with `ColorHex/Hsl/Hsv` for different representations
of rgb based colors. `ColorHsl4f<eSpace::SceneLinear, eAlpha::Premultiplied>`
- Add non RGB spaces/storages ColorXyz.
Reviewed By: JacquesLucke, brecht
Differential Revision: https://developer.blender.org/D10978
Colors are often thought of as being 4 values that make up that can make any color.
But that is of course too limited. In C we didn’t spend time to annotate what we meant
when using colors.
Recently `BLI_color.hh` was made to facilitate color structures in CPP. CPP has possibilities to
enforce annotating structures during compilation and can adds conversions between them using
function overloading and explicit constructors.
The storage structs can hold 4 channels (r, g, b and a).
Usage:
Convert a theme byte color to a linearrgb premultiplied.
```
ColorTheme4b theme_color;
ColorSceneLinear4f<eAlpha::Premultiplied> linearrgb_color =
BLI_color_convert_to_scene_linear(theme_color).premultiply_alpha();
```
The API is structured to make most use of inlining. Most notable are space
conversions done via `BLI_color_convert_to*` functions.
- Conversions between spaces (theme <=> scene linear) should always be done by
invoking the `BLI_color_convert_to*` methods.
- Encoding colors (compressing to store colors inside a less precision storage)
should be done by invoking the `encode` and `decode` methods.
- Changing alpha association should be done by invoking `premultiply_alpha` or
`unpremultiply_alpha` methods.
# Encoding.
Color encoding is used to store colors with less precision as in using `uint8_t` in
stead of `float`. This encoding is supported for `eSpace::SceneLinear`.
To make this clear to the developer the `eSpace::SceneLinearByteEncoded`
space is added.
# Precision
Colors can be stored using `uint8_t` or `float` colors. The conversion
between the two precisions are available as methods. (`to_4b` and
`to_4f`).
# Alpha conversion
Alpha conversion is only supported in SceneLinear space.
Extending:
- This file can be extended with `ColorHex/Hsl/Hsv` for different representations
of rgb based colors. `ColorHsl4f<eSpace::SceneLinear, eAlpha::Premultiplied>`
- Add non RGB spaces/storages ColorXyz.
Reviewed By: JacquesLucke, brecht
Differential Revision: https://developer.blender.org/D10978
Using displacement runs the shader eval kernel, but since OptiX modules are not loaded when
baking is active, those were not available and therefore failed to launch. This fixes that by falling
back to the CUDA kernels.
Issue is that due to the strange definition of render in grease pencil
(meaning should be rendered similar to rendering). This included normal
viewport rendering in OB_RENDER and OpenGL render in OB_RENDER.
For other rendering modes the overlay vertex opacity would be used. This
patch sets this value to 1 when rendering via a scene strip override.
NOTE: that this isn't a good solution as I expect that users want to use
the opacity of the Grease pencil object. Perhaps the GPencil team has a
better solution for it.
Tool settings for sequencer were not initialized, which caused crash
when adding strips.
There was fix for same issue in versioning rB0f81dafe6cec, but
subversion was not bumped, so files with uninitialized tool settings
may still exist.
Add `SEQ_tool_settings_get()` accessor function that will initialize
tool settings if they are missing. Change operator code to use
`SEQ_tool_settings_fit_method_get()` function instead of accessing
tool settings directly
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D11383
This will update active camera based on the maker for each frame.
Reviewed By: Sebastian Parborg (zeddb), Antonio Vazquez (antoniov)
Differential Revision: https://developer.blender.org/D11358
Do not allow a window to be created that has a top position that
can obscure all or part of title bar. Right and Left edges can
still be specified slightly outside of monitor bounds, but top
edge must be clamped to monitor top.
see D11371 for more details.
https://developer.blender.org/D11371
Reviewed by Ray Molenkamp
This fallback is an old hack.
It is difficult to have an orientation convention when several random
factors determine which one should be used.
In this case, to "fix" the problem, a new behavior had to be implemented.
Now the redo when moving the camera in `Camera View` has the default
orientation as `View`.
Deduplicates code by introducing a PlaneDirtortBaseOperation for common logic.
Reviewed By: #compositing, jbakker
Differential Revision: https://developer.blender.org/D11273
Prepare node for conversion to Geometry Nodes.
There should be no functional changes.
Reviewed By: JacquesLucke, LazyDodo
Differential Revision: https://developer.blender.org/D11226
When sound strip is above another strip such as movie strip, it prevents
from rendering movie strip.
This bug was introduced in 0b7744f4da. Function `must_render_strip()`
checks if there is any strip with `SEQ_BLEND_REPLACE` blending and
considers this strip as lowest strip in stack. Sound strips do have this
blend mode set, which caused the bug.
Remove all sound strips and muted strips from stack collection before
checking with `must_render_strip()` function
I'd like to use this file to draw curves from geometry nodes, which
would otherwise require implementing a C API. The changes in this
commit are minimal, mostly just casts and changing to nullptr.
Differential Revision: https://developer.blender.org/D11350
The CoW copy of the node group was not updated correctly after it
changed. Arguably, tagging the node tree could also be part of
`ntreeUpdateTree` (which needs to be called after changes to the
node tree anyway). However, there are many instances where the
depsgraph is tagged explicitly after `ntreeUpdateTree` is called,
so it should be fine here as well. This is similar to what is done
in `snode_dag_update`.
Differential Revision: https://developer.blender.org/D11342
The wrong matrix function was used and overwrote the custom bone shape
scale instead of reading from it.
Reviewed By: sybren
Differential Revision: https://developer.blender.org/D11330
When creating Win32 windows, the sizes and placements can be out by a
small amount, mostly noticeable near monitor edges. This is because
Windows 10 includes a thin invisible border (typically 7 pixels) when
determining position. Therefore the correct values can sometimes be
just outside the monitor bounds, but we clamp them at those bounds.
This patch fixes this by first clamping the requested values to monitor
bounds, adjusting for window chrome with AdjustWindowRectEx(), and then
using those adjusted values in CreateWindowExW().
see D11314 for more details.
Differential Revision: https://developer.blender.org/D11314
Reviewed by Ray Molenkamp
Accessing data through RNA should never implicitely create it. Objects'
and particles' forcefields and collision settings were doing this.
Note that UI code also had to be tweaked to properly handle `None`
(NULL) cases.
Differential Revision: https://developer.blender.org/D11341
This data (the force fields) are expected to always be there, but they
are currently created on the fly by RNA accessors (typically from UI
draw code), which is extremely wrong way to do it.
Differential Revision: https://developer.blender.org/D11341
This is an initial change to speed up the calculation of the Occlude eraser. In the future, we can add more optimizations, but at least this increase speed.
Intead to check always the 3 points, the check is skipped if it's not required.
Base in a solution by Philipp Oeser.
This is related to T88412
Root of the issue was actually hidden deep in depsgraph itself: it would
not properly update all of its COW IDs using a datablock when depsgraph
decides to evaluate or un-evaluate it.
This would lead to evaluated IDs pointing to either:
- orig IDs when there was an evaluated version of those (annoying bug,
but not a crashing one).
- old address of previously evaluated IDs that no longer exists in the
depsgraph (causing the crash from the report e.g.).
This commit adds an extra step at the end of nodes building, that goes
over all of already existing IDs in the depsgraph to check whether they
do one of the two things above, and tag them for COW update if so.
NOTE: This only affects depsgraph (re-)building, not its evaluation.
This remains consistent with the fact that operations that may change
the depsgraph content (like Collection exclusion etc.) need to trigger a
rebuild.
NOTE: Performances: Worst case scenarii, like (un-)excluding a whole
character collection in a production file, lead to 5% to 10% extra
processing time in depsgraph building. Most of it comming from extra COW
processing (in depsgraph's update in `build_step_finalize`), the detection
loop itself only accounts for 1% to 2% of the whole building time.
Maniphest Tasks: T85752
Differential Revision: https://developer.blender.org/D10907
This node can change all faces that use a specific material to use a
different material. Using this node is significantly more efficient
than creating a selection from all faces with a specific material
index and then using the Material Assign node.
Ref T88055.
Differential Revision: https://developer.blender.org/D11325
Share the pointer with the original mesh instead, this matches behavior
of all other objects edit-mode data.
Duplicating the edit-mesh pointer makes updates to edit-mesh require
a COPY_ON_WRITE update, which is currently an expensive operation
(copying the entire mesh).
Notes:
- This change is from 802027f3f8
so the edit-meshes object pointer `BMEditMesh.ob` referenced the COW
version of the object. This pointer has since been removed, so the
copy is no longer needed.
- Having a separate edit-mesh pointer could be used so linked duplicates
could have their own generated meshes. For this to be supported,
many other changes would be needed: see D10920.
This patch adds wavelength node support to Eevee, similar to how
Eevee Blackbody node works, thus it is a little off from Cycles.
Reviewed By: #eevee_viewport, fclem, brecht
Differential Revision: https://developer.blender.org/D11326
Previously, the "follow path constraint" and "follow parented curve"
were clamped. This restriction was lifted in rBcf2baa585cc8
Add back an option to get the old behavior in the "Path animation" settings.
Reviewed By: Sybren
Differential Revision: http://developer.blender.org/D11263
Since version 2.80, the annotations of the Scene strip were not displayed in VSE. Also, the UI panel was`Grease Pencil` and must be `Annotation`
The problem was the offscreen render hasn't evil_CTX and the section of the annotation was never called.
Differential Revision: https://developer.blender.org/D11329
Make `CMAKE_OSX_DEPLOYMENT_TARGET` independent of buildbot settings and
always set to `OSX_MIN_DEPLOYMENT_TARGET`. That fixes the launch error
on OS older than buildbot's.
Remove unused `MACOSX_DEPLOYMENT_TARGET`.
Fix T88419
Diff D11323
Since there is a special callback for assigning the name from the
search box, and the callback doesn't call an operator or something
else that would do an undo push, I think the solution is to do it
manually here.
Another option would be adding the button flag "UI_BUT_UNDO", which
isn't set by default for search buttons, but that gives us ugly names for
the undo steps, so an explicit push is better in this case.
Differential Revision: https://developer.blender.org/D11190
When activated in modal, `translate`, `resize`, `rotate`, `shear` and
`edge_rotate_normal` use a different orientation than the set in scene.
This orientation needed to match since some of these modes can be switched
during operation.
The default orientation for these modes was `V3D_ORIENT_VIEW`.
And this changed when finishing the `translate` and `resize` to
`V3D_ORIENT_GLOBAL`.
But this could cause inconsistencies when inputting values from the
keyboard.
The solution now is to change the orientation when you change the mode.
---
Note: Although the user can expect the value entered to reflect the
orientation set in the scene, it would require a lot of changes and would
not be really useful.
Use bone rotation order to compute the baked rotation. This fixes a bug
introduced in rB0e85d701c654, where the object rotation order was
applied to the bone.
Maniphest Tasks: T88359, T86193
Reviewed By: sybren, GuiltyGhost, #animation_rigging
Differential Revision: https://developer.blender.org/D11282
This node is similar to the Value and Vector node.
It just provides a way to use the same material in multiple nodes
without exposing it outside of a node group.
Differential Revision: https://developer.blender.org/D11305
This adds a new Material Assign node. It can be used to change the
material used by an existing mesh or to assign a material to a mesh
that has been generated from scratch.
Differential Revision: https://developer.blender.org/D11155
This fixes the `Apply Modifier` and `Visual Geometry to Mesh` operator
when a modifier changed materials on the evaluated geometry.
This is necessary since rB1a81d268a19f2f1402f408ad1dadf92c7a399607.
Differential Revision: https://developer.blender.org/D11303
The old geometry nodes evaluator was quite basic and missed many features.
It was useful to get the geometry nodes project started. However, nowadays
we run into its limitations from time to time.
The new evaluator is more complex, but comes with new capabilities.
The two most important capabilities are that it can now execute nodes in
parallel and it supports lazy evaluation.
The performance improvement by multi-threading depends a lot on the specific
node tree. In our demo files, the speedup is measurable but not huge. This
is mainly because they are bottlenecked by one or two nodes that have to be
executed one after the other (often the Boolean or Attribute Proximity nodes)
or because the bottleneck is multi-threaded already (often openvdb nodes).
Lazy evaluation of inputs is only supported by the Switch node for now.
Previously, geometry nodes would always compute both inputs and then just
discard the one that is not used. Now, only the input that is required
is computed.
For some more details read D11191, T87620 and the in-code documentation.
Differential Revision: https://developer.blender.org/D11191
This feature of `select_side_of_frame` was disabled by removing option
from operator property enum but functional code was never removed.
Add back option to use this feature.
Feature was disabled due to keymap issue. Currently this feature doesn't
have keymap assigned.
Due to misunderstanding of how strip duplication works, animation data
was duplicated on all strips when any strip was split.
`SEQ_sequence_base_dupli_recursive()` duplicated data on strip that was
being split, and `SEQ_ensure_unique_name()` duplicated animation on all
strips.
Only duplication should be done with `SEQ_ensure_unique_name()` and only
on right side split strips, because only these strips are duplicated.
Fix issue described in T87678, which was partially a bug and partially
change in intended(at least as far as I can tell) behaior.
Function `evaluate_seq_frame_gen` that was partially responsible for
filtering strips in stack for rendering wasn't working correctly.
Intended functionality seems to be removing all effect inputs from stack
as it is unlikely that user would want these to be blended in. However
there was logic to exclude effects placed into same input, which because
of weak implementation caused, that any effect input, that is effect as
well will be considered to be part of stack to be blended in.
This bug was apparently used to produce effects like glow over original
image.
Even though this is originally unintended, I have kept this logic, but
I have made it explicit.
Another change is request made in T87678 to make it possible to keep
effect inputs as part of stack when they are placed above the effect,
which would imply that blending is intended. This change is again
explicitly defined.
Whole implementation has been refactored, so logic is consolidated
and code should be as explicit as possible and more readable.
`must_render_strip function` may be still quite hard to read, not sure
if I can make it nicer.
Last change is for remove gaps feature code - it used same rendering
code, which may be reason why its logic was split in first place.
Now it uses sequencer iterator, which will definitely be faster than
original code, but I could have used `LISTBASE_FOREACH` in this case.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11301
Recently `SEQ_sort()` function was split so functionality is provided
on per-seqbase basis. After discussion about this split, it turned out,
that per-seqbase operation is only that should be provided, because
RNA API functions need to be able to access arbitrary seqbase
Remove recently introduced function `seq_sort_seqbase` and change
`SEQ_sort` function to operate on seqbase.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11297
After this commit, all geometry node "init" and "update" functions are
at the top of each file, right below the "layout" function. This means
you can always scroll to the bottom of the file to see the entry point,
and the boring boilerplate code is grouped in one section.
With this patch you will be able to add and remove attributes from curve
data inside of geometry nodes. The following is currently implemented:
* Adding attributes with any data type to splines or spline points.
* Support for working with multiple splines at the same time.
* Interaction with the three builtin point attributes.
* Resampling attributes in the resample node.
The following is not implemented in this patch:
* Joining attributes when joining splines with the join geometry node.
* Domain interpolation between spline and point domains.
* More efficient ways to call attribute operations once per spline.
Differential Revision: https://developer.blender.org/D11251
These seem to be causing some stability issues, and really are just not that
useful in practice. Compiling them is slow already, so it does not improve
the user experience much to show an AO preview if it's not nearly instant.
Textures may be important to be able to identify an object. They are also a way
to make many objects look more like when rendered with an advanced render
engine, without being that expensive.
So this seems like a simple way to increase usefulness of the automatic
previews.
While this function should (currently) not be called on linked ID, there
is no reason to treat those differently than local IDs, for the part
that they have in common: needs to be properly sorted.
`id_sort_by_name` would simply not deal properly with linked IDs, could
lead to mixing IDs from different libraries, and unsorted IDs within the
same library.
Note that while sorting of local IDs is fine, currently sorting of
linked IDs is a total unpredictable failure.
Next commit will fix this, ensuring that linked IDs are grouped by their
library, and properly name-sorted within this library group.
This patch turns off the creation of file thumbnails for files that are
offline and therefore not fully-present on the file system. These types
of files - typically cloud-based or stored on slower backup media -
only have their contents available when actually accessed, at which
point there will be a short delay. If we allow thumbnail creation in
this state then all offline files in a folder will be downloaded just
to view a listing, which can take a long time.
Files in this state will instead get a more generic thumbnail that
still indicates file type (icon in center) and that shows offline state
will a special icon at the bottom-left.
Although this currently only affects Windows users, most of this patch
is platform-agnostic. So other platforms inherit this behavior if they
only add FILE_ATTR_OFFLINE attribute to files in this state.
See D11101 for more information.
Differential Revision: https://developer.blender.org/D11101
Reviewed by Julian Eisel
Fix T88167.
Caused by {rB97defd9cd79b6e3ed0e52481a7078107dbe0522b}
`(BLI_rcti_isect_pt` used here to confirm if cursor position is in between active region boundary.
Subtracting min region boundary from the mouse position before the check, fails the condition.
`mval[2]` introduced to hold the region relative mouse position.
Reviewed By: Severin
Maniphest Tasks: T88167
Differential Revision: https://developer.blender.org/D11224
Changes output for geometry info node in Eevee to be consistent with Cycles (w/o osl)
Before this patch Eevee outputs Z-tangent even for hair. This patch changes it to output hair tangent (growth direction). Hair tangent is impossible to derive otherwise from normal or view direction.
Reviewed By: fclem, brecht
Differential Revision: https://developer.blender.org/D10841
This commit allows that the evaluated geometry of an object has
different materials from the original geometry. This is needed
for geometry nodes.
The main thing that changes for render engines and exporters
is that the number of material slots on an object and its geometry
might not match anymore. For original data, the slot counts are
still equal, but not for evaluated data.
Accessing material slots though rna stays the same. The behavior
adapts automatically depending on whether the object is evaluated.
When accessing materials of an object through `BKE_object_material_*`
one has to use a new api for evaluated objects:
`BKE_object_material_get_eval` and `BKE_object_material_count_eval`.
In the future, the different behavior might be hidden behind a more
general C api, but that would require quite a few more changes.
The ground truth for the number of materials is the number of materials
on the geometry now. This is important in the current design, because
Eevee needs to know the number of materials just based on the mesh in
`mesh_render_mat_len_get` and similar places.
In a few places I had to add a special case for mesh edit mode to get it
to work properly. This is unfortunate, but I don't see a way around that
for now.
Differential Revision: https://developer.blender.org/D11236
This is caused by lacking of consideration of
OB_DUPLICOLLECTION in updateDepsgraph() in the
modifier.
Reviewed by: Antonio Vazquez (antoniov)
https://developer.blender.org/D11292
Splitting of effect strip alone wasn't handled properly. Previously
this resulted in duplicating effect strip, and it was broken at least
from 2.79.
Change in rB8ec6b34b8eb2 was intended to allow splitting strips
individually, so it can be used as RNA API function but also so it
requires as little glue logic as possible.
This is fixed by splitting all dependent strips at once in 2 separate
ListBases for left and right strips. Strips can be finally moved into
original `ListBase`.
With this fix it is still possible to split strips individually with
little glue logic. RNA API function could return list of split strips
as well, currently at least one strip in chain will be provided so
chain can be reconstructed on python side.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D10209
Python API function Sequence.move_to_meta() did delete effect chain
when strip with effects was moved.
Use iterator API to query effect strips and move whole chain to meta.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11206
Includes update for OpenVDB file IO, i.e. fixes an issue with
compression flag combination that resulted in random segfaults.
Other changes: Cleanup and formatting.
Read and write generated coordinates (also known as "original
coordinates", "reference coordinates", or "orcos") from and to Alembic.
A custom geometry property named "Pref" is used for (hopefully)
interoperability with Maya and Houdini. For now it's only guaranteed for
Blender-to-Blender.
Export: writing generated coordinates is optional (on by default).
Import: generated coordinates are always read whenever the reading of
vertex data is enabled.
Manifest Task: T88081
Copying an ID outside of BMain should not assume that the datablocks it
uses are now directly linked. This would be wrong e.g. in case that new
no-main ID is copied from a linked data-block and is supposed to remain
a linked data.
Value of the size argument in `strncat` is too large, might lead to a
buffer overflow.
Change the argument to be the free space in the destination buffer
minus the terminating null byte.
Introduced in 5368859a66
Reviewed By: Sergey
Differential Revision: https://developer.blender.org/D11289
The BlenderSync will do quite a bit of work on every sync_data() call
even if there is nothing changed in the scene. There will be early
outputs done deeper in the call graph, but this is not really enough to
ensure best performance during viewport navigation.
This change makes it so sync_data() is only used when dependency graph
has any update tags: if something changed in the scene the dependency
graph will know it. If nothing changed there will be no IDs tagged for an
update in the dependency graph.
There are two weak parts in the current change:
- With the persistent data there is a special call to ignore the check
of the dependency graph tags. This is more of a safety, because it is
not immediately clear what the correct state of recalc flags is.
- Deletion of objects is detected indirectly, via tags of scene and
collections.
It might not be bad for the first version of the change.
The test file used: {F10117322}
Simply open the file, start viewport render, and navigate the viewport.
On my computer this avoids 0.2sec spend on data_sync() on every
up[date of viewport navigation.
We can do way more granular updates in the future: for example, avoid
heavy objects sync when it is only camera object which changed. This
will need an extended support from the dependency graph API. Doing
nothing if nothing is changed is something we would want to do anyway.
Differential Revision: https://developer.blender.org/D11279
- Used try/except instead of a poll function.
- The error case referenced a non-existent error handling module.
Prefer poll functions over exception handling where possible,
also having an operators logic in a try block isn't good practice
as it can hide more serious errors in the code.
Note that duplicate pencil settings access should be moved into a
utility function. This can be part of a separate cleanup.
Resolve ownership ambiguity with shared physics pointers.
Previously, LIB_ID_CREATE_NO_MAIN allowed pointer sharing with
the source ID so physics caches can be shared between original and
evaluated data: (Object.soft.shared & Object.rigidbody_object.shared).
This only worked properly for LIB_TAG_COPIED_ON_WRITE ID's,
as LIB_TAG_NO_MAIN can be used in situations where the original ID's
lifetime limited by it's original data.
This commit adds `LIB_ID_COPY_SET_COPIED_ON_WRITE` so ID's only share
memory with original data for ID's evaluated in the depsgraph.
For all other uses, a full copy of physics data is made.
Ref D11228#287094
lookups
We use the schema so that we can access top level attributes as well.
This is already done for polygon meshes and curves, so this only
modifies the behavior for subdivision objects.
Expand is not expected to update the visibility state of the PBVH, only
the Face Sets IDs. If visibility updates are made accidentally, PBVH
rendering breaks.
In order for this to work properly, the following fixes are needed:
- Expand should always check for active component before attempting to
modify a Face Set ID
- Expand should always check the visibility state on original_face_sets, as
it is the array that contains the visiblilty state that corresponds with the
current state used for PBVH rendering. This implies that after any modification
done by Expand, the visibility state of ss->face_sets and
expand_cache->original_face_sets should match (like in any other tool that
does not modify visibility).
- Expand should never modify the Face Set ID of a poly that is hidden in
expand_cache->original_face_sets.
- When deleting an ID, hidden Face Sets should be skipped when picking IDs for
content filling. This avoids introducing hidden IDs back into the visible
geometry even after updating its visibility state.
Reviewed By: JulienKaspar, JacquesLucke
Maniphest Tasks: T88230
Differential Revision: https://developer.blender.org/D11243
Treat a missing <diffuse> the same as a black diffuse color.
The easiest way to see this bug is with a Collada shader like
```
<constant>
<emission>
<color sid="emission">1 0 0 1</color>
</emission>
</constant>
```
The Collada spec says this should be just
```
color = <emission>
```
ie. red everywhere. The importer slots the red into the Principled Emission socket, but since it leaves the Base Color as the default off-white, this is added to red, and the material looks white-pink in the light and red only in the shadows.
Putting black in the Base Color makes it look red everywhere.
D10939 will also eliminate the much-less-noticeable specular term for this case.
Reviewed By: gaiaclary
Differential Revision: https://developer.blender.org/D10941
Collada shaders with black <specular> should import with Specular=0.
(A missing <specular> is the same as black.)
The general specular conversion is hard, but this case is common and easy.
Fixes the specular for all <constant>/<lambert> shaders, and <blinn>/<phong>
shaders with black/omitted <specular>. Before this they all looked too "shiny".
Reviewed By: gaiaclary
Differential Revision: https://developer.blender.org/D10939
This is giving too bright pixel values, as the sample scaling and random number
sample are wrong. The proper fix for this is complicated. It will be solved in
Cycles X, for now we disable this combination.
Contributed by Intel. On some scenes like classroom with particular integrated
GPUs this speeds up rendering 1.97x. With other benchmarks and GPUs it's
between 0.99-1.14x.
Combining location, rotation and scale channels into a matrix is
a standard task, so while it is easily accomplished by constructing
and multiplying 3 matrices, having a standard utility allows for
more clear code.
The new constructor builds a 4x4 matrix from separate location,
rotation and scale values. Rotation can be represented as a 3x3
Matrix, Quaternion or Euler value, while the other two inputs
are vectors. Unneeded inputs can be replaced with None.
Differential Revision: https://developer.blender.org/D11264
No real functional changes.
When `i` is zero, `filelist_cache_previews_push` was called twice with
the same icon.
This caused the preview to be computed twice when only once is needed.
Before, the camera parameters were calculated only for first frame.
If the camera is animated, these values need to be recalculated in order to get the new camera view position and export the strokes as expected.
Also fixed the export of PDF when the view is not in camera view. PDF export, needs to be done in camera view.
This patch introduces non linear sliders. That means, that the movement
of the mouse doesn't map linearly to the value of the slider.
The following changes have been made.
- Free logarithmic sliders with maximum range of (`0 <= x < inf`)
- Logarithmic sliders with correct value indication bar.
- Free cubic sliders with maximum range of (`-inf < x < inf`)
- Cubic sliders with correct value indication bar.
Cubic mapping has been added as well, because it's used for brush sizes
in other applications (Krita for e.g.).
To make a slider have a different scale type use following line in RNA:
`RNA_def_property_ui_scale_type(prop, PROP_SCALE_LOGARITHMIC);`
or:
`RNA_def_property_ui_scale_type(prop, PROP_SCALE_CUBIC);`
Test the precision, step size and soft-min if you change the scale type
of a property as it will feel very different and may need tweaking.
Ref D9074
Since spline data is stored separately for each spline, the data often
needs to be flattened into a separate array. It's helpful to have the
necessarily-sequential part of that split off into a separate method.
I've found myself using functions like these in quite a few places.
Adds internal API for creating and managing OpenXR actions at the
GHOST and WM layers. Does not bring about any changes for users since
XR action functionality is not yet exposed in the Python API (will be
added in a subsequent patch).
OpenXR actions are a means to communicate with XR input devices and
can be used to retrieve button/pose states or apply haptic feedback.
Actions are bound to device inputs via a semantic path binding
(https://www.khronos.org/registry/OpenXR/specs/1.0/html/xrspec.html#semantic-path-interaction-profiles),
which serves as an XR version of keymaps.
Main features:
- Abstraction of OpenXR action management functions to GHOST-XR,
WM-XR APIs.
- New "xr_session_start_pre" callback for creating actions at
appropriate point in the XR session.
- Creation of name-identifiable action sets/actions.
- Binding of actions to controller inputs.
- Acquisition of controller button states.
- Acquisition of controller poses.
- Application of controller haptic feedback.
- Carefully designed error handling and useful error reporting
(e.g. action set/action name included in error message).
Reviewed By: Julian Eisel
Differential Revision: http://developer.blender.org/D10942
This was a bug uncovered by rB50782df42586.
Previously, the lightcache was always discarded between redraw and forced
to be updated again.
Now we check for update inside the render loop making it compatible with
accumulation motion blur and long exposure.
This was caused by the slight focus gather not being wide enough
for small radii. Now the cast to int will properly round the radius to
the nearest integer.
This is related to T86244 Black Artefacts in EEVEE on Transparent BSDF
This was caused by the bokeh LUT being sampled outside the valid range.
But `texelFetch` is only valid if the sample actually exists. This lead to
undefined behavior.
The fix is to increase `DOF_MAX_SLIGHT_FOCUS_RADIUS` (which just offsets the
LUT along the X=Y axis) to avoid any sample outside the defined range.
operator
This is only used when the preset is set to 'Custom', make that clear in
the tooltip and disable editing the value if another preset is used.
Issue came up in T88155.
Maniphest Tasks: T88155
Differential Revision: https://developer.blender.org/D11210
effect
The way it is now, `ClothVertex->mass` is setup once for the clothObject
and then reused if it already exists [so does not change across frames/
steps] which happens in`cloth_from_object` / `do_init_cloth`, where
`SIM_cloth_solver_init` / `SIM_mass_spring_set_vertex_mass` are called
and set up masses in `Implicit_Data`.
Seems possible to update `ClothVertex->mass` every step in
`do_step_cloth`, however it seems more involved to update the masses in
`Implicit_Data` there as well. The masses from `Implicit_Data` are
accessed in many places, so it would be mandatory to have these masses
kept up-to-date (and even then it is unclear if the solver was designed
to work with these animated or if there are assumptions about this being
stable across the sim).
So propose to remove the (broken/not implemented) animation capabilities
on the property instead.
Maniphest Tasks: T88188
Differential Revision: https://developer.blender.org/D11225
This code in the geometry set header was not directly related to
geometry sets, it makes more sense in the attribute access header.
This makes it clearer that code for geometry components uses attribute
code, rather than the other way around. It also allows adding more
functionality to `BKE_attribute_access.hh` that depends on these things
without including `BKE_geometry_set.hh` there.
The code incorrectly used the size of the second to last segment rather
than the last segment's size. That was a problem when the last segment
is a vector segment but the second to last isn't.
I also used the opportunity to slightly refactor the control point
offsets cache, making it one longer so it also contains information
about the size of the last segment, simplifying other code.
In {rB266cd7bb82ce}, support for muting links was added. It might be
debatable if we define a shader as "having" displacement even if the link
is muted, but after said commit, shader_has_displacement() would return
true but still leave the returned node as NULL.
Now also return false if the link is muted (otherwise the caller would
need to additionally check the returned node as well.)
Maniphest Tasks: T88234
Differential Revision: https://developer.blender.org/D11256
Now the operators work like other areas of Blender using the list of selected objects.
Also, the name has been changed to:
```Layers:
- Copy Layer to Selected
- Copy All Layers to Selected
Materials:
- Copy Material to Selected
- Copy All Materials to Selected```
Reviewed By: mendio, pablovazquez, pepeland
Differential Revision: https://developer.blender.org/D11216
This module exposes the platform utils defined in the GPU module in C.
This will be useful for porting existing code with `bgl` to `gpu`.
Reviewed By: fclem, brecht, campbellbarton
Maniphest Tasks: T80730
Part of D11147
This module exposes the capabilities defined in the GPU module in C.
This will be useful for porting existing code in `bgl` to `gpu`.
Reviewed By: fclem, brecht, campbellbarton
Maniphest Tasks: T80730
Part of D11147
Code would still create an object (without setting up materials), code
for removing unused material slots would then freeze.
Now return/cancel early in case of empty selection.
This came up in T88269 [which is still not fully fixed, transforming
curve edit points clear their GP_STROKE_SELECT flag which now results in
the early exit, should be looked at separately]
Maniphest Tasks: T88269
Differential Revision: https://developer.blender.org/D11252
The term direction was used in 3 different ways in screen editing code,
making it hard to follow:
- 0-3 for as magic numbers mapped to [west,north,east,south].
- `h`, `v` characters for [horizontal,vertical] axes.
- Cycle direction SPACE_CONTEXT_CYCLE_PREV, SPACE_CONTEXT_CYCLE_NEXT
The following changes have been made:
- Add `eScreenDir` for [west,north,east,south], use variable name `dir`.
- Add `eScreenAxis` for [horizontal,vertical] values, use variable name
`dir_axis`.
- Add `eScreenCycle` for existing enum `SPACE_CONTEXT_CYCLE_{PREV/NEXT}`.
- Add macros `SCREEN_DIR_IS_VERTICAL(dir)`,
`SCREEN_DIR_IS_HORIZONTAL(dir)`.
Replacing `ELEM(dir, 1, 3)`, `ELEM(dir, 0, 2)`.
- Move `ED_screen_draw_join_highlight`, `ED_screen_draw_split_preview`
to `screen_intern.h`.
Reviewed By: Severin
Ref D11245
This allows extra options (in-front and stroke order) to be shown when adding line art kind of grease pencil object.
Reviewed by: Antonio Vazquez (antoniov)
Diff: https://developer.blender.org/D11130
- Matches changes in Python 3.x dictionary methods.
- Iterating now raises a run-time error if the property-group changes
size during iteration.
- IDPropertyGroup.iteritems() has been removed.
- IDPropertyGroup View & Iterator types have been added.
- Some set functionality from dict_keys/values/items aren't yet
supported (isdisjoint method and boolean set style operations).
Proposed as part of T85675.
There is a new Texture data-block socket that we can use in Geometry
Nodes now. This commit replaces the texture property of a node and
gives it a texture input socket instead. That increases flexibility.
The texture socket still has some limitations that will be lifted in the
next couple of days (e.g. it's not supported by the switch node and
cannot be exposed the a modifier yet).
Differential Revision: https://developer.blender.org/D11222
This fixes a few "obvious" places where we do unnecessary updates.
Those special cases were added in the early days of geometry nodes
when many updates were missing and we tried to get it to work at all.
There is a fairly high risk that with this change some required updates
will be missing again. Those can be fixed when we find thme.
Some of the update issues might have been fixed by rB58818cba40794905f9323080e60884e090f2d388
and similar changes we added over time.
Differential Revision: https://developer.blender.org/D11238
Improve the "In Betweens" tools:
- Push Pose from Rest Pose
- Relax Pose to Rest Pose
- Push Pose from Breakdown
- Relax Pose to Breakdown
- Pose Breakdowner
These all now use the same new sliding tool:
- Actual visual indication of the blending/pushing percentage applied.
- Mouse wrapping to allow for extrapolation without having to worry
about the initial placement of the mouse. This also means these tools
are actually usable when chosen from the menu.
- Precision mode by holding {key Shift}.
- Snapping to 10% increments by holding {key Ctrl}.
- Overshoot protection; by default the tool doesn't allow overshoot
(lower than 0% or higher than 100%), and it can be enabled by pressing
{key E}.
- Bones are hidden while sliding, so the pose itself can be seen more
clearly. This can be toggled by pressing {key H} while using the tool.
Reviewed By: #animation_rigging, zeddb, sybren, #user_interface, brecht, Severin, looch
Maniphest Tasks: T81836
Differential Revision: https://developer.blender.org/D9054
Sculpting tools are designed to ignore hidden geometry and behave like
hidden geometry does not exist.
When getting the neighbors of a vertex, now this takes into account
hidden geometry to avoid returing neighbors which connected edge is not
visible. This should make corner cases of a lot of tools work properly,
especially when working in low poly meshes when is common to have a
single face loop hidden.
Reviewed By: JacquesLucke
Differential Revision: https://developer.blender.org/D11007
This issue originates from a missing BVH packing for visibility data
when it is modified.
To fix this, this adds update flags to the managers to carry the
modified visibility information from the Objects' modified flag to the
GeometryManager.
Another set of flags is added to determine which data need to be packed:
geometry, vertices, or visibility. Those flags are then used when
packing the primivites.
Reviewed By: brecht
Maniphest Tasks: T87929
Differential Revision: https://developer.blender.org/D11219
When using a tool that is not a brush, the previously used brush
preset will still be active in the tool settings, so the cursor will
draw its custom reviews.
This checks if the current active tool is a brush before drawing its
previews. If it is not a brush tools, it draws a default white cursor.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D9418
This is very similar to rB5613c61275fe6 and rB0061150e4c90d, basically
just exposing a `VMutableArray` method to its generic counterpart. This
is quite important for curve point attributes to avoid a lookup for
every point when there are multiple splines.
An emission texture is currently connected to the Base Color socket. It should connect to the Emission socket, like a constant does.
Reviewed By: gaiaclary
Differential Revision: https://developer.blender.org/D10990
link->multi_input_socket_index, which is used to calculate the links
position on the multi-input socket, was not set.
Now it is set to the sockets current link count.
Review: Jacques Lucke (JacquesLucke)
Differential Revision: https://developer.blender.org/D11082
Heterogeneous lookup is useful when constructing a key in a
map/set is relatively expensive (e.g. `std::string`). When doing
lookups in the map/set, one usually does not want to construct
the type to avoid overhead. Instead, heterogeneous lookup
allows for using a different type (such as `StringRef`) as key.
This change makes it easier to implement heterogeneous
lookup for custom types. Before, one had to specialize
`blender::DefaultHash`. Now, one just has to implement
a `static uint64_t hash_as(value)` on the type itself.
One still has to provide the equality operator in addition
to the hash function of course.
Before, any usage of tbb wrappers used in modifier code would
just fall back to the non-threaded non-tbb version.
We ran into this issue a couple of times in patches.
Sometimes it is useful to find the key that compares equal
to a known key. Typically that happens when the key itself
has additional data attached that is not part of its hash.
Note that the returned key reference/pointer is const, because
the caller must not change the key in a way that changes its
hash or how it compares to other keys.
Unavailable sockets should generally be ignored during evaluation.
They mainly exist because we don't have a better mechanism to turn
some sockets on/off depending on node parameters.
Currently, it is still possible that a link connects an available with an
unavailable socket. This link is not displayed in the ui and should
generally be ignored.
Previously, multiple threads adding information to node ui storage
at the same time resulted in memory corruption. The lock prevents
that, but might potentially become a bottleneck in the future.
For now favour correctness over a potential performance bottleneck.
'Preferences' is the term used elsewhere in Blender so this commit makes
the option more consistent.
In the future, the "default" target could be changed to something more
descriptive.
The new "Close Area" operator can let any neighbor replace the area to
be closed. This patch improves the selection of the best, and most-
aligned, neighbor. It maximizes the ratio of the shared edge lengths,
rather than minimize the absolute amount of misalignment. This follows
our expectations closer because it takes into account the relative
sizes of the areas and their edges.
see D11143 for details and examples.
Differential Revision: https://developer.blender.org/D11143
Reviewed by Campbell Barton
When disk cache path is same as blend file path, with Unix-like systems
blend file can be overwritten by disk cache directory.
This was caused by `BLI_delete(path, false, true)` when path points to
file. On Windows this would result in error message and file would not
be deleted. On Linux, file is deleted and then overwritten with cache
directory.
To further minimize chance of removing blend file, append disk cache
path with `_seq_cache` suffix.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11217
Byte images used `ibuf->float_colorspace` as source colorspace.
This was oversight - `ibuf->rect_colorspace` should be used as source
colorspace.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11223
In e1f3996d74, logic for changing metastrip start and end frame based
on contained strips was removed. This was done intentionally and
incorrect functionality wasn't noticed as drawing code reflected
seemingly correct state.
Original code was mostly correct, because meta strip doesn't store its
internal start and end points. This code was restored with minor
modifications so function `SEQ_time_update_sequence()` is fully self
contained as it is used not only by transform operator.
In addition, drawing glitches that happen when meta content is outside
of meta boundaries were fixed. These glitches were not caused by
e1f3996d74.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11215
Not allowing external direct access to the vector of splines in the
curve will help for things like reallocating custom data when a spline
is added or removed.
* Set colors for the new texture and material sockets
* Material uses the same color used for shading icons
* Texture uses a plum color desaturated enough to not be confused with Vector's violet
* Image socket adjusted to be closer to Texture sockets but darker
* Integer socket toned down in saturation to not stand out so much
(and be closer to float sockets which are gray)
Making this change now during bcon1 to gather feedback from the community,
and because Geometry Nodes needs to use the new texture/material sockets.
This commit exposes the first spline control point attributes. The
implementation incorporates the attributes into the virtual array
system, providing efficient methods to flatten the data into a
contiguous array and to apply changes from a flattened array. This
is only part of the eventual goal, which includes changes to run
attribute nodes separately for each spline to completely avoid copying.
So far `tilt` and `radius`, the two generic attributes common to
all spline types, are implemented. The more complex `position`
attribute is also added. It requires some special handling for Bezier
splines, where the control point handles need to be moved along with
the control points. To make that work I also added automatic handle
recalculation to the Bezier spline.
Differential Revision: https://developer.blender.org/D11187
This was caused by the drivers not optimizing the shader enough to remove
the samplers and data used by closure eval. Removing the lighting loops
from the depth shader fixes the perf regression.
With Constrain to Image Bounds selected, UVs will be constrained to the
correct/closest UDIM if the image is tiled.
UVs will be constrained to the 0-1 UV space if the image is not tiled.
This will override the present behavior of always constraining selected
UVs to the 0-1 UV space (UDIM 1001).
Reviewed By: campbellbarton
Ref D11202
The sockets are not exposed in any nodes yet.
They work similar to the Object/Collection sockets, which also
just reference a data block.
This is part of D11222.
Was causing calculation issues later on in the kernel.
This change catches the most obvious case: missing attribute. The old
code was trying to set tangent to 0, but because it was transformed as
a normal it got converted to non-finite value. This change makes it so
that no transform is involved and 0 is written directly to the SVM
stack.
To cover all cases it will require using safe_normalize() in this node
and in the normal transform function. This is more involved change from
performance point of view, would be nice to verify whether we really want
to go this route.
I've left asserts in the BSDF allocation functions. Don't have strong
connection to them, but think they are handy and are not different from
having an assert in the path radiance checks.
Differential Revision: https://developer.blender.org/D11235
It is possible that BSDF allocation will advance pointer in the
allocation "pool" but will return null pointer if the weight is
too small.
One artist-measurable issue this change fixes is random issues
with denoising: normal pass for denoising could have accessed
non-initialized normal of a closure.
Differential Revision: https://developer.blender.org/D11230
When activated in modal, `translate`, `resize`, `rotate`, `shear` and
`edge_rotate_normal` use a different orientation than the set in scene.
This orientation needed to match since some of these modes can be switched
during operation.
The default orientation for these modes was `V3D_ORIENT_VIEW`.
And this changed when finishing the `translate` and `resize` to
`V3D_ORIENT_GLOBAL`.
But this could cause inconsistencies when inputting values from the
keyboard.
The solution now is to change the orientation when you change the mode.
---
Note: Although the user can expect the value entered to reflect the
orientation set in the scene, it would require a lot of changes and would
not be really useful.
Extracts `nlasnapshot_blend_get_inverted_upper_snapshot()` from
`BKE_animsys_nla_remap_keyframe_values()`
This introduces a new struct member:
`NlaEvalChannelSnapshot->remap_domain` and marks which values of
`blended_snapshot` are processed for remapping/used-for-inverting.
Effectively, it marks which values have successfully been remapped and
can be further used for remapping.
`nlasnapshot_blend_get_inverted_upper_snapshot()`:
output snapshot `r_upper_snapshot` has each channel's `remap_domain`
written to which effectively marks the successfully remapped values.
The only reason a value is not in the remap domain is if inversion
failed or it wasn't marked to be remapped.
`..get_inverted_upper_snapshot()` has a variant `nlasnapshot_blend()`
from {D10220}, but this patch doesn't depend on it at all. A third
variant will later be added `..get_inverted_lower_snapshot()`.
Altogether, these three functions allow solving for any of
(lower_snapshot, upper_snapshot, blended_snapshot) given the other two.
The function `..get_inverted_lower_snapshot()` will also similarly
process the remap domain of the blended and lower snapshot.
added assertions within `nlasnapshot_blend()` and
`..get_inverted_upper_snapshot()` to future proof branches dealing with
blendmode and mixmodes. (suggested by sybren)
No user functional changes
Reviewed By: sybren
Differential Revision: https://developer.blender.org/D10222
This turns links red if no implicit conversion can be made between the
from socket and the to socket. For geometry nodes this happens with
object, geometry, collection, and string sockets that are connected to
a different type. The change is simply implementing a callback that is
already implemented for other node tree types.
Differential Revision: https://developer.blender.org/D11229
Add translation/rotation/scale parameters for custom bones shapes. The
new scale is a 3D vector `custom_shape_scale_xyz`, and replaces the
`custom_shape_scale` float.
Reviewed By: #animation_rigging, sybren, zeddb
Differential Revision: https://developer.blender.org/D10974
Previously we had a lot merge conflicts since we always put the most
recently added node at the bottom. By sorting the list we will have
one fewer merge conflict when a node is added in most cases.
The id-property iterator referenced a PyObject pointer without
increasing it's user count - allowing for errors if the value
goes out of scope during iteration.
Similar to how `GVArray_For_VArray` implements `materialize_impl` to
forward the work to its non-generic virtual array, we can do the same
thing for the mutable version, `GVMutableArray_For_VMutableArray`.
This commit should have no visible changes, since as far as I can tell
the only user of this class does not implement special materialize
methods anyway.
This enables the overlay for instanced geometry.
After this change, objects that are an instance of the current active
object (which are also being modified in the current active mode) won't
fade, which is different from the previous behavior.
Reviewed By: fclem
Maniphest Tasks: T82155
Differential Revision: https://developer.blender.org/D9362
When checking if the mesh has only one Face Set only the current active
component for expand needs to be checked. Otherwhise other components
that won't be modified by Expand that contain other IDs will be taken
into account, making the Face Set deletion go into an infinite loop.
Reviewed By: JacquesLucke
Maniphest Tasks: T88060
Differential Revision: https://developer.blender.org/D11169
Update the "current value" of the Shape Key blend amount when value is
not within the min/max range. New function `rna_ShapeKey_update_minmax`
used to update and clamp the current value.
Reviewed By: mano-wii, lichtwerk, #animation_rigging, sybren
Maniphest Tasks: T54339
Differential Revision: https://developer.blender.org/D11071
When a scene uses cryptomatte the viewport rendering would lead to a
memory leak. The reason was that all image renders (viewport+final)
activated cryptomatte. But is only used for final rendering.
This patch only activates cryptomatte when doing final rendering.
The Cycle-Aware Keying option was added in 2.8 and is used
to allow keyframing over cyclic F-Curves without disturbing
the cycle (e.g. overwriting an end keyframe updates both ends).
This effect is not limited to auto keyframing and is applied
to any key insertion, but when the popovers were rearranged
it was put in the Auto-Keyframing related one for some reason.
This is misleading, especially since because of that the
option is incorrectly greyed out when auto keyframing is
disabled, so move it to the generic Keyframing popover.
Differential Revision: https://developer.blender.org/D11213
Was giving a warning:
```
BKE_spline.hh:293:35: warning: 'interpolate_to_evaluated_points' overrides a
member function but is not marked 'override' [-Winconsistent-missing-override]
```
Mistake in {rBe48c4d73d378}.
Was using the vertex index as a lookup for the loop color (instead of
the loop index).
(Issue was not present in original D1429 btw).
Maniphest Tasks: T88145
Differential Revision: https://developer.blender.org/D11212
These variables and methods should make it easier to loop through buffers elements/pixels. They take into account single element buffers.
Single element buffers can be used for set operations to reduce memory usage.
Usage example: P2078
Reviewed By: #compositing, jbakker
Differential Revision: https://developer.blender.org/D11015
This allows extra options (in-front and stroke order) to be shown when adding line art kind of grease pencil object.
Reviewed by: Antonio Vazquez (antoniov)
Diff: https://developer.blender.org/D11130
The last point of the output was at the same location as the
first point of a cyclic spline. The fix is simple, just account for
the cyclic when choosing the sample edge length, and don't
hard code the last sample.
The spline `length` function assumed that the curve always had evaluated
edges. That is clearly false. This commit adds a check to `length` and a
special case for a single point in the curve resample node.
This commit uses two changes to improve the performance of the point
instance node.
**Prevent Reallocations**
At 64 bytes, the transform matrix for every instance is rather large,
so reallocating the vector as it grows can become a performance bottle-
neck. This commit reserves memory for the instances that will be added
to prevent unecessary reallocations as the instance vector grows.
In a test with 4 million instances of 3 objects in a collection, the
node was about 40% faster, from 370ms to 270ms for the node.
**Parallelization**
Currently the instances are added by appending to a vector. By changing
this slightly to fill indices instead, we can parallelize the operation
so that multiple threads can fill data at the same time. Tested on a
Ryzen 3700x, this reduced the runtime from the above 270ms to 44ms
average, bringing the total speedup to ~8x.
Note that displaying the instances in the viewport is still much slower
than the calculations in node, this change doesn't affect that.
This patch refactors the instance component to make use of the earlier
refactoring in rB4599cea15dcf. Now we don't have to build an array of
instance references the size of the point domain, and we can gather the
possible instances only once and use the same vector for all component
types. Generally the node should be a bit faster and use less memory.
The logic is moved around a bit, especially the hashing of the ID
attribute to pick from the instance list, but the result is unchanged.
Differential Revision: https://developer.blender.org/D11203
Those were mostly just left over from previous work on particle nodes.
They solved the problem of keeping a reference to an object over
multiple frames and in a cache. Currently, we do not have this problem
in geometry nodes, so we can also remove this layer of complexity
for now.
We need to always add a single point to the last cyclic segment that
completes the loop, because that includes the starting point of the
evaluated edge. The existing code forgot about that point.
This node generates a naturally parametarized (even length edge) poly
spline version of every spline in the input. There are two modes,
"Count", and "Length". These are similar to the same options for the
line primitive node in end points mode.
I implemented this instead of a "Sample Points" node, because for this
operation it's trivial to keep the result as a curve, which is nice
since it increases flexibility, and because it can make instancing
simpler, i.e. using the transforms of each evaluated point rather than
requiring the construction of a "rotation" attribute.
Differential Revision: https://developer.blender.org/D11173
Now it's possible to append materials of one grease pencil object into another one. The operator allows active material or all materials.
Also, the Layer Copy To Object has been renamed to Layer Append to Object to keep consistency and now allows to append all layers at once.
Small addition inspired by [this tweet](https://twitter.com/Vorundor/status/1390645286624763909) of a user in a situation I also saw myself in the past.
Showing "(Clipped)" next to the view name in the `Text Info` overlay fits well since it's a per-viewport setting.
{F10059921, size=full}
While on Local view:
{F10059925, size=full}
Multiple viewports:
{F10059946, size=full}
Reviewed By: Severin
Differential Revision: https://developer.blender.org/D11193
Avoids having frames with the word "Frame" on top, resulting in less visual noise.
(users were working this around by adding a space as label name).
Differential Revision: D11193
There need to be more cleanup for ffmpeg 4.5 (ffmpeg master branch).
However this now compiles on ffmpeg 4.4 without and deprication
warnings.
Reviewed By: Sergey, Richard Antalik
Differential Revision: http://developer.blender.org/D10338
We were not assigning the amount of sound channels to the output frames.
Newer ffmpeg releases has sanity checks in place and doesn't fall back
to two channels anymore.
This adds `parallel_for` to the Attribute Curve Map node to improve performance.
Grain size set to 512.
Reviewed By: HooglyBoogly
Differential Revision: https://developer.blender.org/D11194
this is a followup to rB2bd85d9cc623, we cannot forcefully delete
obsolete overrides of object data (meshes etc.), as this implies also
deleting their user object, which might still be a perfectly valid
override, albeit in conflict regarding its obdata ID pointer...
This is the opposite of previous code, which would keep those
'deprecated' overrides arround (often in a dedicated collection), when
they were detected as user-edited.
While this is a safe-ish way to (try to) preserve user-edited data, this
tends to add too much 'trash' data to production scenes, which cleaning
becomes a burden.
Note that user will get warnings in thos cases, and can always choose
not to save the current blend file and go fix the library issue instead.
Code detecting overrides which reference linked data is missing was
actually missing many cases, leading to too much garbage data being kept
around after resync process.
This iterator design provides means to create simple and flexible API
to query and work with collection of strips. It should be used in cases
when conditions require multiple stages of recursive iteration of all
strips or similar complex scenarios.
Quick API overview:
Basic queries are standalone functions that return SeqCollection
Use SEQ_collection_create() and SEQ_collection_free() to construct
such query functions.
Use these functions to get strips of known conditions, like selected
strips, movie strips, muted strips and so on.
Use SEQ_reference_query() when querying strips with relation to
some reference strip. For example to get its effects, strips that have
same type or use same input file and so on.
These aren't standalone functions because often you need to query strips
relative to each strip in collection.
Use SEQ_collection_expand() to query strips relative to each strip
in collection. These will be merged to original collection.
Use SEQ_collection_merge() to merge 2 collections
To iterate collection elements use macro SEQ_ITERATOR_FOREACH()
This API is quite specific, but I think it is best suited for tasks
that are usualy solved in sequencer codebase.
Old sequencer iterator has been completely removed.
SEQ_ALL_BEGIN and SEQ_ALL_END macros re-use new iterator design.
As initial use for this iterator select_grouped_effect_link()
function has been rewritten. It was not only broken, but also it used
DNA fields to aid iterating strips.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D10337
Instead of only drawing images on first start, load them into cache.
This resolves a logical problem when images don't load fast enough,
where the animation would load some frames each time until all images
loaded into cache.
In practice this could play back with severe frame skipping many times
times before all images were loaded making playback smooth.
Part of a fix for T81751.
Originally colorspace of float images was converted using CPU.
GLSL will render images much faster.
Originally image was converted to `global_role_default_byte` space,
disregarding view transform and also display device, which now is
possible to specify. These parameters could be set via commandline to
settings used in Blender, however if they are to be set by users, these
needs to be sanitized.
Right now defaults are assumed for device given for
`COLOR_ROLE_DEFAULT_BYTE`. This should produce same behavior as
implemented before.
Together with D11167 animation player performance should be much better.
This code was mostly copy-pasted from sequencer.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11178
This node has the same functionality as the color and vector curve
mapping nodes in the shader editor. Here is works on every value for
the selected attribute, and it can also output a float value. Other
than that, the implementation is quite straightforward-- almost
completely boilerplate code.
Differential Revision: https://developer.blender.org/D10921
Mode switching passed through when the mode wasn't supported for all
modes except particle edit-mode.
Add a check to ED_object_mode_compat_test to prevent the error message.
The check to include particle edit mode in the object-mode drop-down
didn't match the poll function to edit particle edit mode.
Share the check between both functions.
This option should always be available. Even if it does not affect the
current object depending on its mode, it affects the rest of the geometry
of the scene.
The rest of the options also work this way. For example, a grease pencil
object does not have face orientation, but the ovelay option is still
there for the rest of the scene.
Reviewed By: JulienKaspar, HooglyBoogly
Maniphest Tasks: T83398
Differential Revision: https://developer.blender.org/D10215
Fix trying to use cross product on parallel vectors.
Fix intersection checks failing because we run into floating point
issues with very small numbers.
One of current annoying limitations of Blender re Collections/Objects is
that objects are forbidden to not be instantiated in at least one
collection.
Code ensuring that as a pst-processing step of override creation/resync
operations would be a bit too eager to add those objects to an external
'ad-hoc' collection, which poses several issues (both in term of keeping
the scene well organized, and related to override hierarchy handling).
So now be very conservative and only generate and use external 'storage'
collection for those objects when it is absolutely mandatory.
In pratice, it means this should never happen anymore on any decently
organized data source.
Caused by {rB0d9f79b163ee}.
IDP_SyncGroupTypes was now syncing from src to src (leading to
unexpected operator properties).
Assume this is rather critical, dont know the part of the code well, but
above commit clearly shows a change from 'dest->data.group' to 'src-
>data.group' which shouldnt be there.
Maniphest Tasks: T88030
Differential Revision: https://developer.blender.org/D11171
When the user hovered over a number input field, pressed Enter and then
typed in '0', confirming the input would always cancel the action. This
is because in this particular case `ui_textedit_begin` is called
instead of `ui_numedit_begin`. This function will not set
`data->startvalue` (leaving it at `0`) which will then trigger the
cancel in `ui_apply_but_NUM` which checks if the input changed (by
comparing the entered value with `data->startvalue`).
The fix makes sure that when `ui_textedit_begin` is called on a number
button, the `data->startvalue` is set correctly like in
`ui_numedit_begin`.
Breaking commit: rBSeb06ccc32462beaacbb114d6d0e450b6fc911047
Note: This also affects pressing tab to move to a new number field and
entering '0'. The fix will also cover this case.
Reviewed By: Severin, #user_interface
Maniphest Tasks: T88058
Differential Revision: https://developer.blender.org/D11168
Export UV maps to Alembic on every frame. This makes the export of UV
maps consistent with mesh normals. In the case of static UV maps it may
cause a slight slowdown (since they're exported on every frame now), but
due to the deduplication performed by the Alembic library, the resulting
files will be the same size anyway.
Thanks to @CodyWinch for providing the solution to the problem, and
writing the original patch D8397.
Differential Revision: https://developer.blender.org/D8397
The tooltips for the tracker buttons were all equal (because they all
use the same operator).
This adds the `get_description` operator callback to the
`CLIP_OT_track_markers` operator and returns a more descriptive
tooltip for the different buttons.
Reviewed By: mont29
Maniphest Tasks: T87983
Differential Revision: https://developer.blender.org/D11145
While rB6b18678e34bf did the correct updates, it did it a bit early (not
covering all possible node tree changes).
Now do the ntreeUpdateTree() at the very end of the material import
instead.
Thx @scurest investigating.
Maniphest Tasks: T86314
Differential Revision: https://developer.blender.org/D11159
`SEQ_recursive_apply` and `SEQ_seqbase_recursive_apply` were incorrectly
refactored into `iterator.c` file, but they aren't part and don't use
sequencer iterator.
Functions are moved to `utils.c` file.
Partial fix for T81751 which exposes multiple playback performance
issues. Previously the cache was limited to 30 frames, without a way to
increase the cache for smooth playback with files that are slow to load.
Now the animation plays back smoothly once loaded into cache.
The cache limit from the system preference is used
when the player is launched from Blender.
A new player argument `-c <cache_limit>` was added to support this.
The `file.execute` operator would early-exit because the mouse wasn't hovering
the file list. Caused by 4ba9d7d71e.
Although simpler solutions would have been possible, I decided it's better to
split add a new operator for executing based on the mouse (for double-clicking
files), to separate that from the window level execute operator
(`file.execute`). This allows more control and we can get rid of the implicit
assumption that the keymap would call `file.select` on mouse-press, and
`file.execute` on double-click, for the double-click behavior to work. The cost
is that we execute the file selection & activation logic twice on the
double-click, but that shouldn't be an issue at all.
Also removes the `need_active` property from the `file.execute` operator.
That's ancient and wasn't implemented well anyway.
To be clear, reason this fixes the bug is that `file.execute` works entirely
with the `execute()` callback now and doesn't early-exit based on the mouse
position anymore.
Might trigger warnings about the `need_active` property not being found for
custom keymaps. These can be ignored and the property can safely be removed
from the keymap. I don't expect other keymap breakages.
This was caused by the SSR option resetting the accumulation. But the
render passes were only cleared in the init phase. This means that
when SSR was resetting the `taa_render_sample` the actual renderpasses
would still contains 1 sample. This means the renderpasses were always
divided by the wrong number of samples.
The fix is to clear just before accumulation if the sample is 1.
The fact that it works for motion blur is kind of a blessing. This is because
we check `stl->effects->ssr_was_valid_double_buffer` before resetting the
sampling. So this only happens on the first motion step and does not affect
the rest of the rendering.
Differential Revision: https://developer.blender.org/D11033
This artifact was already present in previous version but was hidden
by the faulty SSS scale.
The issue comes from the translucence using the geometric normal
(computed using fragment shader derivative) leading to poor precision at
depth discontinuity.
Replacing using the same geometric normal reconstruction as the ambient
occlusion pass removes most of the issue.
This was caused by the SSR option resetting the accumulation. But the
render passes were only cleared in the init phase. This means that
when SSR was resetting the taa_render_sample the actual renderpasses
would still contains 1 sample. This means the renderpasses were always
divided by the wrong number of samples.
The fix is to clear just before accumulation if the sample is 1.
The fact that it works for motion blur is kind of a blessing. This is
because we check stl->effects->ssr_was_valid_double_buffer before
resetting the sampling. So this only happens on the first motion step
and does not affect the rest of the rendering.
Reviewed by: jbakker
Differential Revision: https://developer.blender.org/D11033
The render session is keeping track of the scene update, which includes
kernel loading time.
This fixes negative render times reported when CUDA kernels are compiled
at runtime.
A bit fragile logic, can be re-implemented using some user-counted
scope utility classes, so that only outer-most time skip is applied.
Not all python-defined ID properties are overridable (yet), this needs
to be detected by libquery 'foreach id' code, such that those ID
pointers can be ignored by override code when working on override
hierarchies.
Fixes part of the issues found while investigating studio files (namely,
some py-defined ID pointer properties from rigify that are not currently
overridable would cause issues and false detections during resync).
Move the detection/decision of whether an ID pointer should be taken
into account in library override hierarchy processing to the LibQuery
area of code, by introducing a new callback flag.
This allows to factorize the test logic, be explicit in liboverride code
about ID relationships that can be ignored when exploring the override
hierarchy, and adds the possibility to do more checks about pointers to
be tagged as non-overridable in the future.
Note that all but the 'special' ID pointers (loop-back, embedded, etc.)
should be overridable. If some is not, relevant IDType 'foreach_id'
callback code is reponsible to tag it properly.
Python-defined IDProperties however are not systematicaly overridable
(yet), so this should allow us to detect that case and act accordingly
in an incomming commit.
No behavioral change expected in this commit.
Each frame display would add an item to the cache limiting list
without checking if it was already in the list.
Limiting would then free image buffers when the length of the list
exceeded USE_FRAME_CACHE_LIMIT (currently 30).
In practice this meant short animations would free and reload
frames during playback.
In some cases, users start video editing work from General template,
where Filmic view transform is chosen by default. Currently, this causes
issue when working with sRGB footage, which are not properly decoded
into linear working space and final render looks miscolored.
When adding first strip, check it's colorspace and if it corresponds to
colorspace for byte images, ensure that view transform is set to
default value, which is Standard.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D11151
Match the output to the prefilled bug report script's output.
Make the output file self-contained.
Reviewed By: lichtwerk
Differential Revision: https://developer.blender.org/D11144
This patch makes the spline tilts (interpolated to the evaluated points)
affect the evaluated normals, allowing manual control of the rotation of
each profile in the curve to mesh node.
The method is based on Animation Nodes code, which keeps the data in
direction vector form, and rotates around the tangent vector.
Differential Revision: https://developer.blender.org/D11152
While the method of interpolating based on the curve length at each
evaluated point may be useful in some situations, for now it's best to
be consistent with the existing methods in Blender, so this commit adds
a simple linear-based-on-resolution sampling method to bezier splines.
The special case for the interpolation to the last point was being used
for every point in the last segment, because of the rounding. Instead,
make the function slightly more complicated to properly handle the
correct interolation in the cyclic and non-cyclic cases.
Some discard of vertbuf were not correctly followed by discards of
the GPUBatches that were using them. This lead to a use-after-free
situation where GPUBatches would reuse old VBO information. This did
not crash immediatly because the VBO indices were cached by our VAO
caching system. It kept working on some implementation because VBO
reference in the VAO (probably) preventing the VBO from being freed
by reference counting.
This fixes T85977 NVidia: Random crashes in 'DrvPresentBuffers'
The little chevron tab to open a hidden region wouldn't show up in the
Spreadsheet editor. Cause was an incorrect GPU-scissor usage:
While drawing regions, the scissors should be kept enabled, just the
scissor rectangle should be updated - and afterwards reset to what it
was before.
The little chevron tab to open a hidden region wouldn't show up in the
Spreadsheet editor. Cause was an incorrect GPU-scissor usage:
While drawing regions, the scissors should be kept enabled, just the
scissor rectangle should be updated - and afterwards reset to what it
was before.
Reports would change the active emboss of the Block, to make some of its
buttons appear without background. But it didn't reset it so that
buttons added afterwards wouldn't use the correct embossing.
Internally, when using Fill brush a dilate of the filled area was done, but this was hardcoded to 1 pixel.
In some situations, this was not enough, so now the value is accesible in the UI and can be set with different values.
Also, as this value is more used than `Leak Size`, the new Dilate is on Topbar, and Leak Size has been moved to Advanced panel.
In my previous patch https://developer.blender.org/D10171 some code changing the direction the strokes normal was accidentally included. This patch reverts that back to the original normal calculation.
Reviewed By: #grease_pencil, antoniov
Differential Revision: https://developer.blender.org/D11148
Mark NLA/FCurve modifier properties as library-overridable.
It was already allowed to add such modifiers to a library-overridden
object, but then the properties of those modifiers were read-only,
limiting their use.
After copying NLA tracks from one `AnimData` to another, also ensure
that the `AnimData::act_track` and `AnimData::actstrip` pointers are
pointing to the copy rather than the original.
This is a necessary step to allow library overrides on NLA modifiers
without crashing Blender.
The remapping of the pointers is done by looping over the tracks/strips
and comparing pointers. Alternatively, I could update the copy functions
themselves to keep track of those pointers and return them, but IMO that
would produce more spaghetti (they're also used in cases where this
pointer-remapping is not desired).
All changes:
* Include `file.pack_libraries` and `file.unpack_libraries` to the menu
[1].
* Rename "Pack Blender Libraries" → "Pack Linked Libraries" [2].
* Rename "Unpack Blender Libraries" → "Unpack Linked Libraries" [2].
* Rename "Pack All Into .blend" → "Pack Resources" [3]
* Rename "Unpack All Into Files" → "Unpack Resources" [3]
* Rename "☑ Automatically Pack Into .blend" → "☑ Automatically Pack
Resources" [3]
* Rename "Make All Paths Relative" → "Make Paths Relative" [4]
* Rename "Make All Paths Absolute" → "Make Paths Absolute" [4]
* Add separators accordingly
---
[1] - This was never exposed since its original commit rB16411da41e40.
Now that operator not listed in menus don't
show up in the search, this became even more hidden.
[2] - The original name (Pack Blender Library) was not clear enough.
Pose Libraries and Asset Libraries are also technically Blender
libraries.
[3] - The term All was misleading since it didn't include the Linked
Libraries.
[4] - No need to use "All". It is not used in the Report/Find Missing
Files either.
This commit put this in the File > External Data menu.
Differential Revision: https://developer.blender.org/D11109
Text data block were not considered special in the recursive purge
function. So they would get deleted if they had no actual users.
To fix this we instead make text data block use "fake user" so that
addon authors can specify script files that should be removed if nothing
is using it anymore.
Per default, new text object have "fake user" set. So functionality
wise, the user has to explicitly specify that they want the text object
to be purge-able.
Reviewed By: Bastien
Differential Revision: http://developer.blender.org/D10983
Text data block were not considered special in the recursive purge
function. So they would get deleted if they had no actual users.
To fix this we instead make text data block use "fake user" so that
addon authors can specify script files that should be removed if nothing
is using it anymore.
Per default, new text object have "fake user" set. So functionality
wise, the user has to explicitly specify that they want the text object
to be purge-able.
Reviewed By: Bastien
Differential Revision: http://developer.blender.org/D10983
The main goal of this refactor is to not store Object/Collection
pointers for every individual instance. Instead instances now
store a handle for the referenced data. The actual Object/Collection
pointers are stored in a new `InstanceReference` class.
This refactor also allows for some better optimizations further down
the line, because one does not have to search through all instances
anymore to find what data is instanced.
Furthermore, this refactor makes it easier to support instancing
`GeometrySet` or any other data that has to be owned by the
`InstancesComponent`.
Differential Revision: https://developer.blender.org/D11125
This patch adds initial curve support to geometry nodes. Currently
there is only one node available, the "Curve to Mesh" node, T87428.
However, the aim of the changes here is larger than just supporting
curve data in nodes-- it also uses the opportunity to add better spline
data structures, intended to replace the existing curve evaluation code.
The curve code in Blender is quite old, and it's generally regarded as
some of the messiest, hardest-to-understand code as well. The classes
in `BKE_spline.hh` aim to be faster, more extensible, and much more
easily understandable. Further explanation can be found in comments in
that file.
Initial builtin spline attributes are supported-- reading and writing
from the `cyclic` and `resolution` attributes works with any of the
attribute nodes. Also, only Z-up normal calculation is implemented
at the moment, and tilts do not apply yet.
**Limitations**
- For now, you must bring curves into the node tree with an "Object
Info" node. Changes to the curve modifier stack will come later.
- Converting to a mesh is necessary to visualize the curve data.
Further progress can be tracked in: T87245
Higher level design document: https://wiki.blender.org/wiki/Modules/Physics_Nodes/Projects/EverythingNodes/CurveNodes
Differential Revision: https://developer.blender.org/D11091
There was a function to set 2 pixels wide for inverse filling, but this must not be done in the borders of the image.
Now, the borders are checked before set 2 pixels.
Reported for ops.transform.vertex_warp and ops.transform.vertex_random.
Polling will check with ED_transverts_check_obedit, if we know it will
fail e.g. for pose mode, we can do the same check in the UI already.
Maniphest Tasks: T87926
Differential Revision: https://developer.blender.org/D11128
Using scene frame 1 is not reliable in cases when there is a
frame offset is involved. Using frame 1 seems more reliable,
although might still fail under certain circumstances.
More reliable fix would require a deeper change in the data
structure and the logic about frame loading and size detection.
The issue was caused by frame start/offset change triggering clip
reload, which was happening with a hardcoded scene frame index of 1,
which could be outside of the actual clip frames.
Solved by removing source change tag from the frame start/offset
update. The source doesn't really change: the resolution will stay
the same, as well as media type, its duration. So the tag was not
needed.
This precomputes vertex normals in the procedural and caches them in case none
are found in the archive. This only applies to polygon meshes, as subdivision
meshes are yet to be subdivided, so it is useless to do this computation.
The goal here is to speed up data updates between frames, as computing normals
shows up in profiles even for large objects. This saves around 16% of update time
for a production file.
This splits the data reading logic from the AlembicObject class and moves it to
separate files to better enforce a separation of concern. The goal was to simplify
and improve the logic to read data from an Alembic archive.
Since the procedural loads data for the entire animation, this requires looping
over the frame range and looking up data for each frame. Previously those loops
would be duplicated over the entire code causing divergences in how we might
skip or deduplicate data across frames (if only some data change over time and
not other on the same object, e.g. vertices and triangles might not have the
same animation times), and therefore, bugs.
Now, we only use a single function with callback to loop over the geometry data
for each requested frame, and another one to loop over attributes. Given how
attributes are accessed it is a bit tricky to simplify further and only use a
ingle function, however, this is left as a further improvement as it is not
impossible.
To read the data, we now use a set of structures to hold which data to read.
Those structures might seem redundant with the Alembic schemas as they are
somewhat a copy of the schemas' structures, however they will allow us in the
long run to treat the data of one object type as the data of another object
type (e.g. to ignore subdivision, or only loading the vertices as point clouds).
For attributes, this new system allows us to read arbitrary attributes, although
with some limitations still:
* only subdivision and polygon meshes are supported due to lack of examples for
curve data;
* some data types might be missing: we support float, float2, float3, booleans,
normals, uvs, rgb, and rbga at the moment, other types can be trivially added
* some attribute scopes (or domains) are not handled, again, due to lack of example
files
* color types are always interpreted as vertex colors
Workaround for crash when accessing FaceMaps / PaintMask data in
editmode, just disallow access in editmode as is done with UVs.
Same fix as in {rB3e2619b3e72a}.
Maniphest Tasks: T87969
Differential Revision: https://developer.blender.org/D11146
Initial report was mentioning the Classroom demo scene, but this is
probably because the scene was pre-configured to be used with OpenCL.
Would expect any OpenCL compositing to be failing prior to this fix.
The reason why crash was happening is due to OpenCL queue being
released from OpenCLDevice destructor. Is not that obvious, but
when Vector (including std::vector) is holding elements by value
a destructor will be called on "old" memory when vector capacitance
changes.
Solved by making forbidding copy semantic for compositor devices and
forcing move semantic to be used.
Also use emplace semantic in the devices vector initialization.
This commit significantly speeds up many of the attribute nodes when
multiple threads are available in linear situations when parallelism
cannot be achieved elsewhere.
See the differential for a table of timing comparisons tested on a
Ryzen 3700x. For an attribute with 4 million elements, the nodes were
about 3 to 9 times faster.
The changes are not exhaustive, other nodes could still be parallelized
in the future. Also, it would be possible to further optimize the grain
size in `parallel_for`, but I'd rather make sure it isn't too small.
I tested some different values, but also relied on intuition--
increasing grain size for less complex operations and vice versa.
Differential Revision: https://developer.blender.org/D11139
Currently overriding properties within material node trees is not
supported. However there is a workaround that allows feeding values
through drivers via an intermediate custom property, as described
in T82404. The workaround relies on the behavior of the ID copying
code that always patches datablock self-references even without any
overrides.
Unfortunately, this broke during development of 2.93. This happened
because a call RNA_struct_override_matches added in rB2281db72b0157
detects that no override exists, and 'restores' the self-reference
to point to the original datablock.
To avoid this, mark the Material.node_tree property with the
PROPOVERRIDE_IGNORE flag to stop RNA_struct_override_matches
from recursing into the currently unsupported node tree sub-block.
This flag should be removed when this is properly supported.
This was confirmed to fix the workaround and discussed with @mont29.
Shaders are only compiled if they are used by some other Node (Geometry, Light, etc.).
This usage detection is done before updating the Scene, however it fails at detecting
Shaders used by Procedurals not known to Cycles (e.g. ones defined by third party
applications), as Procedurals are only updated after the shaders are compiled.
To remedy this, we now use the Node reference counting mechanism to detect whether a
Shader is used and therefore should be compiled.
This removes `ShaderManager::update_shaders_used` as it is not needed anymore, however,
since it would also update the Shader ids, this is now performed in
`ShaderManager::device_update`, and a new virtual `device_update_specific` method was
added to handle device updates for SVM and OSL.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D10965
This adds a reference count to Nodes which is incremented or decremented
whenever they are added to or removed from a socket, which will help us
track used Nodes throughout the scene graph generically without having to
add an explicit count or flag on specific Node types. This is especially
useful to track Nodes defined through Procedurals out of Cycles' control.
This also modifies the order in which nodes are deleted to ensure that
upon deletion, a Node does not attempt to decrement the reference
count of another Node which was already freed or deleted.
This is not currently used, but will be in the next commit.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D10965
There was a quadratic algorithm extracting triangles from a coplanar
cluster. This is now linear.
Also found and fixed a bug in the same area related to the triangulator
added recently: it didn't get the right correspondence between new
edges and original edges.
Previously the option in the camera's //Object Data Properties > Viewport Display
> Sensor// would not display the sensor in camera view. This seemed to be caused
by the theme color `TH_VIEW_OVERLAY` having zero set for the alpha channel and
alpha blending being active, resulting in no visible output. Hence
`immUniformThemeColorShade(TH_VIEW_OVERLAY, 100);` is replaced with
`immUniformThemeColorShadeAlpha(TH_VIEW_OVERLAY, 100, 255);`.
Reviewed By: mano-wii
Differential Revision: https://developer.blender.org/D11075
Object orientation for thumbnail creation changed to be slightly
oblique (tilted to one side and from above) to better show shape,
especially when axis-aligned. Camera lens changed to 85 to avoid
distortion of close objects like human heads.
see D9940 for details and examples.
Differential Revision: https://developer.blender.org/D9940
Reviewed by Julian Eisel
When saving a file in Edit mode with Multiframe enabled, the render did not include the modifiers.
Now the multiframe is not enabled if it's doing a render.
Resolves occasional glitch/flicker drawing dark buttons in the UI.
Regression in 405a5d3bd7
which removed shader unbinding when the batch is drawn.
GPU_shader_bind could run with the sRGB uniform in an unexpected state.
Reviewed By: fclem
Ref D11124
Multi-overrides of a same linked ID in a same override hierarchy are
currently not supported and can cause all kind of issues.
In some cases they could lead to infinite loop trying to resync the same
ID over and over, this is now prevented.
Found in some Blender studio production files.
Keep track of clog_refs so we can null the pointers when calling
CLG_exit. Otherwise we will run into issues where the code will try to
access freed data.
While it was technically safe to call Map.remove while iterating over
a map, it wasn't really designed to work. Also it wasn't very efficient,
because to remove the element, the map would have to search it
again. Now it is possible to remove an element given an iterator
into the map. It is safe to remove the element while iterating over
the map. Obviously, the removed element must not be accessed
anymore after it has been removed.
This commit adds a new API tha allow to replace the bgl API in the exemples on:
https://docs.blender.org/api/current/gpu.html
**Overview (New API):**
```
gpu.state: active_framebuffer_get
GPUFramebuffer: read_color
GPUOffscreen: texture_color
```
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D11031
Instead of creating different python wrappers for the same GPU object,
return the same `PyObject` created earlier.
This also allows for more secure access to existing GPU objects.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D11044
The formula used to compute the bend did subtraction of two big numbers
to get the position. Changed to find the delta and add that,
by rearranging the formula into a more numerically stable form.
Reviewed By: mano-wii, campbellbarton
Ref D11074
Currently there is an "Auto" option for the domain, this commit adds a
similar option for "Auto" data type, that uses the data type from the
target attribute or the source attribute (in that order).
Ref T87347
Differential Revision: https://developer.blender.org/D10932
Rename BKE_object_runtime_free -> BKE_object_runtime_free_data,
since the runtime pointer is part of the object, only the data is freed.
Leave the data cleared to avoid accidental use,
this is in keeping with other `*_free_data()` functions.
None of these generic poll functions had NULL pointer checks,
since all operators that use these functions expect a valid constraint,
modifier .. etc. Add the NULL check to the poll function.
Ref D11126
Reviewed By: mont29, Severin
If no other object was selected while dragging one in (e.g. from the Outliner
or an object asset from the Asset Browser), all visible objects in the active
view layer would get selected.
Issue was caused by a wrong enum type use.
Mistake in bcdba7c34d.
This adds two new methods:
* `clear` just removes all keys from the vector set.
* `index_of_or_add` returns the index of a key and adds it if has not
been added before.
While fa7ddd0f43 fixed the reported issue,
the possibility of reusing runtime data during curve-to-mesh conversion
remained. Instead of treating the bounding-box as a special case,
clear all run-time data for temporary objects.
Ref D11026
Reviewed By: sergey
The term `verify` doesn't fit with what this function does
and is sometimes used to check data is valid or to control validity
checking as with `RNA_define_verify_sdna`.
use more common term `ensure`.
This adds a callback to bNodeTreeType to check which socket types are
valid for the tree type. Function has been implemented for the normal
tree types, and can be implemented for custom node trees with python,
by adding a `classmethod` to the tree. However, only builtin socket
types are supported.
This is relevant for T87049, but it also has the advantage that it is
now clear which node trees support which sockets. Previously this
was assumed to be known by all developers.
Differential Revision: https://developer.blender.org/D10938
This is just linear interpolation, but it's nice to have an equivalent
to `mix3` for only two values. It will be used for interpolation of
values between bezier spline control points.
Currently when you try to convert a Text-object to Grease pencil from the Object-menu or via the operator in some other way, the Text-object is only converted to a Curve.
This commit converts that curve to a Grease pencil object.
Differential Revision: https://developer.blender.org/D11117
BMesh intersect could leave invalid items in the selection list,
causing a crash. The list is now cleared since boolean is such a
destructive operation, it's unlikely the selection order would be useful.
Thanks to @lukastoenne for finding the root cause.
This patch adds a threshold value to the glow effect in color mode.
Currently, the threshold is hardcoded to 5%.
You can select a color and specify a higher threshold to include
similar colors in the effect.
Note: depends on D10670
Reviewed By: #grease_pencil, pepeland
Differential Revision: https://developer.blender.org/D10672
This patch adds the Randomize options that exist in the Array modifier to the offset modifier.
Currently the patch uses
```
BLI_findindex(&gpf->strokes, gps);
```
to get the index of the current stroke for making each stroke a different seed value. This is how the noise modifier also gets the stroke seed value and it is noted there as well that this method is slow, and should be fixed in the future with another method of getting the stroke index.
Other methods were explored such as using the total number of points of the stroke, but that makes the randomize options incompatible with other modifiers before it such as Multiple Strokes, Array, Build, and Simplify.
{F9591325}
Differential Revision: https://developer.blender.org/D10171
Creating a shallow copy is sometimes useful to get a unique ptr
for a virtual array when one only has a reference. It shouldn't
be used usually, but sometimes its the fastest way to do correct
ownership handling.
When renaming a data-block that is an asset, while the asset is visible in the
Asset Browser ("Current File" asset library), the list wouldn't re-sort items,
breaking the alphabetical sorting.
This was easily possible while changing the data-block name throught the Asset
Browser's sidebar, while in the "Current File" asset library.
Sometimes functions expect a span instead of a virtual array.
If the virtual array is a span internally already, great. But if it is
not (e.g. the position attribute on a mesh), the elements have
to be copied over to a span.
This patch makes the copying process more efficient by giving
the compiler more opportunity for optimization.
FLT_EPSILON was added to the threshold when comparing values,
this caused problems selecting similar small faces since the areas
can be very small in this case.
Also increase the displayed precision so it's easier to use
smaller numbers.
- Allow long enums to break early
- Allow long titles to wrap
- Colloapse hlist to few collumns if needed
This fixes a few of the bigger issues, some more tweaks are still needed
in some places
Instead to create only the Blank object, now a new Layer and a simple material is added.
This is a common request of artists.
Reviewed By: mendio, pepeland
Differential Revision: https://developer.blender.org/D11110
The algorithm that calcualted the direction (inside/outside) of the
polyline was not always returing the correct result. This mean that the
polyline was filled "inside-out".
The fix uses the winding number to calculate the inside and outside.
Reviewed By: antoniov, pepeland
Maniphest Tasks: T87718
Differential Revision: https://developer.blender.org/D11054
In some cases functions were defined with arguments of different array
lengths in headers vs. implementations. This commit fixes some of the
cases I ran into, but probably not all of them.
Issue was caused by anim handle being reset by
`DEG_evaluate_on_framechange()`
Preserve anim handle by backing it up in
`blender::deg::SequenceBackup`
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11059
When frame is rendered to file, bilinear interpolation on scaled images
caused blurred edges.
This is fixed by not doing interpolation with pixels outside of source
buffer.
Such fix doesn't cover cropped images which are currently cropped by
filling area of image with black transparent color, because image buffer
has bigger size than visible image area.
This is fixed by offsetting range used in previous fix by amount, that
would correspond to crop distance. Because of this, image area doesn't
even need to be filled with transparency.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11058
This patch allows you to dynamically control stroke's opacity and thickness using an object for distance reference in the modifier. Fading range is adjustable, and it is compatible with current curve/vertex group selection.
Reviewed By: #grease_pencil, antoniov, mendio
Maniphest Tasks: T82177, T80194
Differential Revision: https://developer.blender.org/D9091
When an aligned (or auto) handle was snapped with only the control
point selected, it would not snap to the correct point, but offset. This
was because the handles were not considered selected. The `TD_SELECTED`
flag was not being set.
The fix makes sure that we include the handles in the selection when
the handle is aligned or auto.
Reviewed By: antoniov
Maniphest Tasks: T86881
Differential Revision: https://developer.blender.org/D11111
Fix for the Arc commands (A/a) to successfully parse the 4th and 5th arguments.
Fix for floats without a specified integer part, like `.123`
File for testing:
{F10042021}
Reviewed By: filedescriptor
Differential Revision: https://developer.blender.org/D11099
This was caused by the new depsgraph persistence.
The GPUbatches we got from the cache being the same for each frame
means that we need to be more careful about cleanning the additional
VBOs references.
Moving the `EEVEE_motion_blur_swap_data` function call at the end of
the loop makes sure the references are cleaned.
Mostly the interesting information about the instances IDs whether they
are -1 or not, but it's still worth displaying them in the editor.
In the future when we can hide or show colums, we could decide to hide
this one by default.
Differential Revision: https://developer.blender.org/D11104
WorkScheduler task model deletes work packages after executing them. The other models don't do so. All models should handle packages the same way.
Reviewed By: #compositing, jbakker
Differential Revision: https://developer.blender.org/D11102
WorkScheduler task model deletes work packages after executing them. The other models don't do so. All models should handle packages the same way.
Reviewed By: #compositing, jbakker
Differential Revision: https://developer.blender.org/D11102
Face set visibility is already flushed to all mesh elements in the right
way in SCULPT_visibility_sync_all_face_sets_to_vertices, calling
BKE_mesh_flush_hidden_from_verts is legacy code from the previous
visibility system that was causing the vertex visibility to take
priority over face sets. When hidding a single loop, all vertices of
those faces are still visibile, so this line was tagging all loop faces
visible. This was leaving mesh/face set visibiltiy in an unconsistent
state.
Reviewed By: JacquesLucke
Maniphest Tasks: T87474
Differential Revision: https://developer.blender.org/D11008
These settings are used though for these strokes (see
'paint_stroke_use_dash'), should be visible in the UI as well.
This was correctly added when dashing was introduced in rB15f82278d5d4
btw., but then messed up in rBfb74dcc5d69d.
Maniphest Tasks: T87816
Differential Revision: https://developer.blender.org/D11096
These settings are used though for these strokes (see
'paint_stroke_use_dash'), should be visible in the UI as well.
This was correctly added when dashing was introduced in rB15f82278d5d4
btw., but then messed up in rBfb74dcc5d69d.
Maniphest Tasks: T87816
Differential Revision: https://developer.blender.org/D11096
This was just not implemented for curve strokes when world spacing was
introduced in rB87cafe92ce2f.
Now do the equivalent of what was done in said commit in
'paint_space_stroke', now in 'paint_line_strokes_spacing' as well.
Maniphest Tasks: T87815
Differential Revision: https://developer.blender.org/D11098
This is a first step towards T87620.
It should not have any functional changes.
Goals of this refactor:
* Move the evaluator out of `MOD_nodes.cc`. That makes it easier to
improve it in isolation.
* Extract core input/out parameter management out of `GeoNodeExecParams`.
Managing this is the responsibility of the evaluator. This separation of
concerns will be useful once we have lazy evaluation of certain inputs/outputs.
Differential Revision: https://developer.blender.org/D11085
This is a first version of an Attribute Transfer node. It only supports two
modes for mapping attributes from one geometry to another for now.
More options are planned for the future.
Ref T87421.
Differential Revision: https://developer.blender.org/D11037
Generally, it would be good to not allow this from happening in the
first place but that is quite tricky because an object does not know
which other object instances it. Similar checks might be necessary
in other places, but this fixes the bug already.
Differential Revision: https://developer.blender.org/D11086
This patch fixes the remaining issues described in T87749. The jitter
was caused by inconsistent rounding when using the floats icon_size
and icon_padding to offset the bound for the text drawing. Using
`round_fl_to_int` leads to consistent results and fixes the jitter
that remained in some buttons with icons, UI lists, and breadcrumbs.
Differential Revision: https://developer.blender.org/D11062
The trouble was that there was a context pointer "modifier" in the
property editor context that returned the active modifier. But the
"modifier" variable was already used in many places, for pointers
that are *not* equivalent to the active modifier.
The context pointer for the active modifier was unecessary anyway.
If we need to access a context pointer for the active modifier in the
property editor then we can add it. Until then it only adds confusion.
Counts of less than one weren't allowed in end points mode mostly to
avoid a division by zero when calculating the delta. It's trivial to
allow a count of one, so this commit does that, with the point placed
at the start location.
This is a change split from the geometry nodes curves branch. Since
curve object modifier evaluation happens in this file, moving it to C++
will be quite helpful to support the `GeometrySet` type. Other than
that, the code in the branch intends to replace a fair amount of this
file anyway, so I don't plan to do any further cleanup here.
Differential Revision: https://developer.blender.org/D11078
This allows us to remove a callback from the modifier type info struct.
In the future the these modifiers might just be replaced by nodes
internally anyway, but in the meantime it's nice to unify the handling
of evaluated geometry a bit.
Differential Revision: https://developer.blender.org/D11080
The Outliner was doing a full rebuild of its tree in response to transform
notifiers. I don't see any reason for this, a simple redraw without rebuilding
should be just fine.
The same optimization could be done for other object notifiers, but I'll check
on them separately.
Caused by {rB278011e44d43}.
Framebuffer management since above commit now seems to require region
bind/unbind in for the operator to be able to redraw correctly without
using the same framebuffer in multiple contexts.
Maniphest Tasks: T87771
Differential Revision: https://developer.blender.org/D11084
This fixes T87666 and T83252.
The boolean modifier and geometry nodes can depend on the geometry
of an entire collection. Before, the modifiers had to manually create relations
to all the objects in the collection. This worked for the most part, but was
cumbersome and did not solve all issues. For example, the modifiers were not
properly updated when objects were added/removed from the referenced collection.
This commit introduces the concept of "collection geometry" in the depsgraph.
The geometry of a collection depends on the transforms and geometry of all
the objects in it. The boolean modifier and geometry nodes can now just depend
on the collection geometry instead of creating all the dependencies themselves.
Differential Revision: https://developer.blender.org/D11053
Though to my knowledge we haven't had a report about this yet, this
looks like a clear oversight-- the ids are just more data stored by the
instances component, and should be cleared and copied like other data.
This might have resulted in incorrect random IDs for instances in
renderers in some cases where the component had to be copied.
Correction to Prefs tooltip mention of 'add-ons' folder in Scripts Dir.
Differential Revision: https://developer.blender.org/D11064
Reviewed by Harley Acheson
Removing mid-operation area re-targeting for safety and predictability.
Differential Revision: https://developer.blender.org/D11066
Reviewed by Campbell Barton
Calculation of bounding rect for multi-input socket was wrong.
Reviewer: Hans Goudey (HooglyBoogly)
Differential Revision: https://developer.blender.org/D11077
Calculation of bounding rect for multi-input socket was wrong.
Reviewer: Hans Goudey (HooglyBoogly)
Differential Revision: https://developer.blender.org/D11077
This is convenient because having a uniform interface is nice, and
because of the similarity to "last".
Differential Revision: https://developer.blender.org/D11076
If an object named for example `Suzanne` is converted to Gpencil, a material called `Suzanne_Fill` will be created for the gpencil fill.
When this material already exists, the new material will be called `Suzanne_Fill.001` and the operator will not see that this material is already present the next iteration. This leads to a new material being created for every polygon.
This commit changes the code to search for a material starting with `ObjectName_Fill` instead of being equal to.
Reviewed By: filedescriptor, antoniov
Differential Revision: https://developer.blender.org/D11067
This reverts commit 9cce18a585.
rB9cce18a5858c broke the build in unforeseen ways, not easily fixable.
revert for now, so master will at least build again.
- Re-order freeing so an instances __del__ method runs before the
`ExtensionRNA` has been freed.
- "remove" functions no longer free the gizmo/gizmo-group memory,
needed so the identifier used when freeing the extension
doesn't use the freed identifier.
This patch fixes jittering text when resizing regions of the UI as
described in T87749. The jitter was caused by the text padding
being stored as an integer which lead to inconsistent rounding.
Most notably this patch fixes the jitter in the new spreadsheet editor,
but not all occurrences of jitter described in T87749 (e.g. in UI lists)
are addressed.
Differential Revision: https://developer.blender.org/D11060
Reset the active strip offset and scale, in the "Set Render Size"
operator (`SEQUENCER_OT_rendersize`). This ensures that the active strip
will actually fit the new render size, which is what the operator is
intended to achieve.
This is complex situation. Tagged ID deletion (used to delete several
data-blocks at once) removes IDs to be deleted from Main.
But when we remove deleted IDs' usages of other IDs (using
`BKE_libblock_relink_ex`), some specific post-process is required on
some types, like Collections. Those post-processes would in some cases
rely on data actually being in Main.
That failing condition would lead in existing code on missing processing
the very ID (collection) we were working on, leading to missing removing
some child collection pointers, leading to the crash (later on in
LayerCollection resync process).
For now we go with an optimization & fix that avoids processing all
collections in Main when we actually know which one we are working one
(case of `BKE_libblock_relink_ex`, but not of
`BKE_libblock_remap_locked`).
This is however yet another demonstration of the need to rework that
whole collection/layer resync process, since it is not only extremely
inneficient currently, but it also requires valid Main/ID state way too
deep into the remapping code.
NOTE: This fix may very well not catch/address all possible fail cases
here, dealing with the double parent/child relationships of collections
is challenging...
Issue reported by @eyecandy from the studio, thanks.
In cases where the same node tree is used in different materials with
small changes, the wrong material could be selected.
Cause: Hair shaders GPU resources weren't updated and used the previous bound data.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D11036
If saving a file using CloseSave dialog, do not close if there is an error doing so, like if read-only.
Differential Revision: https://developer.blender.org/D10722
Reviewed by Julian Eisel
The grid plane was drawn too big on retina displays compared to other screens,
because the factor was multiplied by the native pixel-size, which is 2 for
Retina displays.
Makes the functions (introduced in 557b3d2762) follow existing naming
conventions for the API.
Changes:
`bpy.types.ID.mark_asset` to `bpy.types.ID.asset_mark`
`bpy.types.ID.clear_asset` to `bpy.types.ID.asset_clear`
When the user entered a driver expression like `#frame` into a number
field, Blender would crash sometime after. This is because
e1a9ba94c5 did not handle this case properly and ignored the fact
that the user can enter text into the field.
The fix makes sure that we only cancel if the value is a *number* equal
to the previous value in the field.
Note that this could also be handled by `ui_but_string_set` which would
be a bit nicer potentially. However, I was unable to find a clean way of
passing `data->startvalue` into that function. `uiHandleButtonData` is
only known locally inside `interface_handlers.c`.
Reviewed By: Severin
Maniphest Tasks: T87688
Differential Revision: https://developer.blender.org/D11049
Previously we always had to set attribute values after creating
the attribute. This patch adds an initializer argument to
`attribute_try_create` which can fill it in a few ways, which
are explained in code comments.
This fixes T87597.
Differential Revision: https://developer.blender.org/D11045
Because we use virtual classes (and for other reasons), we had to do a
small allocation when simply retrieving the data type and domain of an
existing attribute. This happened quite a lot actually-- to determine
these values for result attributes.
This patch adds a simple function to retrieve this meta data without
building the virtual array. This should lower the overhead of every
attribute node, though the difference probably won't be noticible
unless a tree has very many nodes.
Differential Revision: https://developer.blender.org/D11047
The code of the snapping system to interact the objects in the scene only
considers instances what comes from "DUPLI" objects.
This commit adds instances coming from Geometry nodes.
Differential Revision: https://developer.blender.org/D11020
Speed effect caused, that some raw frames are re-used for multiple
final frames. When cached final frame is freed due to memory being
full, it tried to free also lower level cached frames that were used
during compositing. Some lower level cached frames were already freed
by different final frame and `BLI_ghash_remove()` failed.
Check if key still exists in hash or if linked keys were overwritten
before removing them.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D10909
Under some circumstances, modifiers are evaluated more than once.
One time to compute the actual output geometry and another time
with `MOD_APPLY_ORCO`. This design probably has to be revisited
at some point in the context of geometry nodes. However, that would
be much more involved than a bug fix.
The issue was that during the second evaluation, the node tree is
evaluated based on a slightly different input geometry. The data
generated during the second evaluation overwrote the cached
data from the first evaluation, resulting in incorrect data that is
shown in the spreadsheet.
The fix for now is to simply not log any data in the second evaluation.
Add a keying set that includes location, rotation, scale, and custom
properties.
The keying set is intentionally not based on the "Whole Character"
keying set. Instead, it is generic enough to be used in both object and
pose animation, making it possible to quickly switch between animating
characters and props without switching keying sets.
This is, according to @rikkert, what the Blender Studio animators need
99.9% of the time.
The `bisect_distance` in the mirror modifier was hard-coded to `0.001`.
This would result in some unexpected behavior like vertices close
to the mirror plane being deleted or merged.
The fix now adds a parameter to the mirror modifier to expose the
bisect distance to the user. The default is set to the previous
hard-coded value to not "change" previous files.
Ref D10201
When cache is strip is invalidated, movie file was reloaded even if it
isn't necessary. This caused significant performance issues when strip
is being dragged under playhead.
This was caused by calling `SEQ_relations_sequence_free_anim()` and it
was introduced as fix for T36124.
When it is necessary to reload file because another API holds reference
to ImBuf, do this explicitly besides cache invalidation.
In `rna_ColorManagedColorspaceSettings_reload_update()` this was already
done, so no change is needed there.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11024
Math for drawing font over byte buffer was incorrect. Effect can be seen
when target buffer is fully black and transparent - this results in font
color being effectively premultiplied, which causes problems when image
is composited further.
Use `blend_color_mix_byte()` and `blend_color_mix_float()` for blending.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11035
Camera object used for rendering reffered to original not evaluated data.
Use `DEG_get_evaluated_object()` to get evaluated camera.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11039
Was showing a simple confirm dialog, even if the file didn't contain
unsaved changes. The confirm dialog would also show up when choosing
"Restore Last Session" from the splash screen right after startup, which
is weird.
Instead show the proper file closing dialog that allows saving, but only
if there are actually unsaved changes.
Always use the defaults here (radius, depth etc), since desired bounds
have been set interactively, it does not make sense to use a different
value from a previous command.
The Cube tool has already seen a fix for this in rB26e5718e29a7, but
Cone/UVSphere/Cylinder/IcoSphere havent.
Maniphest Tasks: T87677
Differential Revision: https://developer.blender.org/D11038
The UV factor of the last point of a cyclic stroke was using the factor of
the first point leading to unwanted scaling artifacts.
The fix sets the uv factor of the last point to the currect value (last UV
factor + length between last and first point).
Reviewed By: antoniov, fclem
Maniphest Tasks: T86968
Differential Revision: https://developer.blender.org/D10850
Properties Editor is a bit more picky (compared to the Outliner for
example) when it comes to listening to grease pencil notifiers -- it
requires the action to be set.
So when adding the notifier in the dopesheet from the channels (done in
`achannel_setting_flush_widget_cb`), now add the `NA_EDITED` action.
Maniphest Tasks: T87640
Differential Revision: https://developer.blender.org/D11025
Some curve modifiers (namely Hook, SoftBody and MeshDeform) can only work
on pre-tesselated spline points.
Before the modifier UI refactor in rB9b099c86123f, users would get the
'Apply on Spline' option, but disabled and with a tip explaining why
this cant be changed. After rB9b099c86123f though, this button was
always enabled [but disfunctional] leaving the user without an
explanation why this has no effect.
Now restore this functionality since it is quite important information.
Additionally, this button now appears to be ON in these cases which
makes more sense from the user perspective (so it does not represent the
actual setting on the modifier -- this would internally be switched ON in
the modifier calculation anyways though, see
'curve_get_tessellate_point')
Differential Revision: https://developer.blender.org/D11029
Adds `mark_asset()` and `clear_asset()` to ID data-blocks, e.g.
`bpy.context.active_object.mark_asset()`. They essentially do the same as the
mark and clear asset operators.
Scripts are generally discouraged from using operators where possible, but we
need to provide API functions to use instead. In this case it means scripts
don't have to override context to pass an ID to the operator.
This was caused by the closure refactor. The radiance being masked
for SSR, we need to not enable SSR when trying to render the specular
color pass.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11028
Move code common to the Whole Character keying sets ("Whole Character" and
"Whole Character (Selected Bones Only)" into a mix-in class. This avoids
the need to use direct assignments like
`poll = BUILTIN_KSI_WholeCharacter.poll`.
No functional changes.
Dragging a number button, then holding the value and pressing escape
would not reset the value correctly.
This was because eb06ccc324 assumed that `data->value` and
`data->startvalue` were set during dragging which they are not.
The fix moves the if statement into the section where we check if a
number was entered (number edit) making sure that we only cancel
if the button was in "string enter" mode and that the value entered
was the same as before.
Reviewed By: HooglyBoogly, Severin
Maniphest Tasks: T87637
Differential Revision: https://developer.blender.org/D11021
If a new action is created (e.g. by inserting keyframes), the header was
not garuanteed to display the right action.
Notifiers were fine here, redraw took place, the editors action was just
not set soon enough for the drawing.
Now make sure this is set correctly by ensuring the animation context is
right (same as the other editor regions do before drawing).
Maniphest Tasks: T86809
Differential Revision: https://developer.blender.org/D10796
Increase range of internal flags & order UI_SEARCH_FILTER_NO_MATCH
within this range, so public button flags aren't accidentally added
that overlap with internal flags.
Python scripts can now define the reason it's poll function fails using:
`Operator.poll_message_set(message, ...)`
This supports both regular text as well as delaying message creation
using a callback which should be used in situations where constructing
detailed messages is too much overhead for a poll function.
Ref D11001
For indirect light rays, don't assume any hit is opaque, rather if it has
transparency or emission do the shading but don't do any further bounces.
Naturally this is slower when there are transparent surfaces, however
without this cutout opacity doesn't give sensible results.
Differential Revision: https://developer.blender.org/D10985
Some persistent data code was disable due to a deeper design issue, which
meant some updates were not communicated to renderers.
Dependency graph updates work in two passes, once where Blender scene
animation updates are done, then app handler scripts can run to make further
scene modifications, and then the depsgraph is updated again to take those
into account.
Previously the viewport would update renderers twice when such app handler
scripts were present. Now both viewport and persistent data rendering update
the renderers only once, accumulating updates from both passes.
The function applying the search used the dummy search info for when
the item doesn't exist even when there was no UI data associated with
the node at all.
A fix is to only add the search menu when there is attribute info
stored for the node. This is something I wanted to do anyway, since
it makes it look more purposeful when there are no attribute info
for a node, less like a bug.
Differential Revision: https://developer.blender.org/D11016
The problem is that each uiBlock needs to be assigned a unique name,
but when there can be multiple modifiers of the same type, we use the
panel sort order for the unique part of the string. However, the most
recent test file has 1200+ panels in the property editor, so 4
characters isn't enough for a unique string.
That's not a situation I expected, but it makes sense, because we don't
remove legacy panels with unused types when loading old files. So they
tend to accumulate a bunch of unused panels. That's why this works fine
with a new property editor, it doesn't all of the extra old panels.
These panels must be stored for the expansion status and order, but
arguably we could cull unused panels on save. However, simply increasing
the length of the unique panel string is a valid fix in this situation.
In the future, we can look into removing unused panels when saving.
Previously, clicking into a number field, changing nothing and then
clicking outside the field again would trigger an update (RNA prop
would be set to the same value again). This could potentially cause
an expensive operation (like a modifer) to run again, even if all the
parameters were identical.
The fix prevents this by treating unchanging values in the field as a
cancel operation.
Reviewed By: Severin
Maniphest Tasks: T87448
Differential Revision: https://developer.blender.org/D10976
This is a first iteration of a switch node. It can only switch between
two inputs values based on a boolean. A more sophisticated switch
node that has an integer selector will probably come later.
Currently, the geometry nodes evaluator does not support lazy evaluation
of individual inputs. Therefore, all inputs will be computed currently.
An improvement to the evaluator will be worked on separately.
Ref: T85374
Differential Revision: https://developer.blender.org/D10460
Prevent drag events from changing the highlighted gizmo
unless the drag event activates the gizmo.
This resolves a glitch where testing a drag event would highlight
at the point the drag was initiated even when the event was not handled.
The previous code had unclear hacks to avoid updating while transforming,
it was also duplicated in two functions causing an inconsistent
initialization of the looptris bvhtree (which could even generate
unpredictable snapping results).
Now, detection update and inicializatiom of common members are contained in
`snap_object_data_mesh_get` and `snap_object_data_editmesh_get`.
Also, the "Hack to avoid updating while transforming" is more evident.
This method is similar to `std::vector::emblace_back` in that it constructs
the new object inplace in the vector, removing the need for a move.
The `_as` suffix is consistent with similar behavior in Map and Set data structures.
It is important to limit the pixels read on `BLI_array_iter_spiral_square`.
Fortunately `ED_view3d_depth_read_cached` was not being called with a
`margin` parameter.
Wrong logic introduced in rB44c76e4ce310.
It is important to limit the pixels read on `BLI_array_iter_spiral_square`.
Fortunately `ED_view3d_depth_read_cached` was not being called with a
`margin` parameter.
Wrong logic introduced in rB44c76e4ce310.
This allows us to build more complex values in-place in the map.
Before those values had to be build separately and then moved into the map.
Existing calls to the Map API remain unchanged.
A virtual array is a data structure that is similar to a normal array
in that its elements can be accessed by an index. However, a virtual
array does not have to be a contiguous array internally. Instead, its
elements can be layed out arbitrarily while element access happens
through a virtual function call. However, the virtual array data
structures are designed so that the virtual function call can be avoided
in cases where it could become a bottleneck.
Most commonly, a virtual array is backed by an actual array/span or
is a single value internally, that is the same for every index.
Besides those, there are many more specialized virtual arrays like the
ones that provides vertex positions based on the `MVert` struct or
vertex group weights.
Not all attributes used by geometry nodes are stored in simple contiguous
arrays. To provide uniform access to all kinds of attributes, the attribute
API has to provide virtual array functionality that hides the implementation
details of attributes.
Before this refactor, the attribute API provided its own virtual array
implementation as part of the `ReadAttribute` and `WriteAttribute` types.
That resulted in unnecessary code duplication with the virtual array system.
Even worse, it bound many algorithms used by geometry nodes to the specifics
of the attribute API, even though they could also use different data sources
(such as data from sockets, default values, later results of expressions, ...).
This refactor removes the `ReadAttribute` and `WriteAttribute` types and
replaces them with `GVArray` and `GVMutableArray` respectively. The `GV`
stands for "generic virtual". The "generic" means that the data type contained
in those virtual arrays is only known at run-time. There are the corresponding
statically typed types `VArray<T>` and `VMutableArray<T>` as well.
No regressions are expected from this refactor. It does come with one
improvement for users. The attribute API can convert the data type
on write now. This is especially useful when writing to builtin attributes
like `material_index` with e.g. the Attribute Math node (which usually
just writes to float attributes, while `material_index` is an integer attribute).
Differential Revision: https://developer.blender.org/D10994
Now if one stroke in the extremes of the stack is selected, other strokes are moved.
Reviewed By: pepeland, Dantti
Maniphest Tasks: T87321
Differential Revision: https://developer.blender.org/D10997
This adds support for mutable virtual arrays and provides many utilities
for creating virtual arrays for various kinds of data. This commit is
preparation for D10994.
Caused by {rB571362642201} where versioning code for new sequencer tool
settings was only done for scenes already having sequencer scene->ed.
If scene->ed was not present, sequencer tool settings were never
initalized for this scene [if the VSE was then used later], leading to
crashes in some places.
Now just use the versioning code to initalize sequencer tool settings
for all scenes not having them yet.
Maniphest Tasks: T87010
Differential Revision: https://developer.blender.org/D10996
Click-drag events that weren't handled would continually be tested
for each mouse-motion event.
As well as being redundant, this added the overhead of querying
gizmos twice per motion event.
Now click-drag is only tested once when the drag threshold is reached.
This mitigates T87511, although the single drag test still causes
the snap gizmo to flicker.
The intention with this API function was to temporarily load
ID's tagged LIB_TAG_TEMP_MAIN,
however the way the `real_main` was used,
these ID's were loaded into the G.main.
`ob->runtime.geometry_set_eval` can contain instances as well.
This only affected instances generated by geometry nodes.
We should probably have a separate function that tells us if an object
has instances or not..
Compilation fails when our OCIO wrapper creates a shader that
transfer first to scene ref and directly after that to display.
This cause is that the GPU resources of both transfers had the same
name. This is fixed by prefixing the resources.
This can be reproduced by loading a movie file (mkv) in the VSE editor.
Reported by Sergey Sharybin.
In the past, custom attributes were rarely used in practice, because the
only way to use them was from Python. Since geometry nodes, more
users started to add their own attributes. Those attributes should not
be removed automatically. It is still possible to remove them in
geometry nodes explictly to improve performance.
This has technically been fixed by rB3e87d8a4315d794efff659e40f0bb9e34e2aec8a,
but the fix there is questionable, because it disables an optimization for vertex groups
entirely. This fix is a little bit more precise in that it only disables the optimization when
the object is used by some geometry nodes modifier.
This flag is needed so PointerRNA structs that aren't
part of the current context can access enum values
without inspecting the context.
This is needed for keymap access, so the keymap UI and keymap
export doesn't depend on the current context.
This changes the following items:
- package name is now `blender-3.0.0-git.09eb04c0a865-windows64`
rather than `blender-3.00.0-git.09eb04c0a865-windows64`
- Fix version resource for blender.exe not building
- Data directories are now `3.0\...` rather than `3.00\....`
- User prefs are now in:
`c:\Users\users\AppData\Roaming\Blender Foundation\Blender\3.0\`
rather than:
`c:\Users\users\AppData\Roaming\Blender Foundation\Blender\3.00\`
- Updating startup & preferences from previous release
has a special exception for 3.0 to check for 3.93 and older.
See T87532
Ref D10986
`make lite` does not use audaspace, so some files will pollute the console with
tons of warnings about audio values not defined.
Reviewed By: nexyon, campbellbarton
Differential Revision: https://developer.blender.org/D10981
The exporting was creating a pdf state for each stroke, but this was necessary only for strokes with opacity.
Now, the state is only created when needed and remove the state variable from class.
Also, avoid exporting invisible strokes.
Blender 3.0 is now in bcon1 (alpha).
There are likely a few places in Blender and the automated building pipeline
that may fail since we are switching our versioning number system.
For example, at the moment the splash and the status bar are showing
3.00.0, and it should show 3.1.0.
I suspect the Python API, version used to report a bug, buildname, are
all wrong too. These will be handled later.
Since these were added, we decided that builtin and reserved name
attributes start with a lowercase letter. We also use "id" already in
a few nodes, so this change will be consistent with that.
When inserting a node on top of a link, the multi input socket
indices weren't updated. This fixes that and keeps the relative
order of the links the same.
Author: Wannes Malfait
Reviewed By: Fabian Schempp
Differential Revision: https://developer.blender.org/D10969
T87356 occured because last_node_hovered_while_dragging_a_link was not
set on node_link_init.
The assert in T87358 failed because the sorting operation was called even
if the drag link contained a link to another socket.
A little one frame jump was caused because frame was refreshed after
picking a link and before sorting happened.
Reviewer: Hans Goudey
Differential Revision: https://developer.blender.org/D10940
This introduces a context path to the spreadsheet editor, which contains
information about what data is shown in the spreadsheet. The context
path (breadcrumbs) can reference a specific node in a node group
hierarchy. During object evaluation, the geometry nodes modifier checks
what data is currently requested by visible spreadsheets and stores
the corresponding geometry sets separately for later access.
The context path can be updated by the user explicitely, by clicking
on the new icon in the header of nodes. Under some circumstances,
the context path is updated automatically based on Blender's context.
This patch also consolidates the "Node" and "Final" object evaluation
mode to just "Evaluated". Based on the current context path, either
the final geometry set of an object will be displayed, or the data at
a specific node.
The new preview icon in geometry nodes now behaves more like
a toggle. It can be clicked again to clear the context path in an
open spreadsheet editor.
Previously, only an object could be pinned in the spreadsheet editor.
Now it is possible to pin the entire context path. That allows two
different spreadsheets to display geometry data from two different
nodes.
The breadcrumbs in the spreadsheet header can be collapsed by
clicking on the arrow icons. It's not ideal but works well for now.
This might be changed again, if we get a data set region on the left.
Differential Revision: https://developer.blender.org/D10931
The fact that geometry from instnances isn't realized when applying
a nodes modifier can be very confusing, especially for new users.
Nodes themselves realize geometry instances implicitly whenever they
need to. We also currently make instances real and convert points to
mesh when a modifier is added after the nodes modifier. With this
commit, we simply do the same thing when applying the modifier.
There are a few downsides though:
- This can be an extremely heavy operations in some cases where
geometry nodes is used to instance heavy geometry.
- We will still have the issues with materials, since instances use
materials from their original objects, but real geometry uses
materials from the modifier object.
It was decided to live with the potential performance downsides
for now, the idea is the upsides of the change are more important,
and people making complicated setups will be more likely to know not
to apply the modifier. In the future there could be a warning if it's
necessary though.
Ref T87083
This will hopefully make it faster to start using the node editor.
The workspace also includes the spreadsheet. Note that while
existing startup files won't have this workspace added, it will be
in the list of options when clicking on the "add workspace" button.
Ref T86499
Now, instead to offset the stroke color to make it visible over fill, the stroke keeps the original color and the fill is offset.
Related to the issue in T87406.
The color of the strokes was shifted when using `Single Color` or
`Object Color`in Solid Display Mode. This was originally done so
that it was still possible to differentiate strokes and fills.
The fix allows for a completly flat shading by checking if the
Lighting Mode is set to `Flat`. If it is, then the colors are not
shifted for the strokes and everything will use the same color.
Reviewed By: antoniov, fclem
Maniphest Tasks: T87406
Differential Revision: https://developer.blender.org/D10957
This patch adds domain and data type information to each row of the
attribute search menu. The data type is displayed on the right, just
like how the list is exposed for the existing point cloud and hair
attribute panels. The domain is exposed on the left like the menu
hierarchy from menu search.
For the implementation, the attribute hint information is stored as a
set instead of a multi-value map so that every item (which we need to
point to descretely in the search process) contains the necessary data
type and domain information by itself. We also need to allocate a new
struct for every button, which requires a change to allow passing a
newly allocated argument to search buttons.
Note that the search does't yet handle the case where there are two
attributes with the same name but different domains or data types in
the input geometry set. That will be handled as a separate improvement.
Differential Revision: https://developer.blender.org/D10623
Within the 2D Animation workspace, the 'Cursor' dropdown currently displays the 'Display Cursor' incorrectly aligned.
This change fixes the alignment issue.
Reviewed By: #user_interface, #grease_pencil, antoniov
Differential Revision: https://developer.blender.org/D10614
GOP size and quality are adjusted for h264 codec.
These new values are based on result of benchmark on 9 random files:
https://docs.google.com/spreadsheets/d/1nOyUGjoVWUyhQ2y2lAd8VtFfyaY1wQNGj1krCCNbk7Y/edit?usp=sharing
Reducing quality to 50 reduces proxy filesize by about 2x on average
and has no significant impact on decoding performance.
Increasing GOP size from 2 to 10 also reduces proxy filesize 2x-3x
while scrubbing is only about 8% slower. It is still around 100FPS
with 1920x1080 media.
This is unfortunately about 50% slower than MJPEG, but this can be
improved with `fastdecode` tune applied to libx264 encoder
Quite surprisingly h264 codec presets had little influence on proxy
building performance as well as proxy filesize. So far it looks that
FFmpeg does initialize encoder in different way then Blender.
This applies mot only for presets but for tune and profile libx264
setting.
Once this issue is resolved, performance of proxies may be optimized
further.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D10897
When trying to export a GPencil object to SVG from outside the camera
view, the expoted file would contain invalid data. This was because the
calculation of the bounding box did not have any objects to
iterate over.
The fix makes sure we create the object list before trying to calculate
the bounding box.
Reviewed By: antoniov
Maniphest Tasks: T87479
Differential Revision: https://developer.blender.org/D10975
Some operations may use no preferredResolution ({0, 0}) when calling determineResolution on inputs to check if they have resolution on their own. See MixOperation or MathOperation determineResolution implementation. In such cases {0, 0} resolution ends up being set when an input doesn't have own resolution, breaking propagation of the original preferredResolution. They don't mean to set it as resolution, it's just a check.
This patch only allows to set valid resolutions (>0). When it's 0 it may be understood as "No preferred or determined resolution" so it should not be set to give output operations another chance of finding a proper resolution by calling determineResolution again with a different preferredResolution.
Test file:
{F9932526}
Reviewed By: #compositing, jbakker
Differential Revision: https://developer.blender.org/D10972
The SB back-end optimizer for the mesa R600g driver corrupts the vertex
shader for widget drawing. This will not be fixed upstream because SB is
getting replaced as part of the new NIR path. This was thought to be an
issue with instancing and an attempted fix was submitted in D8374, but
it did not fix the issue.
This patch reimplements the array look-up part of the code using switch
case as a workaround and removes the old workaround implemented as part
of D8374.
Reviewed By: Clement Foucault
Differential Revision: https://developer.blender.org/D10967
With this patch, users can define custom tooltips for the exposed
properties of their Geometry Nodes Groups.
Currently this custom tooltips are only used in the modifier panel,
but its a long term goal to use it in the node editor.
Reviewer: Hans Goudey
Differential Revision: https://developer.blender.org/D10884
This implements the changes discussed in T87134 for including switch
object funcionality in 2.93. This includes:
- Remove the switch object operator experimental option
- Remove the option for switching objects in Edit Mode.
- Rename switch_object to transfer_mode.
- Enable the operator only in sculpt mode.
- Expose the operator in the Sculpt menu with an eyedropper modal
option.
On later releases, we could revisit enabling the operator in other mode
and object types as well as its place in the UI.
Reviewed By: JulienKaspar, JacquesLucke
Differential Revision: https://developer.blender.org/D10953
Sculpt menus were using the legacy expand operator. This updates all
entries to use the new one.
It does not make sense to use Expand from the menus as it relies on the
brush cursor position, but now all Expand shortcuts are listed in the
UI.
Reviewed By: JacquesLucke
Differential Revision: https://developer.blender.org/D10952
Previously, the bone position outside of "fit to curve length" mode was
incorrect.
It assumed that the curve was completely straight with no bends or
turns. This would lead to bones being scaled down as their final
position would be servery underestimated in some cases.
The solution is to do a sphere -> curve intersection test to see where
to put the bones while still preserving their length. As we are using
the tessellated curve data this essentially boils down to us doing a
sphere -> line intersection check.
Reviewed By: Sybren
Differential Revision: http://developer.blender.org/D10849
Add a `source_archive_complete` target for `make` that creates a source
archive including the source packages of Blender's dependencies.
This expands `make_source_archive.py` to include files from
`${BUILD_DIR}/source_archive/packages/` as well.
Reviewed By: dfelinto
Maniphest Tasks: T86124
Differential Revision: https://developer.blender.org/D10727
This is done to ensure building with newer OptiX SDK releases that add new struct fields gives
deterministic results (no uninitialized fields and therefore random data is passed to OptiX).
Pass `bContext *C` a bit further down the call stack, to prevent
exploding the number of parameters of `ED_preview_icon_render()`. An
upcoming change will require access to yet another context member, and
this can now be done without adding yet another parameter.
No functional changes.
Adds appropriate checks/guards around all the untrusted parameters
which are used for reading from memory.
Validation:
- All the crashing files within the bug have been checked to not causes
crashes any longer>
- A handful of correct .bmp were validated: 3 different files at each
of 1, 4, 8, 24, 32 bpp depth along with a random variety of other 24
bpp files (around 20 in total).
- ~280 million iterations of fuzzing using AFL were completed with 0
crashes. The old code experienced several dozen crashes in first
minutes of running {F8584509}.
Ref D7945
Entering the numbers in the keyframe field was not triggering a viewport
update, so the little green lines down in the cache area were only
change when the user e.g. changed the frame.
Now send appropriate notifiers.
Issue spotted in T86456 by sebastian_k.
Maniphest Tasks: T86456
Differential Revision: https://developer.blender.org/D10963
- Optionally get the error as a single line.
- Support access the error as an allocated string.
- PyC_ExceptionBuffer_Simple was always printing to the `stdout` while
PyC_ExceptionBuffer didn't, now either print to the output.
Without this, callers are unable to do anything with the error string.
This commit cleans up the RNA names of select mirror operators so that
they are all "Select mirror".
This makes the select menu in edit/pose mode consistent regardless of
object type.
Differential Revision: https://developer.blender.org/D7356
This is especially useful when trying to add a node group instance, e.g. via
drag & drop from the Outliner or Asset Browser.
Previously this would just silently fail, with no information why. This is a
source of confusion, e.g. earlier, it took me a moment to realize I was
dragging a node group into itself, which failed of course.
Blender should always try to help the user with useful error messages.
Adds error messages like: "Nesting a node group inside of itself is not
allowed", "Not a compositor node tree", etc.
Adds a disabled hint return argument to node and node tree polling functions.
On error the hint is reported, or could even be shown in advance (e.g. if
checked via an operator poll option).
Differential Revision: https://developer.blender.org/D10422
Reviewed by: Jacques Lucke
Allow users to undo the effect of the "Generate Preview" operator in the
asset browser (`ED_OT_lib_id_generate_preview`). Without this, the
button is too dangerous.
A few strings describing RNA objects were wrong, including copy/paste errors, spelling and case.
Reviewed By: Blendify
Differential Revision: https://developer.blender.org/D10899
Changing window state using taskbar system menu could result in a titleless window.
Differential Revision: https://developer.blender.org/D10812
Reviewed by Ray Molenkamp
Before rBf674976edd88, the flag indicating whether a curve was 2D or 3D was
ignored by Surfaces objects.
So it can be said that Surfaces objects were always 3D.
We could remove updates to 2D on Surface objects, so the behavior is
identical to what it was before.
But this would also cause the return of `data.dimensions` to be misleading,
complicate the code a bit and add a micro overhead.
So the solution here is just to init all Surface objects as 3D.
Surface objects can now be constrained to 2D with the command:
```
data.dimensions = '2D'
```
The mixing function was designed to give correct results for Mix values of
-1, 0, and +1, but the behavior between these points was not linear. This is
unavoidable, because the function depends on both Mix and Mix^2 (by
multiplying value and mf) so they could not cancel out completely.
The new formula simply calculates the weighted sum without trying to invent
a smooth function.
Value for MixGlareOperation is now passed directly without scaling because
it is then easier to use.
Note that the previous formula performed max() twice for both input image
and the result, now there is just one max() per channel because the glare
input can't be negative.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7138
Related to {T77023}. When using many Denoise nodes the memory in OIDN
are allocated up front. A mutex could stall the process until. This
change will allocate the memory after it received the lock.
File output node always received the resolution from the first socket.
When that socket didn't had a link it would use a resolution of 0,0.
What lead to not saving the file at all.
This only effected Multi layer OpenEXR files.
This change would go over all the links to find the first valid
resolution.
In recent refactor the operator sockets were migrated from a std::list to a blender::Vector.
The way how the file output node created the sockets along mapping the sockets could
lead to storing incorrect pointers.
This patch fixes this by defining and mapping the sockets in separate loops.
This also fixes the issue that the width of the "Name" column
when viewing instances does not resize correctly.
Differential Revision: https://developer.blender.org/D10926
- Remove `make_quicky` as on modern systems linking is the main
bottleneck, and there isn't such a gain from partial builds.
- Remove enum generator as `PyC_StringEnumItems` & `EnumPropertyItem`
are used in most places to access enums from Python, otherwise macros
are added via macros.
Previously it was not possible to subdivide the last segment of a cyclic
stroke.
The fix makes sure that the correct number of new points is calculated
correctly and adds the new points to the last segment.
Reviewed By: antoniov
Maniphest Tasks: T87157
Differential Revision: https://developer.blender.org/D10902
Linked override were not properly ignored in some part of the code,
leading to invalid resync results in some cases with recursive overrides
(i.e. overrides of overrides).
Reported by Andy @eyecandy from the studio.
We do not want to copy exiting overrides data from linked ID when
creating its local override (be it either a template, or because linked
ID is itself an override of another lib data).
Note that this was not a very serious issue, would just cause some memory
leak since override data is re-created on newly copied local data
anyway.
These use cases have been very little tested so far, but both complex
production pipeline and future restrictive workflow will make them fairly
common...
Relying on only no-main for that was weak, and inn the end it turns out
we sometimes also need to ifnore override data during copy of Main
data-blocks.
NOTE: The new `LIB_ID_COPY_NO_LIB_OVERRIDE` is also added to the
`LIB_ID_COPY_LOCALIZE` set of flags.
NOTE: The fact that we may now copy liboverrides in some non-main cases
may cause issues in some cases, pretty impossible to track all possible
ones from reading the code... Would not expect too many problem though,
usages of `LIB_ID_CREATE_NO_MAIN` by itself are not so common.
The "orient_matrix" parameter was only used when contraint was defined.
This change will hardly be noticed by the user, but it can be useful for
addon developers.
Possibly resolves T85734.
Tell `FindOSL.cmake` where to find the shader header files. These have
moved from `${LIBDIR}/osl/shaders` to `${LIBDIR}/osl/share/OSL/shaders`
when OSL was upgraded (T85365).
Compostior relies heavilly on overridden methods. The override keyword has been added
in this module and is desired. The benefit of the override keyword is
that it reports an error when not applied to a (base) virtual method and report
what will not match when refactoring the code to be closer to blender style guide.
Reviewed By: sybren
Differential Revision: https://developer.blender.org/D10846
A `DataSource` provides columns for the spreadsheet to display.
Every column has a SpreadsheetColumnID as identifier. Columns
are not generated eagerly anymore, instead the main spreadsheet
code can request a column from a data source with an column
identifier. The column identifiers can be stored in DNA and allow us
to store persistent data per column.
On the user level the only thing that changes is that columns are
not shown in alphabetical order anymore. Instead, new columns
are always added on the left. The behavior can be changed,
however I'd prefer not to automate this too much currently. I think
we should just add operators to hide/reorder/resize columns soonish.
Differential Revision: https://developer.blender.org/D10901
Some object types don't have a geometry component in the depsgraph.
Before, there always was a warning printed when such an object was
used in the object info node (e.g. to get its location).
Points are drawn as half octahedron (aligned to the camera).
Getting the appropriate matrix for facing the camera would fail in in
orthographic view, points were not facing the camera (revealing their
missing other half octahedron)
Maniphest Tasks: T87150
Differential Revision: https://developer.blender.org/D10923
This is a minor change to add some plumbing code
to support custom geo nodes. This is working the
same way as the custom cycles and compositor nodes.
An example add-in is attached to D10784
Reviewed By: JacquesLucke
Differential Revision: http://developer.blender.org/D10784
Where possible, nodes in the "Geometry" category should support all
geometry component types. This adds support for volumes in the
recently added bounding box node, based on functions added in the
previous two commits.
Differential Revision: https://developer.blender.org/D10906
This commit splits of the function that generates a bounding box for a
volume into a new function, so that the min and max coordinate can
be retrieved from volume data without an object. Also some cleanup:
using the float3 type.
Often you need to apply a transformation to a grid without changing the
original, and it's necessary to avoid a deep copy of the actual data.
OpenVDB has a function to do this, this commit simply adds a wrapper
to transform and use that function with blender's `float4x4` data type.
Split from D10906
Previously only attributes of "real" geometry were displayed in
attribute search. This commit adds code to look through attributes
on instances and add those to the search drop-down too.
This required implementing the same sort of recursive traversal as
the realize instances code. The situation is a bit different though,
this can return early and doesn't need to keep track of transforms.
I added a limit so that it doesn't look through the attributes of
too many instanced geometry sets. I think this is important, since
this isn't a trivial operation and it could potentially happen for
every node in a large node tree. Currently the limit is set at 8
geometry sets, which I expect will be enough, since the set of
attributes is mostly not very unique anyway.
Fixes T86282
Diffrential Revision: https://developer.blender.org/D10919
The problem was that you could getting write access to a grid from a
`const Volume *` without breaking const correctness. I encountered this
when working on support for volumes in the bounding box node. For
geometry nodes there is an important distinction between getting data
"for read" and "for write", with the former returning a `const` version
of the data.
Also, for volumes it was necessary to cast away const, since all of
the relevant functions in `volume.cc` didn't have const versions. This
patch adds `const` in these places, distinguising between "for read"
and "for write" versions of functions where necessary.
The downside is that loading and unloading in the global volume cache
needs const write-access to some member variables. I see that as an
inherent problem that comes up with caching that never has a beautiful
solution anyway.
Some of the const-ness could probably be propogated futher in EEVEE
code, but I'll leave that out, since there is another level of caching.
Differential Revision: https://developer.blender.org/D10916
Fix `nullptr` redeference when setting 'orig_data' pointers on CoW copies,
by stopping the loop also when `element_cow == nullptr`. This avoids a
crash of Blender when the original list of pointers is longer than the
CoW list of pointers.
I've also added a `BLI_assert()` that checks for equal lengths of the
two `ListBase`s, so that problems like these aren't hidden away completely.
The root cause of the crash was actually a modifier that was assigned to
an object of the wrong type (an Armature object doesn't support modifiers).
This caused the list of modifiers on the CoW copy to be shorter than the
list of modifiers on the original Object. It's still a mystery how that
object got that modifier in the first place.
Broken by 04e1feb830
Caused by accidentaly moving `SEQ_proxy_rebuild_finish` outside of
`if (nseq->type == SEQ_TYPE_MOVIE)` condition in function
`SEQ_proxy_rebuild_context()`.
To check if an "is_mesh_verts_only" mesh, the overlay engine checks if the
mesh has no "totedge" and has "totvert".
However, sometimes this engine can check the wrong mesh since editmesh
works on `embm->mesh_eval_final`.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D10917
Previously, the spreadsheet editor could only show data of the original
and of the final evaluated object. Now it is possible to show the data
at some intermediate stages too.
For that the mode has to be set to "Node" in the spreadsheet editor.
Furthermore, the preview of a specific node has to be activated by
clicking the new icon in the header of geometry nodes.
The exact ui of this feature might be refined in upcoming commits.
It is already very useful for debugging node groups in it's current
state though.
Differential Revision: https://developer.blender.org/D10875
This adds the ability to get a selection list for both edit mode bones
and pose mode bones.
To do this the selection menu list logic had to be reworked a bit.
Before it only stored the names of objects. This might work will of
objects, however as stated in the code, it might fail for linked objects
(so multiple object can have the same name in some corner cases).
For bones it is a very common occurance where you can have multiple
armature that has the same bone names. So now it also stores the object
and bone pointers for this case.
Reviewed By: Sybren
Differential Revision: http://developer.blender.org/D10570
Fix T85796
The issue was that where_on_path uses a resampled curve to get the data
from the curve. This leads to disconnects between the curve the user
sees and the evaluated location data.
To fix this we simply use the actual curve data the user can see.
The older code needed a cleanup either way as there were hacks in other
parts of the code trying to work around some brokenness. This is now
fixed and we no longer need to clamp the evaluation range to 0-1 or make
helper functions to make it do what we actually want.
Reviewed By: Campbell, Sybren
Differential Revision: http://developer.blender.org/D10898
While the existing behavior worked as intended,
it wasn't possible to have two views on the same file at different
locations.
Since there isn't much use in having two views open at the same location
allow one view to be at a different scroll location.
UI edit-source and selecting a text data block now need explicit calls
to scroll to the cursor location.
Resolves T87284
Without legacy-undo, loading memfile undo only cleared edit-mode data
for mesh and armature object types.
This causes an assertion when loading edit-mode undo steps afterwards
for other object types as the existence of this data made entering
object mode fail.
Resolve this by freeing all objects types undo data from ED_editors_exit
Before this change, object-data could only have
one of the objects referencing it in edit-mode.
Reverting since multiple meshes in edit-mode (for the same object data)
isn't thread-safe as the evaluated edit-meshes are created in the
original edit-mesh, causing a crash updating the depsgraph
for linked duplicates, see: T86767.
While we could support this case, it's a bigger project without
significant benefits, so reinstate the limitation of only
allowing a single object to be in edit-mode for each object data.
This adds back the error from T85974 which caused an assertion
but didn't crash in release builds.
This reverts commit 2b60d7d09c.
This commit essentially touches to post-processing of collections and
objects after resync itself has been done, to ensure their proper
instantiation in the scene:
- Remove a lot of the process in resync case (resynced data are assumed
to be already instantiated in the scene, unlike override creation
case).
- For auto-resync, only do post-processing once after all overrides
have been resynced (doing it after each individual resynced was
causing a lot of instantiation glitches, with a lot of unwanted
extra objects and collections being added to the master collection).
It also deals in a much more reliable way with detection of objects
missing from the scene, by using the new `BKE_scene_objects_as_gset`
utils.
As a bonus this makes auto-resync process slightly faster (only by a few
percents, but that's always good to get).
This is internaly using the code of `BKE_scene_objects_iterator` and
steals its gset. More efficient than using that iterator directly to
rebuild another GSet...
In folded view, some type of data are listed as one icon per item,
others are 'compacted' as a single icon with a counter.
This commit adds bones (edit, normal and pose ones), pose groups and
vertex groups as 'compacted' ones in folded view.
Part of D10855.
This is a minimal, information-only view currently, listing by default
all the override data-blocks, with their user-edited override
properties.
System-generated overrides (like the overrides of pointers to other
override data-blocks) can be shown through a filter option.
Finally, potential info or warning messages from (auto-)resync propcess
are also shown, as an icon + tooltip next to the affected items.
Part of D10855.
This flag is set for liboverride IDs that are detected as no longer
needed by resync process, while having been user-edited, so
auto-handling code cannot silently delete them.
Exposing those to users will be part of the new incoming Override
Outliner view.
Part of D10855.
This removes a lot of unnecessary code that is generated by
the compiler automatically.
In very few cases, a defaulted destructor in a .cc file is
still necessary, because of forward declarations in the header.
I removed some defaulted virtual destructors, because they are not
necessary, when the parent class has a virtual destructor already.
Defaulted constructors are only necessary when there is another
constructor, but the class should still be default constructible.
Differential Revision: https://developer.blender.org/D10911
This code is incompatible with .blend files from subversion 16 and 17.
The ideal would be to create a new subversion when landed rBf674976edd88.
But for now, due to the delay, moving the code to the previous subversion
can solve it.
The function name was not very specific, this makes it clearer that it
works on instances rather than only real geometry. Also use `r_`
prefix for the return argument.
During review of D10730 it was discovered that the "uv" name causes
issues for cycles, which uses it as a default internal data name. While
that could be fixed in the future, there was no particular reason to use
"uv" instead of "uv_map", so we use the latter instead here, which
is consistent with the lowercase naming scheme chosen for attributes.
Some of the BMesh primitive operators have a lot of overhead due to the
fact that they use other operators like extrusion, removing doubles,
and rotation, to build each shape rather than filling arrays directly.
Implementing the primitives directly with the Mesh data structure is
much more efficient, since it's simple to fill the result data based
on the known inputs. It also allows also skip the conversion from BMesh
to Mesh after the calculations.
Speed matters more for procedural operations than for the existing
operators accessible from the add menu, since they will be executed
every evaluation rather than once before a destructive workflow.
| Shape | Before (ms) | After (ms) | Speedup (x times faster) |
| -------- | -------------| ------------| -------------------------|
| Cylinder | 1.1676 | 0.31327 | 3.72 |
| Cone | 4.5890 | 0.17762 | 25.8 |
| Sphere | 64213.3 | 13.595 | 4720 |
The downside of this change is that there will be two implementations
for these three primitives, in these nodes and in `bmo_primitive.c`.
One option would be re-implementing the BMesh primitives in terms of
this code, but that would require `BMesh` to depend on `Mesh`, which
is currently avoided. On the other hand, it will be easier to add new
features to these nodes like different fill types for the top and the
bottom of a cylinder, rings for a cylinder, and tagging the output with
boolean attributes. Another factor to consider is that the add mesh
object operator could be implemented with these nodes, just providing
more benefit for a faster implementation.
As a future cleanup, there is room to share code that generates the
"rings" topology between different nodes that generate meshes.
Differential Revision: https://developer.blender.org/D10730
When having a checkbox and a value both in one row together with an
animation decorator it is questionable whether the decorator should act
on animating the checkbox or the corresponding value. At the moment,
usage in modifiers does not seem to be very consistent:
Here the animation decorator works for animating the checkbox:
- `Build` (greasepencil) > `Factor` (this was reported and is changed in
this patch to act on the value instead of the checkbox)
- `DataTransfer` > `Topology Mapping` > `Max Distance` (this I guess
should also act on the value instead of the checkbox)
- `Edge Split` > `Edge Angle` (questionable)
- `Mirror` > `Merge` (questionable)
- `Screw` > `Merge` (questionable)
- `Wireframe` > `Crease Edges` (questionable)
Here the animation decorator works for animating the value:
- `VertexWeightEdit` > `Group Add/Remove Threshold`
- `Decimate` > `Symmetry`
So in this patch only the behavior in the greasepencil Build modifier UI
is changed, since I think it is quite obvious that one would more often
use the decorator for animating the factor value than for animating the
checkbox.
Maniphest Tasks: T87263
Differential Revision: https://developer.blender.org/D10910
This was added in 37b82a2d26, doesn't have much purpose and doesn't even do
anything in the vast majority of cases. It only affects custom properties. So
at the very least it shouldn't be shown for regular RNA property buttons. But
even for cases where it may do something, we couldn't find a good use-case. If
this operator has use-cases with some add-ons, the add-ons can expose it
differently, e.g. with a little 'x' icon next to the button, like it's done in
the keymap editor.
So removing the context menu entry now. Should this turn out to be a problem,
it can be brought back but much more selectively (only where it actually does
something). Or we could combine it with "Reset to Default".
Differential Revision: https://developer.blender.org/D8727
Reviewed by: Brecht Van Lommel, Hans Goudey
This patch fixes the issue in T86463 that colored collection icons in
the collection search box don't scale according to the zoom level.
The issue was caused by the drawing function for the colored
collection icons not allowing to change the size of the icons.
This patch addresses that by scaling the icon based on the drawing
width.
Reviewed By: #user_interface, natecraddock, Severin
Differential Revision: https://developer.blender.org/D10708
This is caused by rB9f323e9bf79f. When hover shortcuts are used,
we cannot use the modifier from the context (which will be the
active modifier), or they won't be hover shortcuts anymore!
There didn't use to be an editor-level "modifier" context variable,
which is why this worked before.
The fix is simple, just specify the modifier name for this particular
remove button.
Differential Revision: https://developer.blender.org/D10870
The default insert link callback for nodes was trying to move the
existing link away, since it didn't properly handle multi-input sockets
(though the problem wasn't exposed for the join node). We can basically
skip all of this "moving existing links" for multi-input sockets, unless
we are at the link limit.
This commit adds a simple node to output the min and max of an
axis-aligned bounding box for the input geometry, as well a rectangular
prism mesh created from these values for convenience.
The initial use case for this node is a "bounding box boolean", where
doing the boolean with just a bounding box could be signigicantly
faster, for cases like cutting a hole in a wall for a window. But it's
easy to imagine other cases where it could be useful.
This node supports mesh and point cloud data right now, volume support
will come as a separate patch. Also note that there is plenty of room
to improve the performance of this node through parallelization.
Differential Revision: https://developer.blender.org/D10420
Caused by typo in {rB76689e851700}.
Since Palettes and bGPDpalette are not the same size, this would not
only cause a crash in versioning code, but could only go downhill from
here on.
Maniphest Tasks: T87236
Differential Revision: https://developer.blender.org/D10903
The `InputAngle` function uses a flawed algorithm to fix precision issues.
This commit refactor the logic of that function by using BLI utilities.
Differential Revision: https://developer.blender.org/D10880
For debugging purposes to convert the internal state of the
NodeOperationBuilder to a graphviz.
Usage:
std::cout << *this << "\n";
Inside any method of the NodeOperationBuilder.
For curves, we need to tag the curve ID_RECALC_COPY_ON_WRITE for batch
cache update (same as in {rB24b2fe50f3ec}).
Maniphest Tasks: T86932
Differential Revision: https://developer.blender.org/D10826
Tweaking e.g. a field strength would then not use the curve/surface
normal anymore [but the object center instead].
If a curve has a forcefield with effector shape Curve (in code its shape
is PFIELD_SHAPE_SURFACE then), it wil get a SurfaceModifier.
Changing properties will free the SurfaceModifierData's bvhtree and mesh
And these dont get copied along when doing the CoW copy, these are
explicitly set to NULL. So this was also failing for meshes, not just
curves.
Without the mesh & bvhtree though, get_effector_data() will not set the
EffectorData's normal correctly (it is closest_point_on_surface() which
does this). And without the right EffectorData's normal, the effector
will of course work unexpected.
Going in and out of editmode made this work because that goes down this
route:
- BKE_object_handle_data_update
- BKE_displist_make_curveTypes
- do_makeDispListCurveTypes
- curve_calc_modifiers_post
-- BKE_mesh_new_nomain_from_curve_displist
-- we then have our desired updated mesh from the curve
-- this will also call the SurfaceModifiers deformVerts [which - given we
have a valid mesh - will update the bvhtree properly]
Also note that _animating_ the effector actually works, (have not done
the deep dive why this works, assume the curve geometry is updated in
this case)
So, now just carefully tag the curve ID_RECALC_GEOMETRY in
rna_FieldSettings_update for this specific case.
Maniphest Tasks: T84623
Differential Revision: https://developer.blender.org/D10092
Caused by rB0f95f51361d7.
Similar to T85515, T84717, T85543 and their related fixes.
In this case, add missing shading notifier as in rB9274bd457a25.
Reviewed By: brecht, lichtwerk
Maniphest Tasks: T86175
Differential Revision: https://developer.blender.org/D10604
Selecting random percentage would not reliably guarantee the precise
percentage of selected elements now randomize an index array instead
and use the precise number of elements from this array.
Note that this change has only been made to edit-mesh
and should be applied to all random selection operators.
Reading Area.type & ui_type printed an error from `pyrna_enum_to_py`
and returned an empty string.
As empty is a valid value, include it in the property enum list.
For Cycles, when enabling the Persistent Data option, the full render data
will be preserved from frame-to-frame in animation renders and between
re-renders of the scene. This means that any modifier evaluation, BVH
building, OpenGL vertex buffer uploads, etc, can be done only once for
unchanged objects. This comes at an increased memory cost.
Previously there option was named Persistent Images and had a more limited
impact on render time and memory.
When using multiple view layers, only data from a single view layer is
preserved to keep memory usage somewhat under control. However objects
shared between view layers are preserved, and so this can speedup such
renders as well, even single frame renders.
For Eevee and Workbench this option is not available, however these engines
will now always reuse the depsgraph for animation and multiple view layers.
This can significantly speed up rendering.
These engines do not support sharing the depsgraph between re-renders, due
to technical issues regarding OpenGL contexts. Support for this could be added
if those are solved, see the code comments for details.
* Move out of Simplify panel, into Light Paths > Fast Global Illumination
* Add separate boolan setting to enable/disable it separate from Simplify
* Default AO bounces to 1
* Put ambient occlusion distance in this panel as well
Since you can already specify a separate size for X and Y with the
grid node, it makes sense to be able to specify the size separately
for each axis also.
This also avoids some awkward math with a Transform node afterwards
when you want a specific size for each direction.
Versioning (except for animation and drivers) is handled in this commit.
Differential Revision: https://developer.blender.org/D10834
Materials used in grease pencil modifiers have the requirement that they
are already used on the object. In the UI dropdown, this restriction is
ensured by calling uiItemPointerR with appropriate searchptr and
searchpropname, so only giving the user the choice of materials already
used on the object.
From python though, it was still possible to assign materials outside of
this this restriction. This led to reports like T86981 [which have been
partially solved by clamping the material index in the modifier code to
be in the valid range].
Now make sure we dont assign "invalid" materials through RNA by
appropriate RNA pointer functions.
This also adds a proper warning (red, alert) in case of the LineArt
modifier if such a invalid material is still in the file [same as other
modifiers already do].
Differential Revision: https://developer.blender.org/D10873
This resulted in bad motion blur from alembic animation when imported
after rB83980506957c (because it was now OFF by default).
When imported in a version prior to rB83980506957c and saved, this was
still working fine.
Vertex interpolation option introduced in rBbab2260b59c7 was defaulting
to ON (correct) in that commit, but was lost in rB83980506957c.
Now switch this to ON by default again.
Note: if you have a file that was (incorrectly) imported with a version
after rB83980506957c you'll have to enable Vertex Interpolation option
on every MeshSequenceCache modifier by hand.
Maniphest Tasks: T86654
Differential Revision: https://developer.blender.org/D10876
Since these nodes are usually used for more basic operations and the
attribute nodes are used when more complexity is necessary, it makes
sense to give these nodes more accessible defaults-- hopefully this can
make learning about the core concepts of geometry nodes a bit easier.
Before this change, you could have the new sculpt symmetry code and the
older weight paint symmetry code active at the same time. This would
lead to users easily trashing their weigh paint data if they were not
careful when switching between modes.
Now the specific weight paint symmetry code is an exclusive toggle so
the user can't accidentally mirror strokes and vertex groups at the same
time. This also paves the way of supporting Y and Z symmetry in the
future for weight groups mirroring if we decide to add it in the future.
Reviewed By: Sybren
Differential Revision: http://developer.blender.org/D10426
The following primitives needed ME_EDGEDRAW, ME_EDGERENDER edge flags:
* Grid
* Circle
Set flags on the inside edges for grid and circle triangle fan (mesh primitive nodes)
so they are visible and selectable.
Reviewed By: HooglyBoogly
Differential Revision: https://developer.blender.org/D10878
The main goal here is to remove the need for a C API to the code in
`mesh_boolean_convert.cc`. This is achieved by moving `MOD_boolean.c`
to C++ and making the necessary changes for it to compile. On top of
that there are some other slight simplifications possible to the
direct mesh boolean code: it doesn't need to copy the material
remaps, and the modifier code can use some other C++ types directly.
The old path that didn't skip the conversion to and from BMesh for
the exact solver was not yet removed from this file. At this point no
problems have came up in the new implementation, so it's safe
to remove it.
This commit improves the performance of the node by up to 40% in some
cases when there are only two input meshes, mainly by skipping the
conversion to and from BMesh.
When there are more than two input meshes (note the distinction from
"Geometries", a geometry set can have many mesh instances), the
performance is actually worse, since boolean currently always does
self intersection in that case. Theoretically this could be improved
in the boolean code, or another option is automatically realizing
instances for each input geometry set.
Another improvement is using multi-input sockets for the inputs, which
removes the need to have a separate boolean node for every operation,
which can hopefully simplify some node trees.
The changes necessary for transforms in `mesh_boolean_convert.cc` are
somewhat subtle; they come from the fact that the collecting the
geometry set instances already gives transforms in the local space
of the modifier object. There is also a very small amount of cleanup
to those lines, using `float4x4::identity()`.
This commit also fixes T87078, where overlapping difference meshes
makes the operation not work, though I haven't investigated why.
Differential Revision: https://developer.blender.org/D10599
Handle Lattice object the same way as Mesh objects. This is mostly to
execute the `object->data = data_eval;` line, which ensures that the
evaluated mesh is assigned to the evaluated object, and thus prevents
the lattice from un-deforming.
This fixes T86440
As the CU_2D flag is set for nurbs, a Curve can have 2D nurbs mixed with 3D.
But the UI does not allow this mixing. It updates all nurbs to 2D or 3D when set.
So remove this specific flag for nurbs.
This may break old files, since 2D curves with mixed 3D are now set as 3D.
Differential Revision: https://developer.blender.org/D10738
Previously, `node_name` was rarely actually a name of a node. It is set
correctly as node name in `ED_node_tree_push`. However, later on it
was overwritten by the name of an id data block in `node_draw_space`.
Now, the node_name stays the name of the "parent" node. Whereas
display_name is the name that will be displayed in the breadcrumbs.
With this change, the `node_name` can be used to reconstruct the
actual path from the root node tree to the currently visible tree.
Differential Revision: https://developer.blender.org/D10874
The node tree evaluator now calls a callback for every used socket with
its corresponding value(s). Right now the callback does nothing.
However, we can use it to collect attribute name hints, socket values
for debugging or data that will be displayed in the spreadsheet.
The main difficulty here was to also call the callback for sockets in
nodes that are not directly executed (such as group nodes, muted
nodes and reroutes).
No functional changes are expected.
After some back and forth with the GP module and some artists, this
option was deemed not that useful. The use case was considered too
obscure so we'll remove it.
It is still posible to have this functionality by using the vertex
weight modiifers or manually clamping the weights.
If any axis of the scale of an object was zero, transforming failed.
This was because `td->smtx` was set to a zero matrix.
orthogonalize_m3_zero_axes is used to fill in the zeroed axes.
Thanks to @filedescriptor for the initial fix: D10869.
Expose a this function to initialize any zeroed axes
to an orthogonal vector based on other non-zeroed axes.
This functionality already existed for `invert_m#_m#_safe_ortho`,
expose as a public function as it's useful to be able to fill in zeroed
axes of transformation matrices since they may be used in matrix
multiplication which would create degenerate matrices.
Compute a subset of the area light that actually affects the shading point
and only samples points within that.
It's not perfect as the real subset is a circle instead of a rectangle, and
the attenuation is not accounted for. However it massively reduces noise for
shading points near the area light anyway.
Ellipse shaped area lights do not have this importance sampling, but do not
have solid angle importance sampling either.
Ref D10594
This simulates the effect of a honeycomb or grid placed in front of a softbox.
In practice, it works by attenuating rays coming off-angle as a function of the
provided spread angle parameter.
Setting the parameter to 180 degrees poses no restrictions to the rays, making
the light behave the same way as before this patch.
The total light power is normalized based on the spread angle, so that the
light strength remains the same.
Differential Revision: https://developer.blender.org/D10594
The combination of building unit tests and WITH_CYCLES_NATIVE_ONLY did not
correctly detect when AVX/AVX2 support is available.
Differential Revision: https://developer.blender.org/D8201
Fix a segfault by setting the `batch_cache` pointer to `NULL` when copying
a Lattice. That way the copy can get its own batch cache when needed,
preventing a use-after-free.
This condition was in contradiction with comment for function
`ffmpeg_generic_seek_workaround()`.
I have noticed, that formats that seeked well used this workaround.
Problem was that I misunderstood code from `av_seek_frame()` - formats
with `read_seek()` function stil don't use generic seeking method.
This is defined in `seek_frame_internal()`
The "Paste Flipped" operator was missing its own tooltip.
This patch adds the tooltip by implementing the `get_description` callback for
`GRAPH_OT_paste` and `ACTION_OT_paste`.
Reviewed By: sybren
Maniphest Tasks: T87037
Differential Revision: https://developer.blender.org/D10859
`ffmpeg_generic_seek_workaround()` applied negative offset for seqrched
packet timestamp, but proxies always start from 0 and timestamp can be
negative.
Limit timestamp value to 0, because `av_seek_frame()` doesn't accept
negative timestamps and returns with error. This prevents seeking from
working correctly.
The RGN_TYPE_EXECUTE region was zero height - with the "Load & Save"
button drawing outside of those bounds - so it didn't respond to events.
This is because the region started off this size and does not change
with simply adding the buttons. Although it has RGN_FLAG_DYNAMIC_SIZE,
delayed reinit of regions is only currently supported on headers.
This gives the execute region an initial (minimum) vertical size but
also makes the region **hidden** by default.
- Showing Prefs as an editor among others it will show the header
but not the execute region.
- Showing the Prefs in a popup window, hides the header region
and shows the execute region.
Ref D10636
Make use of `StatementMacros` so Python structs are formatted correctly.
The comment about STRINGIFY doesn't seem to be valid anymore,
so this has been left out.
This was added in rB93aeb6b318a7, last changed in rB8f0a44a5d55d,
and removed in rB51b796ff1528. I assume it was a way for curves to
have split edges at corners. As far as I can tell it only ever worked
in Blender Internal, and that has been gone for years.
Another possible route here would be restoring this functionality,
but it's generally preferrable to reduce complexity in this
area of curve code than adding it back, especially in the context
of other improvements planned related to curves in geometry nodes.
Differential Revision: https://developer.blender.org/D9966
The index_data_map was not cleared when clearing a cache, so this would just append
the new correct data to the end of the array instead of overwriting it, which would
cause us to then use outdated indices.
For other "Attribute Name" fields we usually give a more specific name
that relates to what the attribute is actually used for, like "Mask" in
the point separate node. This is a similar situation, and would also
be consistent with the naming planned in D10506.
The geometry nodes modifier did not specify that it needs all custom data layers.
Therefore the modifier evaluation code tagged some layers so that they will not be
copied later on by calling `mesh_set_only_copy` in `mesh_calc_modifiers`.
When boolean's Object also has a modifier that depends back on
the target Object, a crash occurred.
In a case like this, BKE_modifier_get_evaluated_mesh_from_evaluated_object
returns NULL, so just have to protect against that case.
- Don't use `r_` prefix for an argument that's also read from
(this is meant for return only arguments).
- Rename variables to make it clear the distance is pixels,
using manhattan length measurement.
The new Cryptomatte workflow assumes that the picker is used in the new
workflow. The legacy node wasn't working as it would never find a
correct cryptomatte layer. This fix will use the color sampling like we
used to do.
WorkPackages struct was created when scheduled. This patch keeps the
WorkPackages around and stores additional data with the workpackages.
The speedup is to small to notice, but it is needed as preparation
to introduce a faster scheduling method.
Regression in 80cbbd2843.
Unfortunately keeping selection picking behavior as well as
supporting finding the nearest face within a radius requires an
inconsistency between x-ray and back-buffer selection that
doesn't work well using the current arguments.
Resolve by adding an argument that causes the input distance
to be ignored for back-buffer selection.
This is used by selection picking but not the knife tool.
This changes behavior for path-selection in face mode,
which now uses a margin for back-buffer selection.
From my own testing this doesn't seem to be a problem like it could be
for regular selection picking.
The tooltip while dragging a collection in Scene mode in the Outliner
was always "Link inside Collection" even if the action performed was
different. This was because the `collection_drop_init` set the
`from_collection` always to `NULL` if the Outliner display mode was
currently set to Scene.
Commit that introduced this issue:
rB0f54c3a9b75be8f8db9022fb0aeb0f8d0d4f0299
The fix removes the check of the display mode and only sets the
`from_collection` to `NULL` if the ctrl (linking) key is held.
Reviewed By: JacquesLucke
Maniphest Tasks: T86947
Differential Revision: https://developer.blender.org/D10864
When using the smooth operator, the thickness would not be
smoothed on the first iteration. This was becasue the inner loop
was set to run `r * 20` times but `r` starts at 0.
The fix makes sure that "Smooth Thickness" works on the first
iteration by using a fixed value of `20` as the loop limit. This
makes sure that for every iteration, we run the smoothing of the
thickness 20 more times.
Reviewed By: antoniov
Maniphest Tasks: T87082
Differential Revision: https://developer.blender.org/D10867
Use LISTBASE_FOREACH instead.
SEQ_CURRENT_BEGIN did null checks, so now these must be explicit.
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D10823
Files that only contain a doc-string still included the last blank line,
since this normally contains code examples.
There are some cases where only a docstring exists
which made sphinx report warnings.
Use utility functions to decompose data paths and resolve the
RNA property from a data-path.
Replaces in-line string manipulation and RNA access.
This allows more complex data paths to be used, where previously string
literals in a data path could break the simple data-path handling logic.
Flag check for V3D_LOCK_CAMERA used boolean style assignment to a char,
which worked with this flag but could fail if other flags are added
that use this convention in the future.
- Add static type checks for values so any change to DNA types
will need to be made in the navigation gizmo too.
- Move camera lock check from `rv3d` to `v3d`,
as this isn't stored in the region data.
2D gizmo navigation button that toggles 'Lock Camera to View' while in Camera View.
Differential Revision: https://developer.blender.org/D10835
Reviewed by Campbell Barton
Now `ED_view3d_backbuf_depth_validate`, `ED_view3d_draw_depth` and
`ED_view3d_draw_depth_gpencil` are unified in `ED_view3d_depth_override`.
This new function replaces `ED_view3d_autodist_init`.
Also, since `ED_view3d_depth_update` depends on the render context, and
changing the context is a slow operation, that function also was removed,
and the depth buffer cached is now updated inside the new unified drawing
function when the "bool update_cache" parameter is true.
Finally `V3D_INVALID_BACKBUF` flag has been renamed and moved to
`runtime.flag`.
Differential revision: https://developer.blender.org/D10678
Introduced in rBce462fa1 but harmless since curretly only `StructRNA`
without `owner_id` have the `STRUCT_UNDO` flag cleared.
So this commit does not bring any functional changes but it will be
useful for {D10695}.
The problem was produced because the strokes were selected, but the loop to clear this flag before applying cutter was using unlocked layers only, and the protected layer flag was not reset.
Now, the flag is reset for all layers, locked and unlocked.
It was not possible to determine if a socket was multi input previously
with BPY. This patch exposes the flag as a read-only property of a node
socket. This is important for addons which automatically add connections
between nodes.
Differential Revision: https://developer.blender.org/D10847
Earlier last year, the shortcuts on hover were built as a way to regain
speed lost by removing the "Apply" and "Copy" buttons from the panel.
For the active modifier concept introduced for geometry nodes, the
shortcuts were changed to only affect the active modifier.
Based on feedback, this change slowed down many people's interaction
with the modifier stack so the UI team decided to return hover shortcuts
for modifier panels.
The downside of this change is that it looks like the active modifier is
"selected" and it could be confusing that the modifier shortcuts don't
apply to it. We can explore different ways to display the active status
to address this.
Ref T87012
Generic context operators now look-up the RNA properties to extract
their description (when it's available).
Add `bl_rna_utils.data_path.property_definition_from_data_path()`
to handle the details of accessing the RNA property definition.
When the strength is initially set to zero, the shader graph is
optimized out to remove any node which will not be executed because of
this, which removes pretty much every single node, except for the
output. As the graph is empty, the world shader is made invisible to
rays so it is not evaluated in the kernel.
However, when the strength is then modified, the Background is not
updated as the modification happens on the Shader Node and not on the
Background Node, so it is never tagged as modified.
To fix this, we need to tag the Background as modified when its shader
is also modified so the Kernel data is properly updated.
Regression caused by rBbbe6d44928235cd4a5cfbeaf1a1de78ed861bb92.
Separate tile buffers on all devices only need to exist when denoising is active (so any overlap
being rendered simultaneously does not write to the same memory region).
When denoising is not active they can be distributed like all other memory when peer
memory support is available.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D10858
Display the bone axes at the head (root) of the bone by default, instead
of the tail (tip), and add a slider so that it's possible to adjust this
position.
Versioning code is in place to ensure existing files behave the same
(axes shown at tail), whereas new Armatures will be using the new
default (axes shown at head).
Reviewed By: #animation_rigging, #user_interface, Severin, Sybren
Differential Revision: https://developer.blender.org/D7685
When using `bpy.data.actions[action_name, "nonexistant-library"]`,
use the term `filepath` instead of `name` in the error message.
Also increase the size to match the file path length.
Ref D10253
Remove the `BONE_UNKEYED` flag. It was only written (set/cleared) but
never actually read.
Also remove `framechange_poses_clear_unkeyed()` as its only function was
to clear the `BONE_UNKEYED` flag. It wasn't used anywhere either.
The only code that used the flag was the `extract_pose_from_action()`,
which was removed in 2869ce6cfa (2009).
No functional changes.
Minor changes preparing for snap gizmo inclusion.
- Extract `knife_snap_edge_in_angle` into a utility function.
- Check the snap vertex on closest edge instead of the face.
- Add MODE_INVOKING state.
- Remove unnecessary NULL checks.
- Control 'ignore_edge_snapping' while dragging instead of checking
dragging in `knife_snap_update_from_mval`.
Ref D8220
This enables ASAN support when used with VS 16.9
enable as usual in cmake with the WITH_COMPILER_ASAN
option, or when using make.bat just tag on `asan'
to the invocation, ie: `make lite 2019 asan`
MSVC: Asan support for 16.9
This enables ASAN support when used with VS 16.9
enable as usual in cmake with the WITH_COMPILER_ASAN
option, or when using make.bat just tag on `asan'
to the invocation, ie: `make lite 2019 asan`
Differential Revision: https://developer.blender.org/D7794
Reviewed By: brecht, sergey
`av_seek_frame()` failed to seek to nearest I-frame. This seems to be
a bug or not implemented feature in FFmpeg. Looks like same issue as
ticket https://trac.ffmpeg.org/ticket/1607 on ffmpeg tracker.
If seeking is done using format specific function (`read_seek2`)
field of `AVInputFormat` is set, `see av_seek_frame()`, use
`av_seek_frame()` function. Otherwise use wrapper that actively searches
for I-frame packet.
Searching is flexible and tries to do minimum amount of work. Currently
it is limited to equivalent of 25 frames, which may not be enough for
some files, but there may be files with no I-frames at all, so it is
best to keep this limit as low as possible. Previously this problem was
masked by preseek, which was hard-coded to 25 frames. This was removed
in rB88604b79b7d1.
If this approach would be unnecessary for some formats, in worst case
file would be seeked 2 times which is very fast, so there will be no
visible impact on performance.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D10845
Add fit_method argument to new movie and image RNA API functions.
This argument is optional.
ref T86925
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D10816
When adding images with operator, image file path is split into filename
and directory passed to load function in name and path fields of
SeqLoadData struct. This is because when loading images directory
and filenames are split.
RNA API function passes whole path in path filed.
Apart from loading API inconsistency, this causes initial image loading
to fail, so strip resolution is not set. Also name field of SeqLoadData
should be reserved for strip name.
Let operator code concatenate and split filepath when needed so loading
API can be consistent with RNA API and also between strip types.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D10818
As the Snap Gizmo properties are used as return of snapping values and
therefore updated with each change, use callbacks to get these values
through RNA properties.
Differential Revision: https://developer.blender.org/D8525
Makes it slightly easier to test whether the installed module works
or not. Depends on D10656
Ref T86579
Reviewed By: mont29, campbellbarton
Differential Revision: https://developer.blender.org/D10665
Avoid CTest errors and exit codes due to test failures which depend
on Blender executable.
Ref T86579
Reviewed By: mont29, campbellbarton
Differential Revision: https://developer.blender.org/D10656
Old code's `install` step did not guarantee that all script files
will be copied from source folder to build folder _before_ copying from
build folder to site-packages. That caused incomplete installation.
Fixed by copying scripts etc., only once. Also match the behavior on
Linux: install to site-packages only if `WITH_INSTALL_PORTABLE` is OFF.
Also fix install and launch issues with Xcode.
When interpolate, the stroke order was not correct because it was assumed the GHash iter would return the items in the same order of insertion, but this is false.
Now, a list is used to keep the order of the strokes and the Hash is used to get the relationship between strokes as before.
The file was not really corrupted (as in, Blender did everything
correctly while saving). I only did not consider the case when
a .blend file is resaved in an older version before.
The issue was caused by handles not being written to the new mask
spline shape: it was always written as (0, 0), which was making the
handle calculation go wrong later on.
Solved by allocating a temporary array of bezier points and calculating
handles for them. While this is an extra array allocation it is only
done for a small amount of points and it is not in the critical code
path. Having this as an extra array helps to overcome a limitation of
the current mask shape API.
Additionally, disable the interpolation for the shape change:
the spline is brand new, there is nothing to be interpolated there.
When reading metadata from image files the metadata is trimmed to 1024.
For cryptomatte the metadata can contain json data and should not be
trimmed. Resulting additional checks in the manifest parser for
incomplete json data.
You could argue to add an exception for cryptomatte, but that would
still allows misuse. When the direction of this patch is accepted we
should consider removing `maxlen` from `IDP_AssignString` as it
doesn't seem to be used anywhere else.
Reviewed By: #images_movies, mont29, sergey
Differential Revision: https://developer.blender.org/D10825
Since a few releases it is possible to process Blenders images much
more effficiently than before thanks to the `foreach_get`/`foreach_set`
methods. This new template shows how those methods can be used together
with numpy.
Differential Revision: https://developer.blender.org/D9400
Initially this is a typo when solving merge conflict back in 2017
in commit rBf4140f2c8138.
The code was so for 3 years, so it is safer to assume that this is
the new expected design. The affected codepath only comes from the
BKE_object_modifier_update_subframe() which is discouraged from use
and have other issues.
Eventually those offending codepaths will be removed completely.
The grade-schema for now is more localized: remove ID_RECALC_ALL
which seems to be misused and is causing issues like the ones which
are described and discussed in T81340.
On user level there should be no measurable changes. In fact, no
functional changes at all are expected to happen.
Reviewed By: sybren
Differential Revision: https://developer.blender.org/D9220
This makes regular development more pleasant, because one does not have
to fix unrelated clang tidy mistakes when one is in the middle of something.
Before this change, I would usually turn clang-tidy off entirely, but then
forget to turn it on again.
This change has been agreed on by Sergey as well.
Fix the data management bug where evaluation of lattice objects would
write back to the CoW copy of the Lattice ID, even when that copy was
shared between objects.
Each lattice object evaluation now stores its own evaluated data copy
via `BKE_object_eval_assign_data()`.
Reviewed By: sergey
Maniphest Tasks: T78650
Differential Revision: https://developer.blender.org/D10790
Previously, the code expected the id property to have the `IDP_FLOAT` type.
However, when assigning a Python float (which is a double internally)
to an id property, it would change the type to `IDP_DOUBLE`.
The fix is to allow both types in the geometry nodes modifier.
UI hints should only be modified when the depsgraph is active.
Otherwise two threads evaluating the same object in different depsgraphs
can conflict with each other.
This is functionality that isn't accessible via the user interface. The
API allows the creation and modification of an override template that
holds rules that needs to be checked when overriding the asset.
The API is setup that it cannot be changed after creation. Later on when
the system is more mature we will allow changing overrides operations.
NOTE: This is an experimental feature and should not be used in productions.
Reviewed By: mont29, sebbas
Differential Revision: https://developer.blender.org/D10792
Respect the distance argument to EDBM_face_find_nearest,
when zero, sample a single pixel, otherwise sample a region.
Knife uses the selection-buffer to pick a face when the ray-cast failed.
This was meant to allow snapping to nearby faces however as the margin
was ignored, it was only used in edge cases where the ray-cast missed
but the pixel didn't.
Now the face-picking threshold is working as expected.
Note that other callers to EDBM_face_find_nearest have been updated
so set their distance argument to zero so this only impacts the knife.
Regular selection and path select could be modified separately if users
prefer this behavior.
Resolve logical error in edbm_shortest_path_pick_invoke
where any discrepancy between EDBM_unified_findnearest and
edbm_elem_find_nearest caused the active-object to be cleared.
While it's not a problem at the moment, using a larger threshold
for path picking exposes the error.
In this case, the cage location was left zeroed which was then
projected back onto the screen to find the nearest screen space
edge/vertex.
This made snapping to the vertex/edge fail in some corner-cases
where it was intended to work.
This is an implementation of Enhanced Subpixel Morphological Antialiasing (SMAA)
The algorithm was proposed by:
Jorge Jimenez, Jose I. Echevarria, Tiago Sousa, Diego Gutierrez
This node provides only SMAA 1x mode, so the operation will be done with no spatial
multisampling nor temporal supersampling. See Patch for comparisons.
The existing AA operation seems to be used only for binary images by some other nodes.
Using SMAA for binary images needs no important parameter such as "threshold", so we
perhaps can switch the operation to SMAA, though that changes existing behavior.
Notes:
1. The program code assumes the screen coordinates are DirectX style that the
vertical direction is upside-down, so "top" and "bottom" actually represent bottom
and top, respectively.
Thanks for Habib Gahbiche (zazizizou) to polish and finalize this patch.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D2411
Reduce the maximum number of queries to find the face under the
mouse cursor from 6x to 2x on cursor motion.
Calculating the screen-space detail could perform 2x look-ups
which have already been calculated.
Knife snapping logic assumed having a face under the cursor meant
the projected positions were set.
This is not always the case as failure to ray-cast uses the back-buffer
as a fallback.
For projection painting tools besides the `DRAW` tool:
- Don't show the texture from viewport stencil drawing.
- Don't show the texture panel.
Based on D10564 by @lichtwerk with own changes.
Check for indirect ID use after other simple user count checks are made.
Also assert ED_object_base_free_and_unlink_no_indirect_check object
argument isn't indirectly used.
The grease-pencil parent check was enabled when deleting objects,
when previously it was only done when unlinking.
While harmless, the previous logic is correct.
da160dc32d exposed a bug in
`BKE_id_multi_tagged_delete` causing a memory leak in soft-body that
made `physics_softbody` test fail.
Use a loop to remove ID's to keep tests working until T86992 is fixed.
Tracing images to grease pencil objects creates sometimes artifacts, as seen, for example, when tracing the attached image. {F9910821}
I have found the same behavior both in the 2.92 release and in the current 2.93 master.
The artifacts are caused by a variable that's not initialized or updated when finding a point tagged as POTRACE_CORNER.
This patch solves this issue.
Maniphest Tasks: T86967
Differential Revision: https://developer.blender.org/D10832
- comment added in struct `KnifeTool_OpData`
- struct pointer name `lst` updated to `list` ( Guess its more readable )
- `KNF_MODEL_IGNORE_SNAP_ON` updated to `KNF_MODAL_IGNORE_SNAP_ON`
- `KNF_MODEL_IGNORE_SNAP_OFF` updated to `KNF_MODAL_IGNORE_SNAP_OFF`
Ref D10824
The `object_delete_exec` lead to `BKE_library_ID_is_indirectly_used`
being called twice. With this patch deleting is around 20% faster.
Example when deleting 10000 objects:
Current: 35.6s
This patch: 18.8s (updated, last rev 29.7s)
Reviewed By: campbellbarton
Ref D9857
If window is maximized when toggling fullscreen, go back to maximized when done.
Differential Revision: https://developer.blender.org/D10813
Reviewed by Harley Acheson
Changes made:
* Add OpenMP linker flags.
* Copy the libomp.dylib to `2.93/lib/libomp.dylib`.
* Change the `LC_LOAD_DYLIB` item such that
the lib is found at `bpy.so/../../Resources/2.93/lib/libomp.dylib`.
Installation is done by D10664.
Reviewed By: #platform_macos, brecht
Maniphest Tasks: T86579
Differential Revision: https://developer.blender.org/D10657
When the window is already maximized allow covering Windows Taskbar when toggling fullscreen.
Differential Revision: https://developer.blender.org/D10811
Reviewed by Harley Acheson
One can only resync, reset, delete etc. locl overrides, linked ones are
basically regular linked IDs and fully not editable.
Reported by Demeter from the Studio, thanks.
Although "Grid" may not be techincally correct since a grid could be 3D,
it was decided to rename the "Plane" primtive to "Grid". The primitive
node allows subdivisions, so the name is more consistent with the
operator in the 3D view.
Ref T86819
This commit includes a file subversion bump for the versioning.
Changing to a more informative 'Weight Paint' rather than 'WPaint'.
Differential Revision: https://developer.blender.org/D9905
Reviewed by Julian Eisel
Change `float to boolean` and `int32 to boolean` to return false for zero and negative values.
This aligns with how artists would expect these values to work. This is in contrast to what a coder would expect. It was determined on blender.chat that this was a better default. This means that a negative float value would give a boolean false.
Change `Color4f to boolean` to return false for zero and negative grayscale values.
Likewise, for color to boolean, to account for negative value colors, the grayscale value would be used for determining if a colour was false or not.
See {T86454}
Reviewed By: JacquesLucke
Differential Revision: https://developer.blender.org/D10685
Activation distance for multi input sockets had different length
dependend of dpi factor. That caused jumping when activation distance
became shorter than snapping distance.
Fixed by removing activation distance and using same function
which is used to evaluate snapping.
Reviewer: Dalai Felinto
Differential Revision: https://developer.blender.org/D10809
Add a basic OSL shader that shows how inputs
and outputs work and do some simple math with
them.
This template is a happy medium between the
templates we already ship, empty_shader is a
little too bare, and the other templates are
a little "too much" and you end up having to
delete a whole bunch of stuff.
a great starting point for some experimentation!
Differential Revision: https://developer.blender.org/D9933
reviewed by: brecht
The "Enable physics for:" text makes no semantic sense- i.e.
"Enable physics for Fluid". Additionally, the leading text is just not
necessary, this section is just as clear without it.
Differential Revision: https://developer.blender.org/D10537
This patch adds support for the remaining operations of the Vector Math
node within Geometry Nodes. While the operations are already available
in the UI, they hadn't been implemented, yet. With this patch the node
uses the implementation that was added for the Attribute Vector Math
node - similar to how it's handled with the Math node and Attribute
Math node.
Differential Revision: https://developer.blender.org/D10650
Warnings in tooltips were using inconsistent formatting, some in
parantheses, some not, some in caps, others not, some on new lines,
some not, etc.
This patch uses a consistent new line and no capitals for these cases.
Differential Revision: https://developer.blender.org/D9904
Now that we have `ATTR_DOMAIN_AUTO`, it makes sense to use it to skip
automatic domain interpolation. This can make code that depends on the
non-interpolated domain of the attribute a bit simpler.
Added support for compositor tests. Compositor tests can be added, executed and viewed in a similar way to cycles
and other render engines tests.
Running test:
`ctest -R compositor`
Updating test:
`BLENDER_TEST_UPDATE=1 ctest -R compositor`
Viewing test results:
typically saved under `build_folder/tests/compositor/report.html`
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6334
The calculation of the 2D point was wrong when using orthographic mode.
Also small cleanup in float3 variable.
Differential Revision: https://developer.blender.org/D10828
SocketReader was added as an easier to understand interface on top of
the NodeOperation. It was implemented as a base class of the
NodeOperation and adds an additional hierarchy level.
Ths change replaces the abstract class with a typedef. In the end we
want to remove the typedef but will wait for some new nodes before doing
so.
Generalize threading settings in proxy building and use them for encoding
and decoding in general. Check codec capabilities, prefer FF_THREAD_FRAME
threading over FF_THREAD_SLICE and automatic thread count over setting it
explicitly.
ffmpeg-codecs man page suggests that threads option is global and used by
codecs, that supports this option. Form some tests I have done, it seems that
`av_dict_set_int(&codec_opts, "threads", BLI_system_thread_count(), 0)`
has same effect as
```
pCodecCtx->thread_count = BLI_system_thread_count();
pCodecCtx->thread_type = FF_THREAD_FRAME;
```
Looking at `ff_frame_thread_encoder_init()` code, these cases are not
equivalent. It is probably safer to leave threading setup on libavcodec than
setting up each codec threading individually.
From what I have read all over the internet, frame multithreading should be
faster than slice multithreading. Slice multithreading is mainly used for low
latency streaming.
When running Blender with --debug-ffmpeg it complains about
`pCodecCtx->thread_count = BLI_system_thread_count()` that using thread count
above 16 is not recommended. Using too many threads can negatively affect image
quality, but I am not sure if this is the case for decoding as well - see
https://streaminglearningcenter.com/blogs/ffmpeg-command-threads-how-it-affects-quality-and-performance.html
This is fine for proxies but may be undesirable for final renders.
Number of threads is limited by image size, because of size of motion vectors,
so if it is possible let libavcodec determine optimal thread count.
Performance difference:
Proxy building: None
Playback speed: 2x better on 1920x1080 sample h264 file
Scrubbing: Hard to quantify, but it's much more responsive
Rendering speed: None on 1920x1080 sample h264 file, there is improvement with codecs that do support FF_THREAD_FRAME for encoding like MPNG
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D10791
Split seeking section of `ffmpeg_fetchibuf()` function into multiple
smaller functions.
Conditional statements are moved to own funtions with human readable
names, so code flow is more clear.
To remove one branch of seeking, first frame is now decoded by
scanning, which will do only one iteration. So nothing has technically
changed.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D10638
This adds a Clamp node for Geometry Nodes Attributes.
Supports both Min-Max and Range clamp modes.
Float, Vector, Color and Int data types supported.
Reviewed By: HooglyBoogly, simonthommes
Differential Revision: https://developer.blender.org/D10526
As a rather premature optimization from rBbbe6d4492823, Object bounds
were only computed when either the Object or its Geometry were modified.
Prior to rB42198e9eb03b, this would work, as the Geometry was tagged as
modified if the Object's transform was also modified.
Since this tagging is not done anymore due to side effects, and since at
the time bounds are computed Objects were already processed and tag as
unmodified, the check on the modified status was always false.
For now remove this check, so the bounds are always unconditionally
updated. If this ever becomes a performance problem in large scenes with
motion blur, we will then try to find a way to nicely optimize it.
This would only affect BHV2 as OptiX and Embree handle object bounds
themselves.
When an addon has been removed, but its `.pyc` files are still there,
the Python module can still be loaded. However, because `__init__.py` is
missing, it becomes a namespace instead of a module, and its `__file__`
will be set to `None`. As a result, it's impossible to get the mtime
from the file (because there is none).
This should not influence any regularly uninstalled add-on, as that
would just remove the add-on's directory; I ran into the problem when
switching Git branches caused an add-on's Python files to disappear
while keeping the `__pycache__` directory around.
When rendering with cycles at some point the manifest is trimmed. This
leads to incomplete/corrupted metadata. This patch will make sure that
the manifest parser doesn't crash.
This solved the issue when the manifest is trimmed at the start of a
hash. Eg '"Name":"'.
This commit adds a node with a "Map Range" operation for attributes
just like the non-attribute version of the node. However, unlike the
regular version of the node, it also supports operations on vectors.
Differential Revision: https://developer.blender.org/D10344
This adds a new RNA method `Action.flip_with_pose(ob)` to flip the
action channels that control a pose.
The rest-pose is used to properly flip the bones transformation.
This is useful as a way to flip actions used in pose libraries,
so the same action need not be included for each side.
Reviewed By: sybren
Ref D10781
Support a cache for fast RNA path look-ups for RNA-path + index.
Unlike a regular hash lookup, this supports using a single lookup
for the RNA path which is then used to fill an array of F-curves.
Reviewed By: sybren
Ref D10781
This function returns an array of keyed frames with rounding,
to avoid duplicates caused by subtle floating point difference.
Reviewed By: sybren
Ref D10781
The recent commit that changed the size (rB83df3545246aada) left out
a few changed. This patch also adjusts the positioning and UV scale of
the generated plane accordingly.
Differential Revision: https://developer.blender.org/D10822
Purely local IDs would not always be used as proper barriers when
generating override hierarchy data, leading to imporperly trying to
resync override IDs from other hierarchies, which would break/cause
issues in case they would be duplicate overrides of the same linked
data.
Reported by Pablo from Sprite team, thanks.
We do not want to re-generate auto-overrides until everything that
needed to be resynced has been resynced. Otherwise it's a waste of time
and guaranteed loss of some override properties.
Mistake in rB9947f2095610 earlier today.
This will allow retrieving the instance groups from multiple geometry
sets and avoiding needing vectors of vectors to store the results.
This is useful when retrieving instances from a multi-input socket
of geometries.
select
In 'Set/Replace' mode this is not a problem, but 'Extend' or 'Subtract'
modes were useless with the current behavior.
The problem here is that 'node.select' fires before 'node.select_box'
(which is fine) but deselects immediately on click.
This issue has come up before in other editors, see
{T70457}
{rB395dfff103e1}
{rBa8ea1ea1b7d5}
Now delay deselection in empty space to mouse release (same as done in
before mentioned report).
also related:
ref T57918
ref T63994
Maniphest Task: T86867
Differential Revision: https://developer.blender.org/D10801
Replacing a hard-coded font size with font style widget size so it can be set by user.
Differential Revision: https://developer.blender.org/D10819
Reviewed by Dalai Felinto
account
UV coords are scaled by aspects (see UVsToTransData). This also applies
for the Cursor in the UV Editor which also means that for display and
when the cursor coords are flushed (new 'recalcData_cursor_image' was
added for this), these need to be converted each time.
Maniphest Tasks: T86796
Differential Revision: https://developer.blender.org/D10817
This flags moving the cursor in the Image Editor T_2D_EDIT with the
following benefits:
- dissallowing e.g. Z constraints
- improving the header display:
-- it should not use scene units
-- now respects if we are moving in pixel coords or not
part of upcoming fix for T86796
ref D10817
When subscribing to name-changes through the API,
the event doesn't trigger if the object is renamed in the outliner.
Fixed by publishing the RNA changes.
Reviewed By: lichtwerk
Ref D10732
This was only happening without an active object.
The "Operator Cheat Sheet" operator collects info about all operators,
and as part of that executes the callbacks to create dynamic enums. The
callback to enumerate the geometry operators (introduced in
rB370d6e50252b) attribute domains depends on a context object. If this
isn't available, we just have to return `DummyRNA_NULL_items`.
Maniphest Tasks: T86921
Differential Revision: https://developer.blender.org/D10814
Revert "Point users to new location of "Show Group Colors" option",
commit rB6ed6741ee35930bf81fad9a5eb6bb17eea168725.
These deprecation hints were intended for one release only, and thus can
be removed now.
This patch renames two domains:
* `Polygon` -> `Face`
* `Corner` -> `Face Corner`
For the change from `polygon` to `face` I did a "deep rename" where I updated
all (most?) cases where we refere to the attribute domain in code as well.
The change from `corner` to `face corner` is only a ui change. I did not see
a real need to update all code the code for that. It does not seem to improve
the code, more on the contrary.
Ref T86818.
Differential Revision: https://developer.blender.org/D10803
CacheFile writing code would not write generic ID data (call for it has
been missing since the initial commit, rB61050f75b13e).
While potentially affecting other areas (mostly CustomProperties/IDProperties),
this was a critical failure for liboverrides. Also added some workaround
code to allow opening broken files (though the override of the CacheFile
data-block will be lost).
This makes the buttons drawn on nodes for unconnected string and color
sockets draw with the correct 40% ratio for the label split rather than
the old 50% ratio not used elsewhere in Blender anymore. The benefit is
a cleaner look, because the button edges line up better with others, and
a bit more space to type in attribute names.
rBa8a92cd15a52 was adding external libraries includes to `INC`, which is
reserved to internal project includes.
`INC` does not allow duplicates, and when using system libs both PugiXML
and Haru headers are under the same path.
We would not add the "object line art" enum to the operator enum list,
if there were no active object in the scene.
This would make it impossible to call this operator from python code as
the enum would we hidden when we were not in a viewport context.
Always make the operator available, having no active object is not a
strict requirement for the operator to work, so expose it always.
Without `calc_object_remap` turned off in the conversion to and from
BMesh for the primitive nodes, the `CD_ORIGINDEX` custom data layer
has incorrect values. By using a different function to do the conversions,
we can avoid this problem.
Thanks to Jacques for finding the fix here.
Differential Revision: https://developer.blender.org/D10805
This error was produced because now it is possible to have several Lattice modifiers and the Bake was removing the lattice data of all modifiers.
Now the data is only recalculated and removed for the current modifier.
Also some cleanup of comments.
This patch adds support to export and import grease pencil in several formats.
Inlude:
* Export SVG
* Export PDF (always from camera view)
* Import SVG
The import and export only support solid colors and not gradients or textures.
Requires libharu and pugixml.
For importing SVG, the NanoSVG lib is used, but this does not require installation (just a .h file embedded in the project folder)
Example of PDF export: https://youtu.be/BMm0KeMJsI4
Reviewed By: #grease_pencil, HooglyBoogly
Maniphest Tasks: T83190, T79875, T83191, T83192
Differential Revision: https://developer.blender.org/D10482
Added a condition to the poll so that it ignores drag and drop on the button. The Paste Name operator is just not implemented. Doesn't work for shading nodes either.
Reviewed By: HooglyBoogly
Differential Revision: https://developer.blender.org/D10769
Add missing call to ED_node_tag_update_nodetree which solves the missing update on initial drag.
Reviewed By: HooglyBoogly
Differential Revision: https://developer.blender.org/D10769
Move Palettes type with other weird, not-really-data ID types (like
PaintCurves and Brushes), higher in the process.
Those preset-like types may use a lot of other ID types, but should only
be used by UI-related (and Scene, for tool settings) types.
Move Brushes type with other weird, not-really-data ID types (like
Palette and Brushes), higher in the process.
Those preset-like types may use a lot of other ID types, but should only
be used by UI-related (and Scene, for tool settings) types.
Move PaintCurve type with other weird, not-really-data ID types (like
Palette and Brushes), higher in the process.
Those preset-like types may use a lot of other ID types, but should only
be used by UI-related (and Scene, for tool settings) types.
Move `OB_GR` processing just after scenes, and before objects.
This is much more sensible in general, and fixes glitches in auto-resync
process of library overrides in particular.
Using the button context for operators is useful for other cases as well (where
the operator isn't the button operator itself). For example we'll need this for
the asset view UI template, where there will be additional operators that
should be able to act on button context.
Currently the procedural will add an entry to the cache for every frame
even if the data only changes seldomly. This means that in some cases we
will have duplicate data accross frames.
The cached data is now stored separately from the time information, and
an index is used to retrieve it based on time. This decoupling allows
for multiple frames to point to the same data.
To check if two arrays are the same, we compute their keys using the
Alembic library's routines (which is based on murmur3), and tell the
cache to reuse the last data if the keys match.
This can drastically reduce memory usage at the cost of more processing
time, although processing time is only increased if the topology may
change.
In order to update the BVH when only the transformations are changing,
we would tag the Object's Geometry as modified. However, when
displacement is used, and the vertices were not themselves modified,
this would cause us to redo the displacement on already displaced
vertices.
To fix this, use a specific update flag for detecting and notifying that
transformations were modified.
Regression caused by rBbbe6d44928235cd4a5cfbeaf1a1de78ed861bb92.
Make sure we have valid links and link pointers using
'ntreeUpdateTree()'.
Maniphest Tasks: T86314
Differential Revision: https://developer.blender.org/D10793
The problem happened when the selection only allowed "single_islands"
(only vertices are selected, no edges or faces).
The result of `is_zero_v3(v->no)` was erroneously being compared to `0.0f`
This commit corrects the wrong condition and optimizes it by adding a
earlier return when the islands don't need to be calculated.
(It also improves the code's readability by joining some variables in the
`struct TransIslandData`).
This adds a `BLI_assert_unreachable()` macro, that should be used instead
of `BLI_assert(false)` in many places.
* `BLI_assert_unreachable` provides more information than `BLI_assert(false)`
to people reading the code.
* `BLI_assert_unreachable` will print an error to `stderr` in a release build.
This makes it more likely that we will get bug reports when the assumptions
of a developer were wrong.
Differential Revision: https://developer.blender.org/D10780
- Use constexpr for better readability.
- Split in functions per backend.
- Split work scheduler global struct in smaller structs.
- Replaced std::vector with blender::Vector.
- Removed threading defines in COM_defines.h
Linked data should only be duplicated if relevant `eDupli_ID_Flags` flag
is set (`USER_DUP_LINKED_ID`), this is being taken care of by generic
`BKE_id_copy_for_duplicate` function, but for some reasons (?)
`BKE_object_duplicate` was directly using `BKE_id_copy` for itself...
Note that this had especially bad consequences when duplicating
overrides of collections...
It makes more sense to avoid mixing the primitives with other mesh
operations. We considered adding a "Mesh" prefix to the names too,
but decided to prioritize improvements to the node search menu instead,
and only consider that if there is an overlapping prefix name for
another data type.
The crash is caused by an out of bound access in the kernel due to
missing data update when a Volume's voxel data changes. Although the
previous bounding mesh is cleared, the Volume Node was not tagged as
modified, and therefore never rebuilt.
To fix this, tag the Geometries (not just Volumes, to be more robust) as
modified in Geometry.clear().
Regression caused by rBbbe6d44928235cd4a5cfbeaf1a1de78ed861bb92.
Crash occurs at `assert(builder->index_len < builder->max_index_len)`.
Non-bezier nurbs were being created with abnormal values causing the
causing edge count in the draw manager to be incorrect.
This commit also deduplicates and adds descriptions to the code.
Thanks @PratikPB2123 for pointing out where the error is.
Currently viewers and previews only display node trees that have at least one node with fixed resolution size. When all inputs are generated, nothing is displayed in most cases (RGB Node is displayed as a single pixel on previews). By generated I mean inputs not having resolution on their own, they create content dynamically given an output resolution.
This patch adds support for those cases by using an appropriate preferred resolution on Viewers/Previews which propagates to generated inputs as output resolution. Now:
- Viewers will display generated inputs with scene render resolution.
- Previews will display them with scene aspect ratio.
This is consistent with final render result and respects relative space.
The benefit for the user is being able to compose images without any input source. For example for creating mask images or simple backgrounds.
Reviewed By: Jeroen Bakker
Differential Revision: https://developer.blender.org/D10611
The goal of this patch is to remove the boilerplate code required to get
a string property that maps to an allocated char pointer in dna.
Previously, one to to provide three callbacks, all of which are not necessary
anymore for a simple string property.
Currently, when an empty string is assigned, the `set` function will always
allocate it as well, instead of assigning `NULL` to the pointer. Some structs
might support `NULL` while others don't, so this is the safer option for now.
Differential Revision: https://developer.blender.org/D10773
There were multiple cases that could lead to problems like moving meta
strip into itself or into it's children meta strips.
Print error string to console when invalid action is requested.
Cycles, Eevee, OSL, Geo, Attribute
Based on outdated refract patch D6619 by @cubic_sloth
`refract` and `faceforward` are standard functions in GLSL, OSL and Godot shader languages.
Adding these functions provides Blender shader artists access to these standard functions.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D10622
Currently, when a taper object is specified, the radius of the spline is
ignored. This patch adds a new option to control how the taper object
affect the effective radius of the spline. The option allow three modes
of operation:
- Override: The old method. The radius of the spline is ignored and
overridden.
- Multiply: The radius of the spline is multiplied by the taper radius.
- Add: The radius of the spline is added to the taper radius.
Ref D10779
Sorting of links on multi-input sockets where not recalculated after
using the knife operator. Added call to resort function before
cutting operator finishes.
Reviewer: Jacques Lucke
Differential Revision: https://developer.blender.org/D10783
- Rename:
`BKE_gpencil_object_material_get_index_name`, to
`BKE_gpencil_object_material_index_get_by_name`
Matching `BKE_gpencil_layer_get_by_name`.
- Move logic to ensure named materials into a new function:
`BKE_gpencil_object_material_ensure_by_name`
No need for a module to define a single panel, since this is an
object panel it can be included with other object panels.
If centralizing line-art properties is needed in the future,
this can be done in a `*_common` module.
In this case both the mirror object and object itself moving in time may
change the geometry and must be checked.
Differential Revision: https://developer.blender.org/D10757
In some multi-functions (such as a simple add function), the virtual method
call overhead to access array elements adds significant overhead. For these
simple functions it makes sense to generate optimized versions for different
types of virtual arrays. This is done by giving the compiler all the information
it needs to devirtualize virtual arrays.
In my benchmark this speeds up processing a lot of data with small function 2-3x.
This devirtualization should not be done for larger functions, because it increases
compile time and binary size, while providing a negilible performance benefit.
Following some discussion among the geometry nodes team, it was decided
that keeping the primitive nodes simpler and requiring a separate
transform node to move the generated geometry from the origin would
be better.
- It's more consistent with the current general idea of "building
block nodes"
- It makes more sense for the future when it will be possible to
use instancing to control the transforms.
- It reduces UI clutter when the controls are not necessary.
Although it works well in most cases, the algorithm to detect if a point
is within the limits of the camera does not work well in othographic mode.
This commit also adds the option `V3D_PROJ_TEST_CLIP_FAR` (currently unused).
Differential Revision: https://developer.blender.org/D10771
The issue was caused by the prefetch code having LOCK_MOVIECLIP lock
acquired while reading frames from the movie files. The need of the
lock was coming from the fact that `clip->anim` can not be accessed
from multiple threads, so that was guarded by a lock. The side effect
of this lock was that the main thread (from which drawing is happening)
did not have any chance passing through it in the cache code because
the prefetch happens so quickly.
The solution is to create a local copy of the clip with its own
anim handler, so that read can happen without such lock.
The prefetch is slower by an absolute number in seconds (within 10%
in tests here), but it is interactive now.
Re-adds a legacy document icon for macOS 10.14 Mojave that is
consistent with the system generated document icon on macOS 11
Big Sur. It uses the old-style document sheet icon, but includes the
file extension underneath the Blender icon (unlike the previous
legacy document icon that was removed in D10267).
Adds the missing description for the exported type identifier.
Finder now correctly displays “Blender File” instead of “data”
for Blender files.
Differential Revision: https://developer.blender.org/D10746
The root of the issue was caused by the PredictMarkerPosition()
always returning false when tracking backwards. This was making
it so tracker always had to run brute initialization, which is
an expensive operation.
From own timing here:
- Tracking forward takes 0.667637 seconds
- Tracking backward used to take 2.591856 seconds
- Tracking backward now takes 0.827724 seconds
This is a very nice speedup, although the tracking backwards is
still somewhat slower. Will be investigated further as part of
a regular development.
Code rebuilding/ensuring the sanity of the collection hierarchy was not
checking for a same collection being child of the same parent multiple
times.
This was already prevented to happen in code adding collections to other
collections, but not for the remapping case.
In collection/viewlayer synchronization code, in some cases, there are
extra unused view layer collections left in old list after all needed
ones have been moved to the new list.
Found while working on T86741.
Previously, different Random Float nodes would generate different values
depending on where they are in the node group hierarchy. This can be useful,
but should definitely not be the default behavior, because it is very inconsistent
with other nodes.
Previously, the signature of a `MultiFunction` was always embedded into the function.
There are two issues with that. First, `MFSignature` is relatively large, because it contains
multiple strings and vectors. Secondly, constructing it can add overhead that should not
be necessary, because often the same signature can be reused.
The solution is to only keep a pointer to a signature in `MultiFunction` that is set during
construction. Child classes are responsible for making sure that the signature lives
long enough. In most cases, the signature is either embedded into the child class or
it is allocated statically (and is only created once).
Added a first test case for review. This will be the base for future test cases.
The current API is sufficient for what is expected for such a low level API.
One concern is that you need to trigger a save in order to update the library overrides
structure. Not expected from TD/Dev point of view.
Test cases are very important when implementing restrictive mode as it is a second evaluation mode that
has impact on the (current) permissive mode.
Reviewed By: Sebastián Barschkis, Bastien Montagne
Differential Revision: https://developer.blender.org/D10747
The impact of translations on UI performances is neglectable, and this
gives better handling of some odd cases where original language of an
add-on is not English.
Currently 3 of the printf lines use the format specifier "%lu" for data
of type size_t instead of "%zu". While this works, this creates compiler
warnings.
This diff fixes those warnings by using the correct format specifier.
Tested using MSVC, but should be correct on any compliant compiler.
Reviewed By: Sebastian Parborg
Differential Revision: http://developer.blender.org/D10761
- Use `use_` prefix for boolean properties.
- Use `_all` instead of `_everything`,
following existing conventions and the properties UI text.
- Use `object_instances` instead of old term `dupli` / `duplication`.
- Use `clip_plane` instead of `clipping_boundaries`,
matching the RegionView3D property naming.
When a function is executed for many elements (e.g. per point) it is often the case
that some parameters are different for every element and other parameters are
the same (there are some more less common cases). To simplify writing such
functions one can use a "virtual array". This is a data structure that has a value
for every index, but might not be stored as an actual array internally. Instead, it
might be just a single value or is computed on the fly. There are various tradeoffs
involved when using this data structure which are mentioned in `BLI_virtual_array.hh`.
It is called "virtual", because it uses inheritance and virtual methods.
Furthermore, there is a new virtual vector array data structure, which is an array
of vectors. Both these types have corresponding generic variants, which can be used
when the data type is not known at compile time. This is typically the case when
building a somewhat generic execution system. The function system used these virtual
data structures before, but now they are more versatile.
I've done this refactor in preparation for the attribute processor and other features of
geometry nodes. I moved the typed virtual arrays to blenlib, so that they can be used
independent of the function system.
One open question for me is whether all the generic data structures (and `CPPType`)
should be moved to blenlib as well. They are well isolated and don't really contain
any business logic. That can be done later if necessary.
We cannot use `std::variant` yet, because not all of the compilers
we support have a working version of it yet. For now, I just replaced
it with multiple `std::option` which is good enough, because currently
`CellValue` is only used for the cells that are actually drawn in
the spreadsheet.
When the movie wasn't found, uninitialized values would be used for
the original width/height.
Also use int instead of float as the image size is stored as an int.
SetWindowPos must be called after SetWindowLongPtr in order to see changes to window frame style.
Differential Revision: https://developer.blender.org/D10756
Reviewed by Ray Molenkamp
Some generic algorithms from the standard library like `std::any_of`
did not work with all container and iterator types. To improve the
situation, this patch adds various type members to containers
and iterators.
Custom iterators for Set, Map and IndexRange now have an iterator
category, which soe algorithms require. IndexRange could become
a random access iterator, but adding all the missing methods can
be done when it is necessary.
The description was missing from the Quick Liquid operator.
The fix adds the following description:
"Make selected objects liquid"
Reviewed By: sebbas
Maniphest Tasks: T86746
Differential Revision: https://developer.blender.org/D10777
The offset when creating the keyframes was set to `frame_target + i` but
`i` starts iterating from the current frame number.
The fix uses just `i` as the frame number.
Reviewed By: antoniov
Maniphest Tasks: T86745
Differential Revision: https://developer.blender.org/D10772
When joining curves, the resulting curve will inherit the bevel depth of
the active curve, but the radii would stay the same which leads to
changed appearance when joining.
Now compensate for this taking the different bevel depths into account
(if present).
Was a feature request here (and I also think we had reports about this
-- which were usually turned down as not-a-bug):
https://blender.community/c/rightclickselect/bhhbbc/
Differential Revision: https://developer.blender.org/D10752
- Adding effect strip resulted in strip with no name
- Adding sound strip attempted to read `fit_method` RNA property,
that did not exist, causing error messages in console
These issues were introduced in bbb1936411.
SEQ_add_effect_strip used SeqLoadData.image.end_frame to set end
frame. This was mistake introduced in last refactoring patch.
Use effect data, not image data, when adding effect strips.
Reviewed By: sergey
Differential Revision: D10633
When transition effect is placed between 2 adjustment layer strips,
only first adjustment layer was rendered by effect.
Limit timeline_frame range to adjustment strip frame range.
This timeline configuration is technically invalid, because strips
should overlap when using transition effect. This was never restricted
and instead of producing no image, transition effect used first and
last frame of source strip. Many users got used to this "feature" so
I think it make sense to fix this case so it behaves like other strip
types.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D10562
- Make properties alligned
- Add decorate to volume
- Add heading to Grease Pencil and Transparent
Reviewed By: ISS
Differential Revision: https://developer.blender.org/D10663
Add option to override current frame whem transforming strip handles.
Option can be found in View menu of VSE preview, or in timeline when
using backdrop.
Reviewed By: ISS
Differential Revision: https://developer.blender.org/D10424
- Position text in center of image
- Increase default font size so it's more visible
- Increase font size limit
- Increase limit for location, so text can be scolled off screen
- Wrap text by default
- Tweak default box and shadow color
- Change text shadow position
- Text box no longer casts shadow
Reviewed By: ISS
Differential Revision: https://developer.blender.org/D10571
Bold and italic fonts can be switched quickly by presing corresponding
button.
Reviewed By: ISS
Differential Revision: https://developer.blender.org/D10542
`SequenceElement` type `orig_height` and `orig_width` members were
set to incorrect size when using proxies and not set when strip was
added which caused value to be unset.
Since now image dimensions must be read when strip is created,
these members can be initialized. When proxies are used, do not set
original size since it is not guaranteed, that proxies are exact size.
These values are not guaranteed to be up to date or exact. They should
be used for strictly informative purposes.
Reviewed By: ISS
Differential Revision: https://developer.blender.org/D6506
This function can be used to create empty meta strip, which is not
straightforward when using operators. Very useful for import/export
scripts.
Reviewed By: ISS
Differential Revision: https://developer.blender.org/D10661
This function can be used to move strips into meta strips with no
side effects like change of selection state.
Reviewed By: ISS
Differential Revision: https://developer.blender.org/D10759
This visualization of nested node groups makes it easier to debug
some issues. Muted nodes, muted links, reroute nodes and unavailable
sockets are removed from the visualization to keep it clean.
Nested node groups are visualized using colored clusters.
The new clamping works by modifying the lamp internal radius which
then soften the light contribution.
However this does remove more light compare to the old solution.
This is because the clamp now affects the light over a much larger
distance since it is smoother. Old scene needs manual tweaking.
Soft surface shadows were already supported but now we support
soft shadows of the volume themselves.
This is only enabled if the light casts shadow and the scene soft
shadows toggle is enabled.
Previously area lights were just considered as point lights.
We now use a "most representative point" technique that make the
light shape appearant and gives more homogenous result.
This technique is quite cheap but it is not physically correct.
So I came up with a power function to have almost the same intensity
output as cycles in the general case.
Sun lights are treated as distant light source and need to gather
shadowing from the full frustum.
This might have performance impact on certain scenes.
This adds 2 new sliders for light objects that modulates the diffuse
light and the volume light intensities.
This also changes the way volume light is computed using point lamp
representation. We use "Point Light Attenuation Without Singularity"
from Cem Yuksel instead of the usual inverse square law.
Previously we could crash because we would not check if the modifier in
question actually was a line art modifier. We also did not query if the
modifier was disabled.
This patch adds data about instances generated by geometry nodes
to the spreadsheet. The transform data is decomposed into position,
rotation, and scale, and there is a name column to display the name
of the instanced object or collection.
This data is implemented specifically for the spreadsheet, because
we're not sure that we want to expose this data as attributes for the
use elsewhere.
Differential Revision: https://developer.blender.org/D10770
The point separate node was failing in situations where one of the
outputs was empty. In addition, the code was not structured very well.
This new implementation stores less temporary information, is more
geometry component type agnostic, and is more self-descriptive.
It also solves the problems mentioned above.
Fixes T86573
Differential Revision: https://developer.blender.org/D10764
Make options that are usually only useful in edge cases off by default.
These settings would requite extra work and thus the modifier evaluation
would be needlessly slower as most users do not need these options to be
on.
It caused some chaining errors when used in combination with image space
chaining. After some internal discussion, we realized it is not
useful as chaining in image space essentially does the same thing.
Instead of storing those in scne's master collection, which is fairly
annoying, we now add them to a (hidden) specific collection. Easy to
ignore, or check and cleanup.
This is generally what people expect when generating a cone. Note that
this translation currently happens after the rotation, but since the rotation
will likely be removed in the future, that won't be a problem for long.
This patch adds the remaining 6 interpolations for mesh domains.
The new interpolations are:
- Corner / point / polygon to edge
- Edge to corner / point / polygon
After this it is possible to adapt an attribute to and from every
mesh domain. This is simple to test with the "Attribute Convert" node.
Though, as a note for the future, there are still some improvements
possible to the interpolations, like lazily calculating values for the
interpolations where it's possible, and slightly improving the
algorithms used for some interpolations, like using corner angles
for polygon to point.
Differential Revision: https://developer.blender.org/D10765
This is not only potentially extremely expensive, it is also fairly
futile, and code is not designed to handle it currently anyway (could
easily end up in inifinite loops and other crashes).
- Remove use of evaluated poses, instead calculate transformations
into an array which is applied afterwards.
- Only update ID's for poses that have been changed.
This is simply a convenience when using this type. More similar
constructors can be added in the future when they are useful.
Differential Revision: https://developer.blender.org/D10714
Allow python access to the `reset_view` functionality which before
was only available through the menu. This was suggested for
consistency after D10561.
Differential Revision: https://developer.blender.org/D10595
The size in the transform matrices was extra, since it is also
passed as an argument to the BMesh operators.
Differential Revision: https://developer.blender.org/D10763
Previously even if the input goemetry set had no point cloud or no mesh
instances, `geometry_set_realize_instances` would create empty data.
This isn't necessarily bad, but it can complicate things down the line if
there are a bunch of empty components getting passed around.
This was reported for geometry nodes, but was true for all nodetrees
(e.g. after deleting the active node). Geometry node trees just made
this more obvious since they start without an active node to begin with.
Fix provided by @lone_noel, thx!
Maniphest Tasks: T86677
Differential Revision: https://developer.blender.org/D10762
Caused by {rB2917f550caa9} which renamed the entry in the toolsystem,
but not the corresponding keymap.
Maniphest Tasks: T86548
Differential Revision: https://developer.blender.org/D10725
Since {rB46aa70cb486d}, using `NC_SPACE | ND_SPACE_VIEW3D` as notifier is
restricted to space data as a reference. This was still used though for
RNA updates in other places (namely `rna_camera`, `rna_scene`,
`rna_animviz`), and passing NULL would automatically set the notifier
reference to the owner id. Above commit would happily filter these out,
leading to missing refreshes.
Now use more specific notifiers (in case of animviz a new
`ND_DRAW_ANIMVIZ` was added).
This was reported for Camera background images btw.
Fixes T86670.
Maniphest Tasks: T86670
Differential Revision: https://developer.blender.org/D10758
Data-block selectors (or other selectors using the
`UILayout.teamplate_search` or `UILayout.template_search_preview`)
in headers used a fixed size width for the name button. Often that meant
the text would be clipped to fit even though there was plenty of
whitespace surrounding the data-block selector.
Now the name button can increase in width with the contained string,
clamped to some arbitrary and quite big maximum (3 times the minimum
width).
We could further take the available space into account, to reduce the
need for scrolling in the header. But I don't think that is much of an
issue and the name clipping it would re-introduce would probably be more
annoying.
Only use non included objects for potential occlusion queries.
Don't calculate lines for them.
This fixes intersection lines appearing on objects not included by the
lineart modifier.
When doing a box selection in the dopesheet in the header area, the data block type was changed to annotation because the pointer was not usable and produced an unexpected result.
Thanks to @pullup for finding a way to reproduce the bug.
This introduces two functions to the blenloader module, here shown as
calls for brevity:
* `temp_lib_ctx = BLO_library_temp_load_id(real_main, blend_file_path, idcode, idname, reports);`
* Now the data in `temp_lib_ctx->temp_id` can be used (but ought not to
be not assigned to non-temp datablocks).
* `BLO_library_temp_free(temp_lib_ctx);`
The first loads a datablock from a blend file, and returns it as part of
a `struct TempLibraryContext`. This struct contains the temp-loaded ID,
as well as enough information to correctly free everything again.
Differential Revision: https://developer.blender.org/D10736
Transforms in certain cases
This was caused by wrong flag checking in {rB278011e44d43}, which just
seems to be a copy-paste error.
For example, enabeling 'Auto-Offset' in the View menu would lead to CM
being ignored.
Maniphest Tasks: T86219
Differential Revision: https://developer.blender.org/D10751
ADD_CHECK_CXX_COMPILER_FLAG(CXX_WARNINGSCXX_WARN_NO_OVERLOADED_VIRTUAL-Wno-overloaded-virtual)# we get a lot of these, if its a problem a dev needs to look into it.
message(FATAL_ERROR"32 bit compiler detected, blender no longer provides pre-build libraries for 32 bit windows, please set the LIBDIR cmake variable to your own library folder")
@@ -9,24 +9,20 @@ if "%BUILD_WITH_SCCACHE%"=="1" (
if"%WITH_CLANG%"=="1"(
setCLANG_CMAKE_ARGS=-T"llvm"
if"%WITH_ASAN%"=="1"(
)
if"%WITH_ASAN%"=="1"(
setASAN_CMAKE_ARGS=-DWITH_COMPILER_ASAN=On
)
)else(
if"%WITH_ASAN%"=="1"(
echo ASAN is only supported with clang.
exit /b 1
)
)
if"%WITH_PYDEBUG%"=="1"(
setPYDEBUG_CMAKE_ARGS=-DWINDOWS_PYTHON_DEBUG=On
)
if"%BUILD_VS_YEAR%"=="2019"(
setBUILD_PLATFORM_SELECT=-A %MSBUILD_PLATFORM%
)else(
if"%BUILD_VS_YEAR%"=="2017"(
setBUILD_GENERATOR_POST=%WINDOWS_ARCH%
)else(
setBUILD_PLATFORM_SELECT=-A %MSBUILD_PLATFORM%
)
setBUILD_CMAKE_ARGS=%BUILD_CMAKE_ARGS% -G "Visual Studio %BUILD_VS_VER%%BUILD_VS_YEAR%%BUILD_GENERATOR_POST%"%BUILD_PLATFORM_SELECT%%TESTS_CMAKE_ARGS%%CLANG_CMAKE_ARGS%%ASAN_CMAKE_ARGS%%PYDEBUG_CMAKE_ARGS%
is a full-featured 3D application. It supports the entirety of the 3D pipeline - modeling, rigging, animation, simulation, rendering, compositing, motion tracking, and video editing.
is a full-featured 3D application. It supports the entirety of the 3D pipeline - '''
'''modeling, rigging, animation, simulation, rendering, compositing, motion tracking, and video editing.
Use Blender to create 3D images and animations, films and commercials, content for games, architectural and industrial visualizatons, and scientific visualizations.
Use Blender to create 3D images and animations, films and commercials, content for games, '''
r'''architectural and industrial visualizatons, and scientific visualizations.
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
