This fixes a performance regression on Ampere cards, on specific scenes like
classroom. For cycles-x there is little difference, but this is still helpful
for LTS releases, and we need to upgrade at some point anyway.
This is an easy & safe, yet not-so-nice way to address the
LayerCollections vs. Collections hierarchy resync problem.
Currently this resync is enforced everytime something changes in the
Collections hierarchy, which is extremely inneficient, and can even
produce 'loss' of LayerCollection data during complex Collection
processes.
Current example is during Library Overrides resync process. New code:
* Makes resync significantly faster (between 10 and 15%).
* Fixes 'disappearing' layer collections settings on sub-collections'
layers.
NOTE: This is not a proper fix for the underlying issue. However,
implementing and testing the 'lazy update' solution as proposed by
{T73411} requires a significant amount of time (especially in testing
and tracking all places where code would need to ensure LayerCollections
are up-to-date), which is not possible currently.
Differential Revision: https://developer.blender.org/D11889
Mantflow only supports wavelet noise, thus the parameter with only one option is no longer useful.
Differential Revision: https://developer.blender.org/D6770
In the draw module, it's not easy to identify what its header is, and
where the shared functions are.
So move `draw_cache_extract_mesh_extractors.c` and
`draw_cache_extract_mesh_private.h` to the same folder as the extractors
and rename these files to make them more identifiable.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11991
Fix hold offset check causing missing snapping point when strip have
only still frames.
Fix effect strips of transformed strips causing snapping to prevoius
strip positions.
Reviewed By: mano-wii
Differential Revision: https://developer.blender.org/D11948
Label for snapping current frame to strips was cut off and not very readable.
Reviewed By: HooglyBoogly
Differential Revision: https://developer.blender.org/D11951
Cached OCIO processors were not freed, instead the color spaces were freed twice.
Reviewed By: brecht, sergey
Differential Revision: https://developer.blender.org/D12011
Fix for T85517
Bug: Couldn't type space while holding down the shift key in text spaces (e.g. when saving a file, changing the name of object).
Changes: Removing the key combination of Shift + space in `WM_event_is_ime_switch` method.
Reviewed By: harley, mont29
Maniphest Tasks: T85517
Differential Revision: https://developer.blender.org/D10452
Six years ago, Bug T44961 about unwanted spikes had me not do a loop
slide if the angle was too extreme, to avoid unwanted spikes.
The current bug showed that that angle was much too big, and limited
desired behavior in many cases. Changing the angle from 0.25 radians
to 0.0001 radians (about 0.006 degrees) still fixes the original bug
and seems very unlikely to be limiting desired behavior now.
The old modifier code, now just used for Fast, has code in it to
flip faces of arguments when their tranform's negativity differs
from the main object's transform's negativity.
I had neglected to put that logic in when I made the change that
skipped the round trip through BMesh.
Fixing this means that the results are more what the user expects
when some or all operands have negative scales.
This patch addresses the issue raised in T88340.
When entering a pose sliding operator bones would automatically get hidden.
While technically not a bug it was decided that it is too confusing.
Hiding with 'H' is still possible though, just won't happen automatically
Reviewed by: Sybren A. Stüvel
Differential Revision: https://developer.blender.org/D11883
Ref: D11883
Abbreviations are harder to read and understand thus it is best to be direct.
For example without understanding and context it is hard to know exactly what "Len" means.
This change transitions libmv/osd tests to our
blender_add_test_executable macro that explicitly
takes the include directories as a parameter.
This is in preparation for future clean-up of
global include directories.
Differential Revision: https://developer.blender.org/D12012
Reviewed By: sergey
This improve the cloth modeling workflow by allowing you to weld only the
edges that are used for the sewing forces.
Reviewed By: mano-wii, weasel
Differential Revision: https://developer.blender.org/D10710
**Drawing Changes:**
- F-curve drawing for Stretch, Multiply, Length and Frame Number.
- Value drawing when no keyframes for Stretch, Length and Frame Numbers.
General view of the new drawing for each speed effect mode:
{F9796642, size=full}
Detail of the horizontal zero (blue) line in the new `Multiply` mode:
{F9798520, size=full}
Nice to have (but I don't know how):
- Auto adjusting of endframe when using Multiply or Boost.
Differential Revision: https://developer.blender.org/D6110
Addresses T76003. When using VR with Eevee and viewport denoising,
scene geometry could sometimes be occluded for one eye. Solution is
to use a separate GPUViewport/GPUOffscreen for each VR view instead
of reusing a single one for rendering.
Reviewed By: Julian Eisel, Clément Foucault
Differential Revision: http://developer.blender.org/D11858
The core problem is that the geometry nodes modifier sometimes support
"mapping" (i.e. it remembers which new vertices correspond to edit mode
vertices) and sometimes it does not, depending on what the nodes are doing.
Also see rB07ce9910f7cc.
The solution here is that the fallback case in `BKE_mesh_foreach_mapped_edge`
does not call the callback with "invalid" indices.
Differential Revision: https://developer.blender.org/D12007
This might change the rotation of some instances after a Curve to Points.
Unfortunately, there is not much we can do about that, the math before
was just wrong. The forward and up axis stayed the same though.
Differential Revision: https://developer.blender.org/D12006
The problem was that the modifier was reevaluated all the time, even
between showing the attribute search and clicking on the attribute
name. This freed the data referenced by attribute search. The real bug
here was that the dependency graph was tagged for update even
though nothing changed. This was because the spreadsheet thought
its active context has changed and it wanted to compute the new
value to be shown in the spreadsheet.
The reason for the bug was that I confused how the tree-path of a
node editor works. The second element in the tree path contains
the name of the group node in the root tree that we're in (instead
of the first element).
Differential Revision: https://developer.blender.org/D12009
When the new "need_ids" flag is false and the output type is not
one of the valid BMesh kinds, there is no need to propagate even
a dummy id to all of the faces.
Share functionality for single and multi-threaded edge-split tagging.
Remove logic that ensured vert & loop indices in bm_mesh_edges_sharp_tag
(missing from fd9fc809b7).
Improves control over the XR reference space by using the stage ref
space (user-defined tracking bounds) instead of local ref space
(position at application launch), if available. Also adds an
"absolute tracking" session option to skip applying eye offsets that
are normally added for placing users exactly at landmarks.
By enabling absolute tracking, users can define the tracking origin
in a way that is not linked to the headset position. Instead, the
tracking values given by the XR runtime are left unadjusted and a
user can manually calibrate an "origin" landmark object to adjust to
their real world space.
Can be useful for applications that use external tracking systems
and those that primarily only need to use controllers and not the
headset (e.g. motion capture).
The absolute tracking option requires an update to the VR
Scene Inspection addon to be accessible by regular users.
Reviewed By: Julian Eisel
Differential Revision: http://developer.blender.org/D10946
This was added in 0b7f581397
but seems not to be needed as the assignment was never correct
since only one corner on either side of the smooth edge had the
vertex normal written to it.
Merge the sharp edge tagging into bm_mesh_loops_calc_normals,
this has the advantage that edge tagging can be performed as part of
walking over each vertices edges - instead of tagging in a separate loop.
Even though this will tag edges twice (once for each vertex),
the computation isn't heavy as it's only calculating a dot-product
between the two face users to compare the angle.
This change combined with 4ba06ad0a8
makes BM_loops_calc_normal_vcos around 5.68x faster,
with an overall speedup over 2.6x when transforming a high poly mesh.
(tested on a system with 32 cores).
Reviewed By: mont29
Ref D11970
Supported multi-threading for bm_mesh_loops_calc_normals.
This is done by operating on vertex-loops instead of face-loops.
Single threaded operation still loops over faces since iterating
over vertices adds some overhead in the case of custom-normals
as the order used for accessing loops must be the same as iterating
of a faces loops.
From isolated timing tests of bm_mesh_loops_calc_normals on high
poly models, this gives between 3.5x to 10x speedup,
with larger gains for meshes with custom-normals.
NOTE: this is part one of two patches for multi-threaded auto-smooth,
tagging edges as sharp is still single threaded.
Reviewed By: mont29
Ref D11928
Extract the slider gui implemented for the pose slide tools.
Generalise it so it can be used by other tools as well.
Reviewed by: Sybren A. Stüvel
Differential Revision: https://developer.blender.org/D9314
Ref: D9314
With this commit, node warnings added to nodes during evaluation
(not "Info" warnings) will also draw in the modifier. In the future
there could be a "search for this node" button as well.
Differential Revision: https://developer.blender.org/D11983
Because these handles are calculated lazily, we need to make sure they
are calculated before switching to a manually positioned mode.
I doubt it would ever be necessary, but theoretically this could happen
on a per-point level, to avoid calculating handles not in the selection.
**Changes:**
- New enums correspond to 4 modes: `Stretch`, `Multiply`, `Frame Number` and `Length`.
- "`Multiply Factor`" has been removed;
- Value corresponding to "`use as speed`" enabled is now the value appended to the `Multiply` enum;
- Value corresponding to "`use as speed`" disabled is now the value appended to the `Frame Number` enum;
- Value corresponding to "`Scale to Length`" enabled is now the value appended to the `Length` enum;
- Except `Stretch` each mode has now its respective control values.
Differential Revision: https://developer.blender.org/D11856
It's the case of Image or Movie Clip node when not selecting any
source or an empty one.
Render methods expect an output buffer with size, only render
operations with resolution.
When exiting editmode, customdata [and thus attributes] are copied from
**BMesh** to **Mesh** (in `BM_mesh_bm_to_me`).
When adding attributes, these were always added via
`CustomData_add_layer_named` only.
Instead (if we are in mesh editmode), we need to go through
`BM_data_layer_add_named` to properly update the **BMesh** (see
`update_data_blocks`) so we have stuff available to copy back to
**Mesh** when exiting editmode.
Same is done for removing (also going through BMesh).
This is now done, DomainInfo is updated to point to BMesh (instead of
Mesh) customdata when in editmode.
Maniphest Tasks: T90026
Differential Revision: https://developer.blender.org/D11998
This node takes a curve and a point selection and allows you to set the
specified (or all) points left/right or both handles to a given type.
Differential Revision: https://developer.blender.org/D11992
* Allow specifying a folder and automatically setting the proper executable
name depending on the operating system
* Use executables from configs for listing devices instead of a blender
command being available
On Windows, using `bpy.data.orphans_purge` with some arguments (eg: `do_recursive=True`) does not produce the expected results. This is due to arguments not being parsed correctly on this platform with the current code.
The proposed fix is based on how other functions with boolean attributes are exposed to the Python API.
Reviewed By: #python_api, mont29
Maniphest Tasks: T89733
Differential Revision: https://developer.blender.org/D11963
This was reported in T90026 for attributes, but was also true for:
- UVMaps
- Vertex Colors
- Sculpt Vertex Colors
- Face Maps
For Vertex groups and Shapekeys this was already done (in that their
remove poll would check if there is a vertex group or shapekey to begin
with), now make this consistent across all mentioned types.
Thx @vvv for the initial patch (where this was done for attributes only)
ref T90026
Reviewed By: HooglyBoogly
Maniphest Tasks: T90026
Differential Revision: https://developer.blender.org/D11990
I'm trying to move away from general files with lots of things in them,
and instead have many small & focused files. I find that easier to
work with since everything has clear responsibilities, even if there is
some minor overhead in managing all these files.
I also try to differentiate more clearly between public and internal
files. So source files and internal headers are in a `intern/`
sub-directory, public functions are in a number of headers one level
higher.
For convenience and to make this compatible with our existing general
headers in `editors/include`, I made the `ED_asset.h` there include all
these public headers.
This is of course a bit of an experiment, let's see how it works in
practice.
Also corrected the name of `ED_asset_can_make_single_from_context()`.
The `ibo.tris` extraction in multithread is currently only done if the
mesh has only 1 material.
Now we cache a map indicating the index of each polygon after sort and
thus allow the extraction of tris with materials in multithreaded.
As caching is a heavy operation and was already being performed in
multi-thread for triangle offsets, no significant improvements are
expected.
The benefit will be much greater when we can skip updating the cache
while transforming a geometry.
**Profiling:**
||master:|PATCH:
|---|---|---|
|large_mesh_editing_materials:|Average: 13.855380 FPS|Average: 15.525684 FPS
||rdata 9ms iter 36ms (frame 71ms)|rdata 9ms iter 29ms (frame 64ms)
|subdiv_mesh_final_only_materials:|Average: 28.113742 FPS|Average: 28.633599 FPS
||rdata 0ms iter 1ms (frame 36ms)|rdata 0ms iter 1ms (frame 35ms)
1.1x overall speedup
Differential Revision: https://developer.blender.org/D11445
This was an open TODO, I wanted to have code for translating asset
library references from and to enum values in a central place, and
access that in the same way from both the Asset Browser and the
Workspace RNA code.
* Adds own file for the related functions.
* Adds doxygen comments.
* Updates RNA callbacks to properly use these functions.
* Let these functions call each other, avoid duplicating logic.
Since recently it's possible to access assets from outside the
File/Asset Browser, via the asset view template. So we are slowly
moving away from file space specific code to dedicated asset system
code. I introduced `AssetLibraryReference` as a duplicate of
`FileSelectAssetLibraryUID`, with a plan to delete the latter in a
separate cleanup commit. That's exactly what this commit is.
This will cause Asset Browsers to open with the default "Current File"
Asset Library. We could avoid that, but it's a minor issue really.
This new variable was introduced with 7898089de3. We don't usually use
an `active` prefix variable. Plus, this makes the name match the one of
the Asset Browser active library variable, so we can use the
`rna_def_asset_library_reference_common()` helper for both.
This will cause Asset Views to open with the default "Current File"
Asset Library. We could avoid that, but it's a minor issue really.
Currently the wording is a bit unclear: it doesn't specify //what// the particles will be converted into. This clarifies it by stating what the particles will be converted into: they will either be converted to a mesh or the instances will be made real.
Reviewed By: Blendify
Differential Revision: https://developer.blender.org/D11795
Fix an incoherence between the Eevee Materials menu and the Cycles Materials menu :
Eevee :
{F10230448}
Cycles :
{F10230449}
Simply Fixed by replacing the Cycles UI code by the Eevee UI code.
Thanks to @Brainzman for helping me create this diff and translate
Reviewed By: Blendify
Differential Revision: https://developer.blender.org/D11979
Just add a check for whether the mesh has faces when retrieving an
attribute on the corner domain. In the future there could be an info
message in the node in this case, since maybe it's not intuitive.
Would previously pass a few properties that are available via the
asset-handle now. This asset-handle is also required for some of the
asset API, e.g. the temporary ID loading. This will probably be needed
before too long.
For this to work, the utility function needs to be callable without
context, which is only needed for a File Browser specific hack anyway
(doesn't apply to this usage of it).
While the asset-handle design is supposed to be temporary (see
35affaa971), I prefer keeping the fact that it's nothing but a file
entry pointer an implementation detail that is abstracted away. So this
introduces getters for the file data we typically access for
asset-handles.
Note that the current asset-handle design is temporary, see
35affaa971. I still prefer this to be const, as code outside the
asset-list/file-list code should never mess with the file data of an
asset.
The asset handle design is only temporary (see 35affaa971) and this
RNA property is only needed for internal, technical reasons of the asset
view template. So although not nice, at least make it clear in the RNA
property description that this should not be used.
Using const indexes and offsets helps to make the logic less sequential,
which is hopefully easier to understand and possibly easier to parallelize
in the future. Also order return arguments last.
If there were lots of selected objects without an existing rigid body,
we would add rigid bodies to them one by one.
This would be slow in python, now we instead do this as a batch
operation in C.
On my (Intel) MacBook it used to take 60 seconds and with this change it
takes about 0.3 seconds.
Reviewed By: Sebastian Parborg
Differential Revision: http://developer.blender.org/D11957
Remove the assumption of the pose library that Action groups are named
after the bones in the armature. Even though this assumption is correct
when the keys are created by Blender, action groups can be renamed. Keys
created by Python scripts can also use arbitrary group names.
Since there is more code in Blender making this assumption, and looping
over selected bones is also a common occurrence, this commit contains
some generic functionality to aid in this:
- `BKE_armature_find_selected_bones`: function that iterates over all
bones in an armature and calls a callback for each selected one. It
returns a struct with info about the selection states (all or no bones
selected).
- `BKE_armature_find_selected_bone_names(armature)` uses the above
function to return a set of selected bone names.
- `BKE_pose_find_fcurves_with_bones()` calls a callback for each FCurve
in an Action that targets a bone, also passing it the bone name.
Convert `pose_backup.c` (in C) to `pose_backup.cc` (in C++). This will
make future improvements easier. For now, it's the same code with just
some additional explicit casts (C++ doesn't allow implicitly casting
`void *`), `NULL` changed into `nullptr`, and some other simple changes.
No functional changes.
When done from the Properties Editor, the context's modifier should be
used (this is where the button is located), when done from elsewhere,
the active modifier is still the way to go (since the context modifier is
not available then)
Maniphest Tasks: T89982
Differential Revision: https://developer.blender.org/D11972
The node tagged polys normals dirty, but the function to calculate the
normals didn't clear the dirty flags for polys. Now clear the poly and
corner dirty normal flags.
The test forgot to set the new need_ids field, which luckily
exposed a bug in the C api for delaunay when that field is false.
Fixed the bug and the test, and added a test for the need_ids false
case.
Wording on the UI, slider consistency and material mask switches layout.
Reviewed By: Sebastian Parborg (zeddb)
Differential Revision: http://developer.blender.org/D11839
Also use `const Curve *` instead of `const Object *`, since the
function works at a lower level than objects anyway.
And also remove another unused function. Since this section of code
for converting curves to meshes will likely be replaced, it's nicer to
see which parts actually remain used at this point.
Function signatures for snap callbacks used `const` incorrectly
which was hidden by casting function types.
This made it seem as if the input arguments wouldn't be change and
wouldn't be initialized.
Name return arguments with an `r_` prefix, order them last,
remove function casts and correct `const` usage.
When fixing issues, seeing operation results can be helpful for
detecting which operation went wrong.
This commit adds an option for exporting all operations results to
image files.
Exceptions are:
- Output operations: They are already exported or can be seen in UI.
- Constant operations: There are too many and is rarely useful.
They are exported to "<temp session folder>/COM_operations/"
with filenames "<operation class name>_<operation id>.png".
Only works on full frame execution mode.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11722
Adds full frame implementation to this node operation.
No functional changes.
2.4x faster than tiled fallback.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11677
Adds full frame implementation to this node operations.
No functional changes.
2.3x faster than tiled fallback on average.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11686
Adds full frame implementation to this node operation.
No functional changes.
No performance changes.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11698
Adds full frame implementation to this node operation.
No functional changes.
No performances changes.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11750
Adds full frame implementation to this node operation.
No functional changes.
3x times faster than tiled fallback.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11635
Currently we mostly iterate buffer areas using x/y loops or through
utility methods extending from base classes.
To simplify code in simple operations this commit adds wrappers for
specifying buffer areas and their iterators for raw buffers with any
element stride:
- BufferRange: Specifies a range of contiguous buffer elements from a
given element index.
- BufferRangeIterator: Iterates elements in a BufferRange.
- BufferArea: Specifies a rectangle area of elements in a 2D buffer.
- BufferAreaIterator: Iterates elements in a BufferArea.
- BuffersIterator: Simultaneously iterates an area of elements in an
output buffer and any number of input buffers.
- BuffersIteratorBuilder: Helper for building BuffersIterator adding
buffers one by one.
For iterating areas coordinates it adds `XRange` and `YRange` methods
that return `IndexRange`.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11882
Although currently only the asset list code uses the asset library
reference wrapper, it can stand on its own and may be used in more
places in the future. So I prefer to give it its own source & header
file.
Also removed unused includes, added proper namespaces as per our C++
style guidelines, and removed an unnecessary TODO comment.
No operator or macro should be missing description. But if they do, then
they should use NULL pointer, and not an empty string.
This behavior was already enforced (through an assert) for operators,
previous commit made it the same for macros.
Code dealing with macro operators missing description field was slightly
different than the one from Operator registration.
This lead to invalid memory accesses in some python introspection cases
like the i18n messages extraction code in `bl_i18n_utils` module.
The code used `Spline::LookupResult` in a way that referred to evaluated
points and control points interchangeably. That didn't affect the logic,
but the code became harder to read. Instead, introduce a local struct
to contain the data in a more obvious way.
Some uses of delaunay_2d_calc don't need to know the original verts,
edges, and faces that correspond to output elements.
This change adds a "need_ids" value to the CDT input spec, default true,
which tracks the input ids only when true.
The python api mathutils.geometry.delaunay_2d_cdt gets an optional
final bool argument that is the value of need_ids. If the argument
is not supplied, it is true by default, so this won't break old uses
of the API.
On a sample text test, not tracking ids save about 30% of the runtime.
For most inputs the difference will not be so dramatic: it only really
kicks in if there are a lot of holes.
Regression introduced in {rBbfa3dc91b754}.
`ID_RECALC_GEOMETRY` should tag all operations of the `GEOMETRY`
component and not just the operation of node `GEOMETRY_EVAL_INIT`.
This node implements shortening each spline in the curve based on
either a length from the start of each spline, or a factor of the
total length of each spline, similar to the "Start & End Mapping"
panel of curve properties.
For Bezier curves, the first and last control points are adjusted
to maintain the shape of the curve, but NURB splines are currently
implicitly converted to poly splines.
The node is implemented to avoid copying where possible, so it outputs
a changed version of the input curve rather than a new one.
Differential Revision: https://developer.blender.org/D11901
After rB3b6ee8cee708 by @HooglyBoogly vertex groups were moved
to mesh data, and versioning code was provided to upgrade old
files. However, it fails to consider the case of linked duplicates
having different name lists, and dependent on the object order
can cause some of the names to be lost. This can even be all of
them, if there is a duplicate without any names, which can be
easily created by lazy Python code.
To fix this, change the code to use the longest available name list.
Differential Revision: https://developer.blender.org/D11958
Using part of a patch from Erik Abrahamsson, this replaces the
use of linked lists for original id tracking by Sets.
I had thought that the lists were unlikely to grow to more than
a few elements, but when the mesh has a lot of holes (whose
original ids go *outside* the hole, and therefore, most of the
mesh), this assumption can be very wrong.
On a Text regression test, the time went from 11.67s to 0.16s
with this fix. I also tested to make sure that Boolean didn't
slow down with this, and found it actually had a very slight speedup.
Using Sets exposed a dependency on the ordering of the items
in the id lists, luckily caught by a mesh intersect regression test,
so fixed that.
Scaling down images could create images with a width or height of zero.
Clamp at 1 to prevent a crash, also add an assert to scaling functions.
Ref D11956
`split_multicam` used split operator, where if more strips than
multicam were selected, all would be split, which is undesirable.
Add `Sequence.split()` RNA API function. to split individual strips.
Function accepts `frame` and `split_method arguments`. Returns right
strip after splitting.
In case when strip being split have effects, these will be split too, so
no invalid state should be created.
Selection is not handled, this is by design up to user.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11926
This will help enable development on optimizations to the perimeter
calculation here. Using C++ data structures like Array can make the
code easier to read as well.
Longer term, this can help improve integration with attributes
and possibly the new curve code (since strokes and curves are
quite similar in theory).
Differential Revision: https://developer.blender.org/D11941
We need to be more strict about trying to retrieve a list of vertex group
names from objects now, as only three object types support them.
This commit adds a check for vertex group support in a few places, the
data transfer operator/modifier, copying vertex groups to selected
objects, and the vertex group remove and clear functions.
Differential Revision: https://developer.blender.org/D11947
When a vertex group is used to limit the influence of the modifier
to a subset of vertices, binding data for vertices with zero weight
is not needed. This wastes memory, disk space and CPU cycles.
If the vertex group contents is known to be final and constant,
it is reasonable to optimize by only storing data group vertices.
This has to be an option in case the group can change.
Supporting this requires adding a vertex index field and spliting
the vertex count into mesh and bind variants, but both happen to
fit in available padding. The old numverts field is renamed to the
new bound vertex count field to maintain the array length invariant.
Versioning is used to initialize the other new fields.
If a file with sparse binding is opened in an old blender version,
it is corrupted into a non-sparse bind with vertex count mismatch,
preventing the modifier from working until rebind.
Differential Revision: https://developer.blender.org/D11924
`POINT_CACHE_RESET` pointed to `GEOMETRY_EVAL_INIT` while
`GEOMETRY_EVAL_INIT` pointed to `POINT_CACHE_RESET`.
Now `POINT_CACHE_RESET` points to the same nodes pointed to by
`GEOMETRY_EVAL_INIT` thus avoiding the dependency cycle.
The crash happens because `GPU_offscreen_create` is called with `err_out` `NULL`.
This patch proposes a solution within the `GPU_offscreen_create` itself
and raises an error report in the interface if a menu is called with
dimensions beyond what is supported.
Ref T89782
Maniphest Tasks: T89782
Differential Revision: https://developer.blender.org/D11927
The `action_preview_render()` function used to just render, but now it
also temporarily applies the pose. Its comment is now updated for this.
No functional changes.
Replaces current ID Mask node Anti-Aliasing operation by SMAA
operations with default settings as proposed by Jeroen Bakker.
SMAA produces smoother edges.
Reviewed By: manzanilla
Differential Revision: https://developer.blender.org/D11881
In driver editor, vertically flip the value debug lines to align
them with the timeline header values. This makes it easier to read
the values. Also set the line width explicitly, which was incorrect
in some cases.
Differential Revision: https://developer.blender.org/D8877
This move was already prepared with 788d380460 and 26b098c04f. The
template is quite big already, better to give it its own file. Plus it
could use some C++ features like RAII and maybe some more object
oriented code. I plan further refactoring there.
It turns out `BKE_mesh_copy_parameters` can be called while other
tools are running calculations, which meant that it was called at the
same time as `armature_deform_coords_impl`. Beause of that, we
shouldn't do any freeing (of the old vertex group names) there.
Since the materials are copied in the "for_eval" version anyway,
it seems to make sense to copy the vertex group name list there also.
Fixes T89877, and also the failing `deform_modifiers` test.
Differential Revision: https://developer.blender.org/D11936
The new Asset Browser-based pose library is partially implemented in an
add-on. This commit enables the add-on by default, as the old pose
library was built-in and thus always enabled.
The ability to disable the add-on is there mostly for cases where
people/studios want to use their own custom pose library.
Correct cases where the X-axis of the bone (in pose space) aligns with
the pose-space Y or Z-axis. In these cases the decomposition of the
matrix fails, and a negative scale of the X-axis turns into a 180°
rotation around the Y-axis. An extra -1 scale to the X and Z axes of the
resulting matrix seems to fix things.
The asset view UI template is a mini-version of the Asset Browser that
can be placed in regular layouts, regions or popups. At this point it's
made specifically for placement in vertical layouts, it can be made more
flexible in the future.
Generally the way this is implemented will likely change a lot still as
the asset system evolves.
The Pose Library add-on will use the asset view to display pose
libraries in the 3D View sidebar.
References:
* https://developer.blender.org/T86139
* https://code.blender.org/2021/06/asset-browser-project-update/#what-are-we-building
* https://code.blender.org/2021/05/pose-library-v2-0/#use-from-3d-viewport
Notes:
* Important limitation: Due to the early & WIP implementation of the
asset list, all asset views showing the same library will show the
same assets. That is despite the ID type filter option the template
provides. The first asset view created will determine what's visible.
Of course this should be made to work eventually.
* The template supports passing an activate and a drag operator name.
The former is called when an asset is clicked on (e.g. to apply the
asset) the latter when dragging (e.g. to .blend a pose asset). If no
drag operator is set, regular asset drag & drop will be executed.
* The template returns the properties for both operators (see example
below).
* The argument list for using the template is quite long, but we can't
avoid that currently. The UI list design requires that we pass a
number of RNA or custom properties to work with, that for the Pose
Libraries should be registered at the Pose Library add-on level, not
in core Blender.
* Idea is that Python scripts or add-ons that want to use the asset view
can register custom properties, to hold data like the list of assets,
and the active asset index. Maybe that will change in future and we
can manage these internally.
As an example, the pose library add-on uses it like this:
```
activate_op_props, drag_op_props = layout.template_asset_view(
"pose_assets",
workspace,
"active_asset_library",
wm,
"pose_assets",
workspace,
"active_pose_asset_index",
filter_id_types={"filter_action"},
activate_operator="poselib.apply_pose_asset",
drag_operator="poselib.blend_pose_asset",
)
drag_op_props.release_confirm = True
drag_op_props.flipped = wm.poselib_flipped
activate_op_props.flipped = wm.poselib_flipped
```
So far all UI lists had to be defined in Python, this makes it possible
to define them in C as well. Note that there is a whole bunch of special
handling for the Python API that isn't there for C. I think most
importantly custom properties support, which currently can't be added
for C defined UI lists.
The upcoming asset view UI template will use this, which needs to be
defined in C.
Adds a new file `interface_template_list.cc`, which at this point is
mostly a dummy to have a place for the `ED_uilisttypes_ui()` definition.
I plan a separate cleanup to move the UI-list template to that file.
This button type shows a preview image above centered text, similar to
the File Browser files in Thumbnail Display Mode or the default Asset
Browser display. In fact we may want to port these over to use the new
button type at some point.
Will be used by the asset view UI template that will be added in a
following commit. That is basically a mini version of the Asset Browser
that can be displayed elsewhere in the UI.
If a text button is activated that is not in view (i.e. scrolled away),
the scrolling will now be adjusted to have it in view (with some
small additional margin). While entering text, the view may also be
updated should the button move out of view, for whatever reason. For the
most part, this feature shouldn't be needed and won't kick in, except
when a clicked on text button is partially out of view or very close to
the region edge. It's however quite important for the previously
committed feature, that is, pressing Ctrl+F to start searching in a UI
list. The end of the list where the scroll button appears may not be in
view. Plus while filtering the number of visible items changes so the
scrolling has to be updated to keep the search button visible.
Note that I disabled the auto-scrolling for when the text button spawned
an additional popup, like for search-box buttons. That is because
current code assumes the button to have a fixed position while the popup
is open. There is no code to update the popup position together with the
button/scrolling.
I also think that the logic added here could be used in more places,
e.g. for the "ensure file in view" logic the File Browser does.
Adds an operator invoked by default with Ctrl+F that while hovering a UI
list, opens the search field of the list and enables text input for it.
With this commit the search button may actually be out of view after
Ctrl+F still. The following commit adds auto-scroll to solve that.
A downside is that in the Properties, there also is Ctrl+F to start
the editor-wide search. That's not unusual in Blender though (e.g.
scolling with the mouse over a UI list also scrolls the list, not the
region).
Makes it possible to create tooltips for UI list rows, which can be
filled in .py scripts, similar to how they can extend other menus. This
is used by the (to be committed) Pose Library add-on to display pose
operations (selecting bones of a pose, blending a pose, etc).
It's important that the Python scripts check if the UI list is the
correct one by checking the list ID.
For this to work, a new `bpy.context.ui_list` can be checked. For
example, the Pose Library add-on does the following check:
```
def is_pose_asset_view() -> bool:
# Important: Must check context first, or the menu is added for every kind of list.
list = getattr(context, "ui_list", None)
if not list or list.bl_idname != "UI_UL_asset_view" or list.list_id != "pose_assets":
return False
if not context.asset_handle:
return False
return True
```
Add improved arrow key walk navigation in grid layout UI List templates.
Pressing up or down walks the active item to the adjacent row in that
direction, while left and right walk through the items along the columns
wrapping at the rows.
Note from Julian:
In combination with the following commit, this has the important
limitation that the list's custom activate operator won't be called when
"walking over" an item that is scrolled out of the list. That is because
we don't actually create any buttons for those that could be used for
the handling logic. For our purposes of the pose libraries that should
be fine since the asset view list is always made big enough to display
all items. Solving this might be difficult, we don't properly support
nesting boxes with proper scrolling in regular layouts. It's all just
hacked a bit for UI-lists to work. Overlaps quite a bit with T86149.
Differential Revision: https://developer.blender.org/D11063
For pose libraries, we need to be able to apply a pose whenever
activating (clicking) an item in the Pose Library asset view and blend
it by dragging (press & move). And since we want to allow Python scripts
to define what happens at least when activating an asset (so they can
define for example a custom "Apply" operator for preset assets), it
makes sense to just let them pass an operator name to the asset view
template. The template will be introduced in a following commit.
This new layout type is meant for the upcoming asset view UI template.
With it it is possible to show big asset previews with their names in a
responsive grid layout.
Notes:
* The layout is only available for C defined UI lists. We could expose
it to Python, but I think there are still some scrolling issues to be
fixed first. (The asset view template doesn't use scrolling for the UI
list.)
* I'd consider this a more usable version of the existing `GRID` layout
type. We may remove that in favor of the new one in future.
This is more of a first-pass refactor for the UI list template. More
improvements could be done, but that's better done separately. Main
purpose of this is to make the UI list code more manageable and ready
for the asset view template.
No functional changes for users.
* Split the huge template function into more manageable functions, with
clear names and a few structs with high coherency.
* Move runtime data management to the template code, with a free
callback called from BKE. This is UI data and should be managed at
that level.
* Replace boolean arguments with bit-flags (easily extendable and more
readable from the caller).
* Allow passing custom-data to the UI list for callbacks to access.
* Make list grip button for resizing optional.
* Put logic for generating the internal UI list identifier (stored in
.blends) into function. This is a quite important bit and a later
commit adds a related function. Good to have a clear API for this.
* Improve naming, comments, etc.
As part of further cleanups I'd like to move this to an own file.
When generating a preview image for a pose, temporarily apply it to the
armature. Contrary to the usual pose application, this ignores the
selected bones and always applies the entire pose.
Introduce new pose library, based on the Asset Browser. Contrary to the
old pose library (in `editors/armature/pose_lib.c`), which stored an
entire library of poses in an `Action`, in the new library each pose is
its own `Action` datablock. This is done for compatibility with the
asset browser, and also to make it easier to attach preview images,
share datablocks, etc. Furthermore, it opens the door to having
animation snippets in the pose library as well.
This commit contains the C code for the pose library; in order to fully
use it, an addon is required as well (which will be committed shortly).
Add function `BKE_pose_apply_action_blend()`, which blends a given
Action into current pose. The Action is evaluated at a specified frame,
and the result is applied to the armature's pose.
A blend factor can be given to blend between the current pose and the
one in the Action. Quaternions are interpolated with SLERP; it is
assumed that their FCurves are consecutively stored in the Action.
This function will be used in the upcoming new Pose Library.
New function `BKE_pose_apply_action_all_bones()`, which will be
necessary for the upcoming pose library v2.0.
This renames the function `BKE_pose_apply_action` to
`BKE_pose_apply_action_selected_bones`, to reflect that it only works on
selected bones, to contrast it to the new function.
Add operator 'Open Blend File' to the Asset Browser. This operator:
- starts a new Blender process,
- opens the blend file containing the asset,
- monitors the new Blender process, and when it stops,
- reloads the assets to show any changes made.
This is an editor-level abstraction for the `BLO_library_temp_xxx()`
API for temporary loading of data-blocks from another Blend file. It
abstracts away the asset specific code, like asset file-path handling
and local asset data-block handling.
Main use-case for this is applying assets as presets that are based on
data-blocks, like poses. Such preset assets are an important part of the
asset system design, so such an abstraction will likely find more usage
in the future.
Adds the following to `bpy.types.FileSelectEntry`:
* `id_type`: The data-block type the file represenets, if any.
* `local_id`: The local data-block it represents, if any (assets only).
And the following to `bpy.types.AssetHandle`:
* `local_id`: The local data-block the asset represents, if any.
This kind of information and the references are important for asset related
operators and UIs. They will be used by upcoming Pose Library features.
Implements a basic, WIP version of the asset list. This is needed to
give the asset view UI template asset reading and displaying
functionality.
See:
* Asset System: Data Storage, Reading & UI Access - https://developer.blender.org/T88184
Especially the asset list internals should change. It uses the
File/Asset Browser's `FileList` API, which isn't really meant for access
from outside the File Browser. But as explained in T88184, it does a lot
of the stuff we currently need, so we (Sybren Stüvel and I) decided to
go this route for now. Work on a file-list rewrite which integrates well
with the asset system started in the `asset-system-filelist` branch.
Further includes:
* Operator to reload the asset list.
* New `bpy.types.AssetHandle.get_full_library_path()` function, which
gets the full path of the asset via the asset-list.
* Changes to preview loading to prevent the preview loading job to run
eternally for asset views. File Browsers have this issue too, but
should be fixed separately.
It's useful to know where an asset is stored in, before this there was no way
to tell this. This could probably be displayed nicer in the UI but we're
currently unsure how. But at least the information is there now.
With temporary I mean that this is not intended to be part of the
eventual asset system design. For that we are planning to have an
`AssetRepresentation` instead, see T87235. Once the `AssetList` is
implemented (see T88184), that would be the owner of the asset
representations.
However for the upcoming asset system, asset browser, asset view and
pose library commits we need some kind of asset handle to pass around.
That is what this commit introduces.
Idea is a handle to wrap the `FileDirEntry` representing the asset, and
an API to access its data (currently very small, will be extended in
further commits). So the fact that an asset is currently a file
internally is abstracted away. However: We have to expose it as file in
the Python API, because we can't return the asset-handle directly there,
for reasons explained in the code. So the active asset file is exposed
as `bpy.context.asset_file_handle`.
For the Asset Browser, this returns the active asset library of the
Asset Browser, otherwise it returns the one active in the workspace.
This gives simple access to the active asset library from UI code and
Python scripts. For example the upcoming Pose Library add-on uses this,
as well as the upcoming asset view template.
This per-workspace active asset library will be used by the asset views
later. Note that Asset Browsers have their own active asset library,
overriding the one from the workspace.
As part of this the `FileSelectAssetLibraryUID` type gets replaced by
`AssetLibraryReference` which is on the asset level now, not the
File/Asset Browser level. But some more work is needed to complete that,
which is better done in a separate commit.
This also moves the asset library from/to enum-value logic from RNA to
the editor asset level, which will later be used by the asset view.
Update vertex weights between simulation steps if they have changed.
This allows for animated vertex weights in the cloth sim.
Reviewed By: Sebastian Parborg
Differential Revision: http://developer.blender.org/D11640
The `__enter__` function of the `bpy.data.libraries.load` context manager
was storing a pointer to a stack-allocated variable, which was subsequently
used in the `__exit__` function, causing a crash. This is now fixed.
Thanks @Severin for the patch.
If text strips have the same start frame but are stacked on top of each
other in different channels the order in which they are written in the
.rst file was random before.
Reviewed By: Richard Antalik
Differential Revision: https://developer.blender.org/D11903
Mainly naming (also droping the `layer_collection` in favor of just
`layer` for internal code, this is clear enough and much shorter). Add
proper parent/child identifiers, `r_` prefix for parameters also used as
return values, etc.
Also made some parameters const.
The original refactor for vertex groups (3b6ee8cee7)
forgot to bump the minimum file requirement.
I'm also bumping the subversion to 12 so everyone can switch to a
working subversion number.
Differential Revision: https://developer.blender.org/D11931
This makes node trees with long links that cross other nodes easier to work with.
Dimmed links will be ignored by various modal operators like cut and reroute insertion.
Differential Revision: https://developer.blender.org/D11813
We were manually setting the compiler flags
for C++17 support for this previously. CMake
can do this for us in a uniform way without
having to worry about compiler specifics.
Setting these flags manually somehow brought
out some unwanted behaviour (CMake switching
back to C++14) in the nightly CMake builds.
Unsure if that's a CMake bug or planned
new behaviour for future version, but best
to play it safe.
These flags are supported since CMake 3.1
so should not break anything.
Reviewed by: Campbell Barton
Differential Revision: https://developer.blender.org/D11891
This logic is from the curve sundivide node, used to add points with
proper handles in between two existing points. However, the same logic
is used for trimming of Bezier splines, and possibly interactive point
insertion in the future, so it's helpful as a general utility.
The logic is converted to depend on a bezier spline instead of being
static. A temporary segment spline can be used for the latter use case.
This adds id_properties_clear() and id_properties_ensure() functions
to RNA structs. This is meant as an initial change based on discussion
in review of D9697. However, they may be useful in other situations.
The change requires refactoring the internal idproperties callback to
return a pointer to the IDProperty pointer, which actually turns out
to be quite a nice cleanup.
An id_properties attribute could be added in the future potentially.
Differential Revision: https://developer.blender.org/D11908
The general graphing mechanism will create one graph for each output
variable. So it's not limited to time and memory, but that is what the
Cycles tests now output.
The BKE_gpencil_stroke_add_points API function worked well for
creating the primitives in the add object menu, but it expected a
specific data format that doesn't make sense in a dynamic context.
As evidence of that we can see the way source data was duplicated
in the line art file just to use this API function.
This commit solves that problem in two ways:
- Clean up the line art function (this should make it faster too).
- Move/rename the function so its intended use is more clear.
Differential Revision: https://developer.blender.org/D11909
Non-pointer-like properties that are not editable should never generate
override operations.
While harmless (those would never be applied back anyway), better not
clutter override operations list, and also enjoy the symbolic
performances improvement here.
NOTE: Pointer-like properties (pointers and collections) remain
processed as usual here since they usually imply recursivity. We could
make an exception to the exception for ID pointers, but for now I don't
think this is worth it.
Issue would happen when the original, linked data already had 'Disk Cache'
setting enabled. Override would then see no difference with linked data,
and not create any rule for it (as expected).
Root of the issue was that in Cloth modifier copy code, those disk cache
settings were not copied at all, so every time local overrides were
re-generated by copying linked data, those flags would be reset to their
default values.
NOTE: this might exist in other PointCache usages as well, but this code is
in such a bad state that I'd rather do minimal strictly needed changes
there, on a case-by-case basis. Proper recode of that whole system is
wayyyyy out of scope here.
When image strip is added from python using `image_strip_add` operator
and directory path is not terminated with slash, last part of directory
was ignored.
Use `BLI_join_dirfile` instead of simple string concatenation.
Constraining to X axis caused snapping to not work at all. Constraining
to Y axis caused snapping indicator to be drawn, when snapping doesn't
occur.
Reviewed By: mano-wii
Differential Revision: https://developer.blender.org/D11898
nodes or sockets" error
rBfe22635bf664 introduced a utility to check for this (but it was always
returning true).
This wasnt a problem in master (since it is unused there), but in the
2.93 branch, this utility is actually used and the error results in all
geometry nodetrees to appear with the "Node group has unidentified nodes
or sockets" message (and being unusable).
Now return false in has_undefined_nodes_or_sockets if all nodes and
sockets have been successfully checked.
This commit then needs to end up in the 2.93 branch.
Maniphest Tasks: T89851
Differential Revision: https://developer.blender.org/D11911
Socket inspection helps with debugging a geometry node group.
Now, when hovering over a socket, a tooltip will appear that provides
information about the data in the socket. Note, socket inspection only
works for sockets that have been computed already. Nodes that are not
connected to an output are not computed.
Future improvements can include ui changes to make the tooltip look
more like in the original design (T85251). Furthermore, additional
information could be shown if necessary.
Differential Revision: https://developer.blender.org/D11842
Use more direct access to next/previous vertices.
- `BM_edge_other_vert(l_curr->e, l_curr->v)` -> `l_curr->next->v`.
- `BM_edge_other_vert(l_curr->prev->e, l_curr->v)` -> `l_curr->prev->v`.
Add asserts to keep the intention clear.
Adds full frame implementation to this node operations.
No functional changes.
2.5x faster than tiled fallback on average.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11690
Not causing issues in current master because all buffer areas are at
(0, 0) position and `Extend` is not used. But areas may be at any
position in future developments and it will crash.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11784
Link sockets are always connected to inserted translate or scale
operation `Color` sockets even when they have different data type.
This causes crashes on full frame mode when operations read inputs
with non expected datatypes.
Because data type conversions need to be executed before, convert
resolutions must ensure same datatypes are linked.
This commit moves the storage of `bDeformGroup` and the active index
to `Mesh`, `Lattice`, and `bGPdata` instead of `Object`. Utility
functions are added to allow easy access to the vertex groups given
an object or an ID.
As explained in T88951, the list of vertex group names is currently
stored separately per object, even though vertex group data is stored
on the geometry. This tends to complicate code and cause bugs,
especially as geometry is created procedurally and tied less closely
to an object.
The "Copy Vertex Groups to Linked" operator is removed, since they
are stored on the geometry anyway.
This patch leaves the object-level python API for vertex groups in
place. Creating a geometry-level RNA API can be a separate step;
the changes in this commit are invasive enough as it is.
Note that opening a file saved in 3.0 in an earlier version means
the vertex groups will not be available.
Differential Revision: https://developer.blender.org/D11689
When adding texts or various simple effects I often copy-paste strips
to reuse properties from a template such as font or position. I assume
this is common workflow. Issue with this workflow is, that active strip
is not changed after pasting, so when adjusting property, it is original
strip that is being modified.
This is not issue when duplicating strips - selection state is
transfered to duplicate strips, such that duplicate of active strip is
set to be active and duplicate of selected strip is set to selected.
Implement same selection transfering behavior in paste operator, that
exists in duplicate operator.
Since strip can be deleted after copying, it is not possible to rely
on sequencer state. This is true even when pasting strips to different
scene. Therefore active strip name must be stored in clipboard.
Reviewed By: sergey, Severin
Differential Revision: https://developer.blender.org/D11781
Caused by fba9cd019f, then fixed by 0e4245bc28, but without
subversion bump, so some files were still broken after fix.
Repeat fix again, but this time also bump subversion.
Reviewed By: mont29
Differential Revision: https://developer.blender.org/D11864
Use the same partial-update functions used by transform when
editing vertex locations with the number buttons.
This avoids unnecessary calculations for normals and tessellation.
This gives around 1.44x overall speedup on high poly meshes.
Support for begin/update/end callbacks allowing state to be cached
and reused while dragging a number button or slider.
This is done using `UI_block_interaction_set` to set callbacks.
- Dragging multiple buttons at once is supported,
passing multiple unique events into the update function.
- Update is only called once even when multiple buttons are edited.
- The update callback can detect the difference between click & drag
actions so situations to support skipping cache creation and
freeing for situations where it's not beneficial.
Reviewed by: Severin, HooglyBoogly
Ref D11861
In most cases the undo system was loading undo steps twice.
This was needed since some undo systems (sculpt, paint, text)
require stepping out of the current undo step.
Use a flag to limit this to the undo systems that need it.
This improves performance for other undo systems.
This gives around 1.96x speedup in edit-mesh for high-poly objects.
Reviewed by: mont29
Ref D11893
- Tag the object data instead of the object when decoding
(this avoids duplicating mesh object-data on each undo-step).
- Calculate face normals as part of multi-threaded tessellation.
This gives ~11% speedup with 1.5x million polygons.
Rename:
- EDBM_mesh_free -> EDBM_mesh_free_data
BKE_editmesh_free -> BKE_editmesh_free_data
Since this doesn't free the edit-mesh pointer.
- BKE_editmesh_free_derivedmesh -> BKE_editmesh_free_derived_caches
Since this no longer uses derived-mesh, match naming for the related
object function BKE_object_free_derived_caches.
Also remove `do_tessellate` argument from BKE_editmesh_create,
since the caller can explicitly do this if it's needed,
with the advantage that it can be combined with normal calculation
which is faster on high-poly meshes.
All platforms support the proper format string and it's already used in
various other places.
Differential Revision: https://developer.blender.org/D11460
During a mesh transformation in edit mode (Move, Rotate...), only part of
the batch cache needs to be updated.
This commit allows only update only the drawn batches seen in
`BKE_object_data_eval_batch_cache_deform_tag` if the new
`ID_RECALC_GEOMETRY_DEFORM` flag is used.
This new flag is used in the transforms operation for edit-mesh and
results in 1.6x overall speedup in heavy subdiv cube.
Differential Revision: https://developer.blender.org/D11599
This might be an artistic choice, but round end caps are supposed to be
a "half circle centered at the end point of the line" as documented
here: https://docs.blender.org/manual/en/dev/render/freestyle/
parameter_editor/line_style/strokes.html#caps
They are a shashed half circle instead.
This patch makes this pure half circles [and also fixes the case where
thickness of beginning was used for both beginning and end of the
stroke]
Maniphest Tasks: T88015
Differential Revision: https://developer.blender.org/D11340
This commit adds a curve primitive node for creating squares,
rectangles, trapezoids, kites, and parallelograms. It also includes
a mode where the four points are just vector inputs.
Differential Revision: https://developer.blender.org/D11665
In this bug report it resulted in rendering animations stopping too early,
but this affected more areas.
After the previous cleanup commit, it becomes clear that frame and ctime
values were mixed up.
Confusingly, BKE_scene_frame_get did not match the frame number as expected by
BKE_scene_frame_set. Instead it return the value after time remapping, which
is commonly named "ctime".
* Rename BKE_scene_frame_get to BKE_scene_ctime_get
* Add a new BKE_scene_frame_get that matches BKE_scene_frame_set
* Use int/float depending if fractional frame is expected
It always bothered me that we'd do the `BLI_assert(... || !"message")` trick to
print a message alongside the assert, while it should be trivial to have a way
to pass an extra string as additional argument.
This adds `BLI_assert_msg()` with a second argument for a message. E.g.:
```
BLI_assert_msg(
params->rename_id == NULL,
"File rename handling should immediately clear rename_id when done, because otherwise it will keep taking precedence over renamefile.");
```
On failure this will print like this:
```
0 Blender 0x00000001140647a3 BLI_system_backtrace + 291
[...]
13 Blender 0x00000001092647a6 main + 3814
14 libdyld.dylib 0x00007fff203d8f5d start + 1
BLI_assert failed: source/blender/editors/space_file/file_ops.c:2352, file_directory_new_exec(), at 'params->rename_id == ((void*)0)'
File rename handling should immediately clear rename_id when done, because otherwise it will keep taking precedence over renamefile.
```
Reviewed by: Sybren Stüvel, Jacques Lucke, Sergey Sharybin, Campbell Barton
Differential Revision: https://developer.blender.org/D11827
The 'collection' property is flagged PROP_ID_REFCOUNT, so the
modifiers foreachIDLink functions should walk with IDWALK_CB_USER
(instead of IDWALK_CB_NOP).
Otherwise the modifier wont be included as a user for the collection
(e.g. on file read); removing the collection from the modifier will
decrement usercount though (which in worst case scenario makes the
collection orphan and will result in data loss)
Maniphest Tasks: T89765
Differential Revision: https://developer.blender.org/D11877
When editing vertices with number buttons, tag the mesh, not the object.
This prevents the evaluated mesh being re-created for the object
and is correct as the mesh is being edited not the object.
Note that all tags for updating object geometry should be checked
to see if this change should be applied there too.
From a simple test on a high-poly mesh this gives around 1.3x
overall speedup.
After cache implementation line art apply will not show strokes properly, now fixed.
# Conflicts:
# source/blender/gpencil_modifiers/intern/MOD_gpencillineart.c
The Pattern and Search display options in the Clip Editor display settings
are independent and should not be grayed out since those options
remain relevant even with path display turned off.
Alternative solution were propoesed in D11630 and D11715
but each of those patches had downsides.
This solution is the simplest and does not break muscle memory.
Channel packed images should not have their RGB affected by alpha.
rendering in Cycles and Eevee was fine already, but displaying these was
not right in the Image and Node editors.
Not 100% sure what to do for the "Color and Alpha" mode, but I guess
this should stay like it was before (applying the alpha).
"Color", "R", "G", and "B" modes were changed to not have color be
affected by alpha though.
ref. T89034
Maniphest Tasks: T89034
Differential Revision: https://developer.blender.org/D11871
Fly navigation has always had this option. They is particularly useful
when users use "Trackball" as their orbit method.
For walk navigation this works as a one off option. Not as a toggle like
for fly navigation.
Differential Revision: https://developer.blender.org/D11863
Correctly reset `prev` and `next` pointers of action group FCurves when
separating them into distinct `ListBase`s per `bActionGroup`.
These `NULL` pointers are necessary to temporarily demarcate the start &
end of the `bActionGroup::channels` list. Having them still point to
other FCurves caused ordering issues when moving curves towards the
start/end of a group.
This commit corrects the above issue and adds versioning code to rectify
any ordering issues that may have been caused. For this purpose the
`BKE_action_groups_reconstruct()` function is rewritten to avoid relying
on the `bAction::curves` list order or `prev` link integrity.
Differential Revision: https://developer.blender.org/D11811
Simple upgrade of OpenXR to 1.0.17. A version bump was enough, no
Blender code had to change.
Reviewed By: LazyDodo, mont29
Differential Revision: https://developer.blender.org/D11848
It was possible that render buffers and scene kernel data will be out
of sync because reset and scene update happens in different locks.
This is similar issue we've fixed in the Cycles X branch, so backported
relevant changes from there.
This change removes what seems to be unused feature kernel.
Differential Revision: https://developer.blender.org/D11828
This change makes the behavior consistent between shortcut and
option from space clip's header.
The only caveat is that the "Lock to Selection" is removed from the
Display popover. This is because it is rather hard to make operator
to render same as regular checkbox. However, shouldn't be a problem
because the setting in popover was redundant.
Differential Revision: https://developer.blender.org/D10423
The node group interface panels were still implemented in C.
Now they ported over to python for easier maintenance.
Differential Revision: https://developer.blender.org/D11834
In addition to _object_ animation, now _object data_ (mesh, curve, ...)
animation can now be made single user as well.
This came up in T89369 and while it is possible to do this via the
Outliner [where all actions have to be selected individually], this
seems to be more convenient to be done from the 3DView.
note: usercount of the action is checked now, if it single-user already,
no copy takes place (same thing could/should be done for
single_object_action_users as well).
note2: obdata is made single user as well (otherwise duplicated actions
will be assigned to the same shared obdata - which does not result in
unique animadata which is what we are after here)
ref. T89369
Maniphest Tasks: T89369
Differential Revision: https://developer.blender.org/D11683
Previously, this did not compile in VS 2017, because
`new T(initializer_())` would try to call the copy constructor of `T`.
Now, `initializer_` will construct the `T` inplace.
Renaming compositor node in search panel "View Switch" to "Switch View"
for better consistency.
Reviewed By: dfelinto
Differential Revision: https://developer.blender.org/D11717
Using the "Camera Fit Frame to Selected" operator didn't work for Grease Pencil
objects. The same issue caused grease pencil preview thumbnails to be useless
(e.g. when using "Mark Asset" on a Grease Pencil object).
Reason was that there was no logic to handle grease pencil data and its strokes
for the object display-point iterators used for the "Frame Selected" logic.
Addresses T89656.
Reviewed by: Antonio Vazquez, Campbell Barton
Differential Revision: https://developer.blender.org/D11833
The original assumption that the `modifyMesh` function is only
called when the modifier is applied was wrong. There are still a
couple of other places calling it through `BKE_modifier_modify_mesh`.
Now there is an extra check that makes sure instances are only
realized when the modifier is actually applied.
The backdrop image gizmo was not following the backdrop image, it
needs to be refreshed whenever the view changes. The region init
callback is executed whenever the region size changes, so that should be
a reliable place to do that.
Reported as part of T87591.
In a collapsed hierarchy, some type of data are listed as one icon per
item (2.79 did this for all data),
others are 'compacted' as a single icon with a counter (also indicating
if the active item is down that collapsed hierarchy) from 2.80 on.
Not quite sure if {rB92dfc8f2673e} was meant to do this (if it was, this
was not working though because relevant code would only get executed for
object hierarchies it seems), so now this is done for bones as well.
Fixes T88413.
Maniphest Tasks: T88413
Differential Revision: https://developer.blender.org/D11404
Walk rotate speed for tablets was being modified at runtime by scaling
the user preference mouse_speed by a constant factor; this consolidates
scaling into the the compile time tablet scale factor in walkApply.
No change in behavior.
When renaming an ID somewhere in the UI after marking it as asset, it would
often get lost in the Asset Browser (scrolled out of view). It would also get
deactivated.
This patch makes sure that if an asset is active whose ID gets renamed, it is
kept active and visible. That is important for a fast, uninterrupted asset
creation workflow, where users often rename assets while working in the asset
browser.
Old code stored the new file-name to identify a file after re-reading the
file-list after the rename. For assets that doesn't work because there may be
multiple assets with the same name. Here the simple solution of just storing
the pointer to the renamed ID is chosen, rather than relying on the file-name
in this case. (Should be fine with undo, since the ID * reference is short
lived, it's not stored over possible undo steps. If it turns out to have
issues, I rather switch to a rename_id_uuid, but keep that separate from the
file->uid).
Reviewed by: Sybren Stüvel
Differential Revision: https://developer.blender.org/D11119
TBB includes in windows.h which will by
default define min/max macro's by default,
which collide with stl's min/mac functions.
this change instructs windows.h not to
add the offending macros
(Note: This is an alternative version for D9994 by @Schiette. The commit
message is based on his description.)
Currently, when a new directory is created it is not selected.
Similarly, when renaming an existing file or directory, it does not
remain active/highlighted blue after renaming.
This change makes sure the file or directory is always selected after
renaming, even if the renaming failed or was cancelled.
This has some usability advantages:
- Open the newly created directory without having to select it (ENTER).
- If you make a naming mistake, you can immediately fix that (F2)
without having to click it again.
- If you create a directory and forget to name it, you can fix that
(F2) without having to select it.
- This is consistent with many common File Browsers.
Further, selecting the item even after renaming failed or was cancelled
helps keeping the file in focus, so the user doesn't have to look for it
(especially if the renaming just failed which the user may not notice).
In other words, it avoids disorienting the user.
Also see D11119 which requires this behavior.
We could also always select the file/directory on mouse press. This
would make some hacks unnecessary, but may have further implications. I
think eventually that's what we should do though.
`filesel.c` seems like the place that should contain file selection
functions. Previously it was in `file_ops.c` because that was the only
file that actually used it. But a followup commit needs it from a
different file.
- Added create_expected_object if it doesn't exist.
- Added a counter condition for recusrion.
- Added optional test_name in basemesh test.
- Added code of previous MeshTest to SpecMeshTest.
- Changed default value of apply_modifier(s) flag and do_compare flag.
Many ui features for geometry nodes need access to information generated
during evaluation:
* Node warnings.
* Attribute search.
* Viewer node.
* Socket inspection (not in master yet).
The way we logged the required information before had some disadvantages:
* Viewer node used a completely separate system from node warnings and
attribute search.
* Most of the context of logged information is lost when e.g. the same node
group is used multiple times.
* A global lock was needed every time something is logged.
This new implementation solves these problems:
* All four mentioned ui features use the same underlying logging system.
* All context information for logged values is kept intact.
* Every thread has its own local logger. The logged informatiton is combined
in the end.
Differential Revision: https://developer.blender.org/D11785
The geometry nodes modifier uses id properties for the inputs to node groups.
That is because the set of properties changes depending on which geometry
node group is selected.
The animation was not updated correctly because the `ANIMATION_EVAL`
depsgraph node was not evaluated, because nothing depended on it in the
depsgraph. This patch makes sure that the proper link to the geometry
component is inserted.
Differential Revision: https://developer.blender.org/D11831
This issue happened because of some miscommunication during the original
patch review. Since we have the parent element in other outliner modes
(Blender File, Data API) I was on the fence here. Rolling this back now
so that when Show All View Layers is off we don't see the current View
Layer top element.
The fix is simple, but it was better to format the code around to
follow the style in this file.
Differential Revision: https://developer.blender.org/D11830
Rename the mesh circle to "Mesh Circle", mesh line to "Mesh Line",
and mesh subdivide to "Mesh Subdivide". Previously they looked exactly
the same in the search menu, and the nodes themselves had the same
label. This is a "deep" rename that also renames internal defines and
function names to match the UI.
This node is quite similar to the curve to points node, but creates
points for only the start and end of each spline. This is a separate
node because the sampling from the curve to points node don't apply,
and just for ease of use.
All attributes from the curves are copied, including the data for
instancing: tangents, normals, and the derived rotations. One simple
use case is to make round caps on curves by instancinghalves of a
sphere on each end of the splines.
Differential Revision: https://developer.blender.org/D11719
DNA parsing assumed any identifier which starts with
(`struct`, `unsigned`, `const`) was that identifier.
So a struct called `constTest foo;` would be parsed as `est foo;`.
Add utility function to check identifiers are not
part of a larger identifier.
This also supports skipping these identifiers in any order.
Reviewed By: LazyDodo
Ref D11837
The curve resample node neglected to copy attributes to single point
result splines. That could have caused errors if some of the splines
in the result had only one point but others had more.
Use "Snap Playhead to Strips" option to enable playhead snapping.
Change behavior of CTRL key to invert snapping similar to transform operator.
Currently this option is disabled by default. It makes editing quite unpleasant
for me personally, but ideally I should gather feedback from more users.
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D11745
This field was used to reference "parent" meta strips in `sequences_all`
RNA collection iterator functions.
Use `SeqIterator` wrapped in `BLI_Iterator` to iterate over elements.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11793
Work scheduler needed initialization and execution models are
not created during constant folding. This moves work execution
method to execution system.
This patch horizontally centers the text inside the progress bar
widget. It is currently left-aligned and offset to the middle, which
doesn't center properly.
see D11205 for details and examples.
Differential Revision: https://developer.blender.org/D11205
Reviewed by Julian Eisel
The menu lists all socket types that are valid for the node tree.
Changing a socket type updates all instances of the group and keeps
existing links to the socket.
If changing the socket type leads to incorrect node connections the
links are flagged as invalid (red) and ignored but not removed. This is
so users don't lose information and can then fix resulting issues.
For example: Changing a Color socket to a Shader socket can cause an
invalid Shader-to-Color connection.
Implementation details:
The new `NODE_OT_tree_socket_change_type` operator uses the generic
`rna_node_socket_type_itemf` function to list all eligible socket types.
It uses the tree type's `valid_socket_type` callback to test for valid
types. In addition it also checks the subtype, because multiple RNA
types are registered for the same base type. The `valid_socket_type`
callback has been modified slightly to accept full socket types instead
of just the base type enum, so that custom (python) socket types can be
used by this operator.
The `nodeModifySocketType` function is now called when group nodes
encounter a socket type mismatch, instead of replacing the socket
entirely. This ensures that links are kept to/from group nodes as well
as group input/output nodes. The `nodeModifySocketType` function now
also takes a full `bNodeSocketType` instead of just the base and subtype
enum (a shortcut `nodeModifySocketTypeStatic` exists for when only
static types are used).
Differential Revision: https://developer.blender.org/D10912
The wayland support requires the following development packages:
libwayland-dev, wayland-protocols, libegl-dev, libxkbcommon-dev,
libdbus-1-dev, linux-libc-dev
Partially revert 7fc220517f, as it
introduced two issues:
- Deleting keys without active keying set was no longer possible, and
- there was no more confirmation popup.
Pressing {key Alt I} in the 3D Viewport now executes
`ANIM_OT_keyframe_delete_v3d`, adjusted to suit both T88068 and T89592:
- If there is an active keying set, delete keys according to that keying
set.
- Otherwise, behave as `ANIM_OT_keyframe_delete_v3d` did before, that
is, delete all keyframes of the selected object and in pose-mode also
of selected bones.
Remove duplicate code from the `ANIM_OT_keyframe_delete` operator. The
actions of the removed code are already performed by the preceding
`keyingset_get_from_op_with_error()` call.
No functional changes.
Graphs are usually large, needing a lot of horizontal scrolling and
they can include more information for debugging.
This patch makes graph more compact horizontally by splitting
labels in lines and removing namespaces.
Furthermore it adds following information:
- Operation ID.
- SetValueOperation float value.
- Optionally, operation node name.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11720
Currently there is no clear way to know if an operation is constant
(i.e. when all rendered pixels have same values). Operations may
need to get constant input values before rendering to determine
their resolution or areas of interest. This is the case of scale, rotate
and translate operations. Only "set operations" are known as
constant but many more are constant when all their inputs are so.
Such cases can be optimized by only rendering one pixel.
Current solution for tiled implementation is to get first pixel
from input. This works for root execution groups, others
need previous groups to be rendered.
On full frame implementation this is not possible, because buffers
are created on rendering to reduce peak memory and there is
no per pixel calls.
This patch evaluates all operations that are constant into primitive
operations (Value/Vector/Color) before determining resolutions.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11490
ld: warning: could not create compact unwind for _ff_rl_init_vlc: stack
subq instruction is too different from dwarf stack size
Similar to rB2de5de57c58521862e0fecc95fc474ea347b7468
Differential Revision: https://developer.blender.org/D11796
Note that this name is essentially never used anywhere, besides as 'information'
mostly accessible from python console. Those embedded IDs are not in Main, so they
are not accessible by name ever, and mostly unusable from animation perspective
(either drivers or fcurves).
Therefore, no breakage is expected in user scripts or addons, nor when
loading in older versions of Blender.
Reviewed By: dfelinto, brecht
Differential Revision: https://developer.blender.org/D11812
During initialization of the platform a debug message is generated and
interpreted by de callback. Here the platform is checked what requires
an initialized platform.
Fixed by giving the platform check less priority in the check.
This option will determine visibility on either render or the viewport
visibility. Same for modifer settings. So it will either evaluate the
depsgrah with DAG_EVAL_RENDER or DAG_EVAL_VIEWPORT.
This not only makes it more flexible, it is also a lot
clearer which visibility / modfier setting is taken into account (up
until now, this was always considered to be DAG_EVAL_RENDER)
This option was always present in the USD exporter, this just brings
Alembic in line with that.
ref. T89594
Maniphest Tasks: T89594
Differential Revision: https://developer.blender.org/D11820
When introduced in {rB61050f75b13e} this was actually working (meaning
it checked the Outliner OB_RESTRICT_RENDER flag and skipped the object if
desired).
Behavior has since then been commented in rBae6e9401abb7 and apparently
refactored out in rB2917df21adc8.
If checked, it seemed to be working (objects marked non-renderable in
the Outliner were pruned from the export), however unchecking that
option did not include them in the export.
Now it changed - for the worse if you like - in rBa95f86359673 which
made it so if "Renderable Objects" only is checked, it will still export
objects invisible in renders. So since we now have the non-functional
option with a broken/misleading default, it is better to just remove it
entirely.
In fact it has been superseeded by the "Visible Objects" option (this
does the same thing: depsgraph is evaluated in render mode) and as a
second step (and to make this even clearer) a choice whether
Render or Viewport evaluation is used can be added (just like the USD
exporter has). When that choice is explicit, it's also clear which
visibility actually matters.
This is breaking API usage, should be in release notes.
ref. T89594
Maniphest Tasks: T89594
Differential Revision: https://developer.blender.org/D11808
From patch D11780 from Erik Abrahamsson.
It parallelizes making the vertices, destruction of map entries,
finding if the result is PWN, finding triangle adjacencies,
and finding the ambient cell.
The latter needs a parallel_reduce from tbb, so added one into
BLI_task.hh so that if WITH_TBB is false, the code will still work.
On Erik's 6-core machine, the elapsed time went from 17.5s to 11.8s
(33% faster) on an intersection of two spheres with 3.1M faces.
On Howard's 24-core machine, the elapsed time went from 18.7s to 10.8s
for the same test.
Adds full frame implementation to this node operations.
No functional changes.
1.3x faster than tiled fallback.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11764
Adds full frame implementation to this node operation.
No functional changes.
1.4x faster than tiled fallback.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11765
Adds full frame implementation to this node operation.
No functional changes.
1.7x faster than tiled fallback.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11766
Adds full frame implementation to this node operation.
No functional changes.
1.5x faster than tiled fallback.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D11767
Also replace integer with bool in Ghost API when only used as boolean,
and uint8* with char* in Ghost API when variable is a string.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D11617
Signed-off-by: Nicholas Rishel <rishel.nick@gmail.com>
This node creates a poly spline line in one of 2 modes:
- Line between two points
- Start Point, Direction, and Length
Both modes create splines with only start and endpoints.
A resample node can be used afterward to increase the point count.
Differential Revision: https://developer.blender.org/D11769
This patch adds a very simple node that explicitly converts a float to
an int. While this may seem redundant, it would offer 2 benefits to the
current requirement to use implicit float conversions:
1. It makes the node tree's intent more clear and self-documenting
(especially if changes in the future require integer inputs).
2. It eliminates undefined behavior in current/future nodes from float
inputs by guaranteeing that the input is an integer.
The node offers a variety of rounding techniques to make it more flexible.
Differential Revision: https://developer.blender.org/D11700
This would prevent loading a parent that would only be referenced by
children during a linking operation.
Looks like this missing bit of code has been there since the stone ages,
it is fairly baffling to find that such critical low-levels mistakes can
survive decades in a codebase...
Note that such fully-indirectly linked parent object is not instantiated
in scene currently, this is fairly bad I think, but kind of a different
issue.
This patch extends D11695 to provide full support for Chinese and Korean
input on macOS.
Chinese input notes:
You can input symbolic characters (such as '! , '$') during Chinese input.
The difference from Japanese input is that multiple `insertText` may be
called with a single key down (`keyDown` method). This happens when you input
a symbolic character (such as '! , '$') during conversion.
The conversion is confirmed (`insertText`) and the symbolic character is
entered (`insertText`). To solve this problem, I have `result_text` to
concatenate the strings and store them in `result`.
Korean input notes:
Korean does not display a conversion suggestion window.
Like Chinese, Korean input may call multiple `insertText` methods. Also,
in Korean, the previous confirmation (`setMarkedText` and `insertText`) and
the next conversion is processed (`setMarkedText`) may be called
simultaneously with a single key down (`keyDown` method).
For example:
1. press g ㅎ (`setMarkedText`)
2. press k 하 (`setMarkedText`)
3. press t 앗 (`setMarkedText`)
4. press k 하세 (`setMarkedText`, `insertText`, `setMarkedText`)
Fixed so that the `insertText` and the last `setMarkedText` are processed.
Also, if a control character (such as Arrow, Enter) is input during Korean
input, the conversion will be confirmed (`setMarkedText`, `insertText`) and
the original control character will be processed.
In other words, if you press the left arrow key while typing in Korean, the
cursor will move to the left after the character is confirmed. Therefore, I
modified the `keyDown` method so that the `handleKeyEvent` is called again
after the `insertText` is processed in the `interpretKeyEvents` method.
Differential Revision: https://developer.blender.org/D11699
Blender did not support to input East Asian characters (Chinese, Japanese,
Korean) on macOS. This patch adds support for Japanese input, by implementing
the appropriate processing for the NSTextInputClient protocol.
Technical notes:
* The conversion candidate window is drawn by the input method program calling
`firstRectForCharacterRange`.
* The string before confirmation (called `composite` in blender) is handled in
the `setMarkedText` method called by the input method program.
* The string after confirmation (called `result` in the blender) is processed
in the `insertText` method called by the input method program.
Ref T51283
Differential Revision: https://developer.blender.org/D11695
IME conversion candidate window was displayed at the mouse position, instead of
below the cursor or text selection.
Blender need to tell the input method program where the conversion candidate
window is during Japanese and Chinese input.
In macOS, the `firstRectforCharacterRange` is called when input by the input
method starts, and the position of the conversion candidate window is
specified. Therefore, it is necessary to set the position of the conversion
candidate window before input starts. This patch changes it so that the position
of the conversion candidate window is always set when the cursor is drawn.
Differential Revision: https://developer.blender.org/D11697
File Browser code uses the term "params" for its file selection parameters a
lot. Avoid confusion/ambiguity by calling the notifier listener parameters
"listener_params".
Went back to using Blender's polyfill for triangulation, which is much
faster (time for a 3.1M face boolean went from 103s to 48s).
Had to put in detection for the case that needs the exact triangulator
(bug T86805), and also a fix for non-convex quads (bug T89330).
Blender needs to tell the input method program where the conversion
candidate window is during Japanese and Chinese input.
In macOS, there are displays where the window size and the native pixel size
are different, so the candidate window may appear in an unnatural position.
This patch converts the cursor position x and y for matching macOS window
coordinate. On Windows, GHOST_GetNativePixelSize returns 1, so it has no effect.
Differential Revision: https://developer.blender.org/D11696
When inserting text using IME on a button, the character after the cursor is
displayed before the cursor.
This bug seems to have occurred during the refactoring in D765.
Differential Revision: https://developer.blender.org/D11072
This allows converting between different `destruct_ptr` types (which is
just a `std::unique_ptr` with a custom deleter).
The most common use case is to convert from a derived type to
the type of the base class.
These are scripts for benchmarking Blender features on real-world .blend
files. They were originally written for benchmarking Cycles performance, and
were made generic so they can be used for more Blender features.
The benchmarks can be run locally by developers. But the plan is to also run
these as part of continuous integration to track performance over time.
Currently there are tests for Cycles rendering and .blend file loading.
Documentation:
https://wiki.blender.org/wiki/Tools/Tests/Performance
Main features:
* User created configurations to quickly run, re-run and analyze a selected
subset of tests.
* Supports both benchmarking with existing builds, and automatic building of
specified git commits, tags and branches.
* Generate HTML page with bar and line graphs from test results.
* Controlled using simple command line tool.
* For writing tests, convenient abstraction to run a Python function in Blender
with arguments and return value.
Ref T74730
Differential Revision: https://developer.blender.org/D11662
This adds a viewer node similar to the one in the compositor.
The icon in the headers of nodes is removed because it served
the same purpose and is not necessary anymore.
Node outputs can be connected to the active viewer using
ctrl+shift+LMB, just like in the compositor. Right now this collides
with the shortcut used in the node wrangler addon, which will
be changed separately.
As of now, the viewed geometry is only visible in the spreadsheet.
Viewport visualization will be added separately.
There are a couple of benefits of using a viewer node compared
to the old approach with the icon in the node header:
* Better support for nodes that have more than one geometry output.
* It's more consistent with the compositor.
* If attributes become decoupled from geometry in the future,
the viewer can have a separate input for the attribute to visualize.
* The viewer node could potentially have visualization settings.
* Allows to keep "visualization points" around by having multiple
viewer nodes.
* Less visual clutter in node headers.
Differential Revision: https://developer.blender.org/D11470
Using BMesh operators through the edit-mesh API created a full copy
of the mesh so it was possible to restore the mesh in case
one of the operators raised an error.
Remove support for automatic backup/restore from the EDBM_op_* API's
as it adds significant overhead and was rarely used.
Operators that need this can use the BMBackup API to backup & restore
the mesh in case of failure.
Add warning levels to BMO_error_raise so operators can report problems
without it being interpreted as a request to cancel the operation.
For high-poly meshes creating and freeing a full copy is an expensive
operation, removing this gives a speedup of ~1.77x for most operators
except for "connect_verts" / "connect_vert_pair"
which still uses this functionality.
Previously, any face groups that could not be merged into a face
caused the entire operation to report an error and do nothing.
Now these cases are skipped over, dissolving faces where possible.
- Only create arrays with groups of two or more faces.
- Remove raising exception for zero length arrays.
- Remove redundant exception check (assert there is no exception).
- Use a struct for face array & it's length instead of a NULL
terminated array (removes the need to count faces in a loop).
- Replace '[mce]' with "Mike Erwin".
- Remove references to turn-table author as it isn't useful information,
the author was credited in the commit message.
Now, The weight is used in the randomize parameters of the offset modifier.
This is useful to generate effects like explosions.
Related to the new Vertex Weight modifiers.
A point of confusion about this node is that it doesn't work on the
output of the mesh circle primitive node. This patch adds a warning to
help with that. This avoids adding a warning when the geometry set
input has no mesh.
Differential Revision: https://developer.blender.org/D11771
As noted in a comment now, these functions only update a cache, so they
don't change the logical state of the mesh, which is "it will have the
data when necessary." Using a const argument will help const correctness
when accessing an object's evaluated mesh.
This code was written for the File Browser together with the Asset Engine
design, that is not part of the Asset Browser/System design anymore. Updated
comments accordingly.
`FileDirEntryRevision` was actually used, but I removed it and moved the used
members to the parent `FileDirEntry`, since there is no concept of revisions
currently.
There should be no functional changes.
Add a new transformation space choice for bone constraints, which
represent the local transformation of the target bone in the constraint
owner's local space.
The use case for this is transferring the local (i.e. excluding the
effect of parents) motion of one bone to another one, while ignoring
the difference between their rest pose orientations.
The new option replaces the following setup:
* A `child` bone of the `target`, rotated the same as `owner` in rest pose.
* A `sibling` bone of the `target`, positioned same as `child` in rest
pose and using Copy Transforms in World Space from `child`.
* The `owner` bone constraint uses Local Space of `sibling`.
(This analogy applies provided both bones use Local Location)
Differential Revision: https://developer.blender.org/D9493
This constraint can be naturally viewed as a prototype for a future
4x4 matrix math node (or subset thereof), since its basic semantics
already is matrix assignment. Thus it makes sense to add math options
to this constraint to increase flexibility in the meantime.
This patch adds support for several operations that would be useful:
- An option to remove shear in the incoming target matrix.
Shear is known to cause issues for various mathematical operations,
so an option to remove it at key points is useful.
Constraints based on Euler like Copy Rotation and Limit Rotation
already have always enabled shear removal built in, because their
math doesn't work correctly with shear.
In the future node system shear removal would be a separate node
(and currently Limit Rotation can be used as a Remove Shear constraint).
However removing shear from the result of the target space conversion
before mixing (similar to Copy Rotation) has to be built into
Copy Transforms itself as an option.
- More ways to combine the target and owner matrices.
Similar to multiple Inherit Scale modes for parenting, there are
multiple ways one may want to combine matrices based on context.
This implements 3 variants for each of the Before/After modes
(one of them already existing).
- Full implements regular matrix multiplication as the most basic
option. The downside is the risk of creating shear.
- Aligned emulates the 'anti-shear' Aligned Inherit Scale mode,
and basically uses Full for location, and Split for rotation/scale.
(This choice already existed.)
- Split Channels combines location, rotation and scale separately.
Looking at D7547 there is demand for Split Channels in some cases,
so I think it makes sense to include it in Copy Transforms too, so that
the Mix menu items can be identical for it and the Action constraint.
Differential Revision: https://developer.blender.org/D9469
To some degree these are changes in preparation of further Asset Browser
related changes, see D11119. But also, the current UUID design was written for
the old Asset Engine design, which isn't part of the current Asset
Browser/System design anymore.
And lastly, "UUID" are a well established standard
(https://en.wikipedia.org/wiki/Universally_unique_identifier) which this
implementation didn't follow. What we have here is more of an index, or a
unique identifier (https://en.wikipedia.org/wiki/Unique_identifier).
So this does the following changes:
* Renames "UUID" to "UID"
* Changes the type of the UID to (a typedef'ed) `uint32_t`, which is more than
enough for our current asset system design and simplifies things.
* Due to the new type, we can avoid allocations for hash-table storage.
* Add/use functions for UID handling
Note that I am working on a major rewrite of the file-list code. Meanwhile we
want to keep things sensible.
his new modifier allows to generate weights base on:
* Angle of the stroke relative to object or world orientation. For example, if the value is 90, the maximum weights will be for vertical lines and minimum for horizontal lines.
* Distance to Target object. The distance calculated is normalized to get valid weights between 0 and 1.0.
The weights are created in an existing vertex group and the data can be replaced or mixed with the existing value to combine different weight effects. The minimum parameter, allows to define the minimum weight generated. This is useful to avoid very low weights.
The generated weights can be used in any modifier. For example, the angle weight value can be used to mimic FreeStyle Caligraphy modifier using the weight with the thickness modifier.
Also some modifier has been changed to inlude a new option to use the weights as factor of the effect.
As result of this change, the fading option has been removed from Thickness and Opacity modifiers because this can be done using the new modifier, it's not logic to repeat the same.
Reviewed By: mendio, filedescriptor
Differential Revision: https://developer.blender.org/D11604
Generally the evaluated mesh should not be changed, since that is the
job of the modifier stack. Current code is far from const correct in
that regard. This commit uses a const variable for the reult of
`BKE_object_get_evaluated_mesh` in some cases. The most common
remaining case is retrieving a BVH tree from the mesh.
In the Win32 platform our setTitle() can properly assign a Unicode
utf-8 window title. Unfortunately our getTitle() will only read regular
8-bit character strings. This means that we can never compare what we
set to what we get. This patch updates getTitle() to use Unicode-aware
GetWindowTextLengthW and GetWindowTextW.
see T88909 for an example of this affecting user experience.
Differential Revision: https://developer.blender.org/D11782
Reviewed by Ray Molenkamp
Handle width calculation was incorrect in drawing code. This caused
handles to be invisible when zoomed out.
After fixing math, handles become too large, so now they are constrained
to quarter of strip width, which feels more natural and represents
clickable area more closely.
`sequence_handle_size_get_clamped()` did not return size in pixels, but
in 2D-View space, this comment was corrected.
SeqCollection wasn't freed.
It wasn't easy to find culprit so added argument to
SEQ_collection_create() to pass function name.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D11746
Address initial feedback:
- Use checkboxes instead of radio buttons
- Hide snapping distance control from UI
- Tweak snapping line color - use selected strip color, 50% transparency. Similar to other editors
- Draw 2px thick line, since strip outline is also 2px thick
Reviewed By: HooglyBoogly
Differential Revision: https://developer.blender.org/D11759
When multiple File or Asset Browsers would load at once (e.g. when loading a
file with two File Browsers open) and they would load multiple directories or
.blend files (using the Recursions option in the File Browser or loading an
asset library with multiple .blends), often only one File/Asset Browser would
correctly load all files. Others would be incomplete or entirely empty. That
was because of a race condition, where the directories or .blend files would be
loaded concurrently and the first one that finished would cancel the other
ones. This again happened because they used the job system with the same
"owner", which by design makes all jobs with the same owner cancel as soon as
the first is finished.
Address this by making sure they have different owners. That is, not the scene
anymore, but the filelist the job belongs to. Doesn't make much sense to use
the scene as owner for scene-unrelated file loading anyway.
Steps to reproduce were:
* Open two File Browsers as regular editors.
* In the Display Settings popover, set "Recursions" to 2 or 3 levels.
* Navigate to a directory with plenty of subdirectories in both File Browsers.
* Save the file.
* Reload the file, one of the File Browsers likely has an incomplete file list.
Alternatively, use Asset Browsers and open an asset library containing multiple
.blends.
This patch writes the timecode in the .srt file relative to the start
frame of the scene. If the timecode is global but scene does not start
at frame 0 the subtitles don't match if they get loaded in an external
video player. Muted strips will be ignored. Don't allow negative
timecodes in .srt.
Reviewed By: Richard Antalik
Differential Revision: http://developer.blender.org/D11762
The error codes could be used to look up messages from a table
of messages however this wasn't especially useful.
Now all calls to BMO_error_raise must inclue a message.
Edit-mesh partial update logic assumed translate didn't need normals
to be recalculated (for faces with all vertices being transformed).
However translate can optionally rotate which requires
all transformed normals to be updated.
Check for this case and use the previous partial-update method
when it modified extra geometry, so the normals are properly reset.
Further updates need not recalculate them.
Supporting both object & edit-mode is more involved.
Both cases are now supported with object mode tracking the last-used
state for rotation so it's only reset once when rotation is disabled.
Normals now includes many functions including normal splitting &
custom normal manipulation split this into it's own file
to centralize related functions.
This node has two modes: the first mode computes a circle from three
locations and a resolution. The second takes radius and resolution.
The first mode also outputs the center of the computed circle as
a vector.
Differential Revision: https://developer.blender.org/D11650
This changes `UI_but_func_tooltip_set` so that it allows passing a custom free function, which has two benefits:
* The caller can pass `null` to indicate that the value should not be freed.
* Arbitrary c++ data can be passed to the callback (before the struct had to be trivially destructible).
I added `uiFreeArgFunc` and used it in other places where appropriate.
Differential Revision: https://developer.blender.org/D11738
Caused by {rBfba9cd019f21}.
Above commit reordered toolsettings snapping flags but missed remapping
these for the UV toolsettings in versioning code.
Differential Revision: https://developer.blender.org/D11756
Added a loop to add each category geo_node tests
Each geometry node test is an individual test. It can be run
separately by ctest command. For example ctest -R geo_node_mesh
This will run all mesh tests.
The extra overhead is each test adds to the total tests.
Updated the script, adding support for cmd arguments
Added exception errors for try except, formatting, adding comments
Improved the workflow, updated error messages. Re-running of the
test after BLENDER_TEST_UPDATE.
Merge branch 'geo-node-testing' into soc-2021-geometry-nodes-regression-test
Differential Revision: https://developer.blender.org/D11611
Added exception errors for try except, formatting, adding comments
Improved the workflow, updated error messages. Re-running of the
test after BLENDER_TEST_UPDATE.
Each geometry node test is an individual test. It can be run
separately by ctest command. For example ctest -R geo_node_mesh
This will run all mesh tests.
The extra overhead is each test adds to the total tests.
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.