When EEVEE is rendering multiple samples via
eevee_draw_scene, the command submission and in-flight
memory pressure would grow until all samples completed,
due to lack of intermediate flushing of GPU work and memory.
This patch adds a command flush and memory clear for this case
which occurs with high TAA sample counts during saving, similar
to the process in EEVEE_render_draw.
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#107221
- Use off-screen drawing when reading from the front-buffer isn't
supported.
- Add a capabilities flag for reading the front-buffer which is always
disabled on WAYLAND.
- Add GPU_offscreen_read_pixels_region, used for reading a sub-region of
an off-screen buffer - use for color-picking a single pixel.
Fix from [0] with conflicts resolved, worked around [1] not being
applied by checking if Wayland is in use via WM_ghost_backend().
[0]: 6cc2c16d06
[1]: 4e51008a82
It was broken in two ways:
- bpy_bmlayercollection_iter passed PY_SSIZE_T_MIN, while
PY_SSIZE_T_MAX was needed.
- bpy_bmlayercollection_subscript_slice() contained an
off-by-one error.
Pull Request: blender/blender#107165
The file extension was copied into a buffer without checking it's size.
While large extensions aren't typical, some callers used small fixed
size buffers so an unusually named file could crash.
Original commit was [0], applied with merge conflict fixed.
[0]: 5294758830
Previous fix to resolve GPU hang which could occur in the
Metal backend caused additional flickering to occur as
as side effect, due to removal of required execution
dependencies in certain places.
This patch resolves both problems by only removing the
GPU hang dependency stall when additional synchronization
primitives are used along-side the global sync primitive.
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#106914
Blender 3.5 has a performance regression in the image engine
that made the image engine 3-4x slower then 3.4. The cause of
this was the new way how panning was implemented.
This PR disables the new panning for now as a short term fix.
In the future the panning and improvements we did ensured
better performance when dealing with higher resolution images.
But the regression for regular images weren't acceptable.
This fix might introduce other performance regressions on
lower end systems.
In the future we still want to improve the performance to
get back to Blender 3.0 performance, but that requires more
work and has a different priority.
Pull Request: blender/blender#106803
Pull Request: blender/blender#107213
Caused by 2eeec49640.
Above commit would early out when falling through the specialized
greasepencil selection operator to view3d_select_exec. But in order to
select posebones in grease pencil weightpaint mode, we still have to
continue with view3d_select_exec.
Now check this special case [with convenient
`BKE_object_pose_armature_get_with_wpaint_check`] and DONT early out in
that case.
Should go into 3.3 LTS as well.
Pull Request: blender/blender#107076
The crash can occur in the following situations:
- Attempt to open a corrupted EXR file
- Attempt to save an EXR file under a non-existing directory.
The root cause is not really clear: for some reason the OpenEXE API on
the Blender side can not catch OpenEXE exceptions by a constant
reference to a std::exception, although it can by a constant reference
to an Iex::BaseExc.
This does not seem to be an issue with the OpenEXR library itself as
the idiff tool from our SVN folder catches the exceptions correctly.
It is also not caused by the symbols_apple.map as erasing it does not
make the problem go away.
It could still be some compiler/visibility flag which we were unable
to nail down yet.
The proposed solution is to add catch-all cases, mimicking the OIIO
tools. This solves the problem with the downside is that there are
no friendly error messages in the terminal. Those messages could be
brought as part of the workaround by additionally catching the
Iex::BaseExc exception. But probably nobody relies on those error
prints anyway, so added complexity in the code is likely does not
worth it.
Pull Request: blender/blender#107184
The selection (box select, click select...) had an offset when selecting keys in the timeline.
That was because the function to get the NLA mapping ignored the timeline.
Pull Request: blender/blender#106904
Occlusion query buffers not being cleared to zero resulted in
erroneoues selection in certain situations.
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#107135
Runtime this information is stored in the active_color_attribute and
default_color_attribute strings on the mesh, however when saving it
is still saved in the old format with flags on the CustomData layers.
When converting from the strings to the layers not all flags were
properly cleared from the CustomData layers, leading to multiple
layers having the CD_FLAG_COLOR_ACTIVE/RENDER flag.
Caused by uninitialized `ToolSettings` `GpPaint` [which was later
accessed in `BKE_gpencil_palette_ensure`].
Not 100% sure why `ToolSettings` `GpPaint` is properly initialized in a
default startup fille, but for some files, this was not the case (as in
the report)
See 22462fed00 for a similar commit.
Now initialize `ToolSettings` `GpPaint` (alongside `GpVertexPaint`) when
entering grease pencil vertex paint mode.
Should probably go into LTS releases as well.
Pull Request: blender/blender#107131
Due to floating point differences between importance sampling and
texture evaluation, disagreeing on whether or not a ray lies within
the sun disc.
* Use the same input values for geographical_to_direction() in
sky_radiance_nishita() and kernel_data.background.sun.
* The mathematical operations in pdf_uniform_cone() were adjusted to
match sky_radiance_nishita().
Pull Request: blender/blender#106764
The extrude node resizes an existing mesh, but doesn't initialize new
data for most non-generic data types like shape keys or freestyle tags,
or custom normals. The split edges node doesn't process some
similar vertex data too.
In the future this data can become generic attributes, or it can be
supported in the nodes anyway. But now the new data is un-initialized
after being allocated.
Fixes#106926
CustomData_get_active_layer_index() was used by accident. But that
returns the CustomData layer index (in all layers) as opposed to
CustomData_get_active_layer(), which returns the active UV layer.
Pull Request: blender/blender#106644
This PR reverts the breaking part of the #106535. This part doesn't seem
to be required to fix the HD4400-HD5500 issue.
Might also fix#106844.
Pull Request: blender/blender#106887
After investigating the crash logs it looked like the macro
unrolling wasn't working on Windows systems with these GPUs.
Macro unrolling was changed in order to cross compile to Metal and
in the future to Vulkan. The macro unrolling in OpenGL can be removed
by using a different naming scema.
This PR removes the macro unrolling by changing the generated GLSL
code:
**Before**
```
layout(std140) uniform _probe_block
{
ProbeBlock probe_block;
};
```
**After**
```
layout(std140) uniform probe_block
{
ProbeBlock _probe_block;
};
```
Some tweaks had to be done to the Eevee-shaders to make sure that
the macro unrolling is done correctly and could be compiled using
legacy opengl drivers.
Fix: #106278Fix: #106555
(and others)
Pull Request: blender/blender#106535
Changes to viewport state to resolve texture paint color
selection introduced a side effect wherein the correct
attachment size of a framebuffer was reset. This size is
needed when scissor regions are disabled to return the
state to its correct default. When this default was wrong,
certain operators would have incorrect offsets.
To maintain consistency with the OpenGL backend, the
Metal backend independently tracks the raw attachment
size using default_width/height. This will also reset to zero
when attachments are all removed, unlike other state which
may be retained.
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#106857
Before 6c774feba2, the maximum number of UV maps was
enforced throughout Blender. Now it has to be enforced in the few places
that actually have the limit. In the future the limit of the subsurf
could be lifted, but that isn't done now to keep the crash fix simple.
Since the positions of the final mesh are modified rather than a
separate array after d20f992322, the final mesh has
to become the deformed mesh after the deform modifiers, rather
than the input mesh. The arrays can't be shared anymore, but that
performance loss will be solved by implicit sharing shortly.
Give errors in a few cases:
- The mesh has no UV map
- The faces have no area
- The applied modifier has no curve data (it may have been modified)
Use errors instead of cancelling the operator completely so the
operator can still do something useful when many meshes are
selected and only some fail.
Pull Request: blender/blender#106823
This is mistake in the refactor of the DNA storage which unified
the camera and object storage. The bit which was missed from the
initial logic is that the autosave does not use regular file
write.
Detect this in the do-versioning code and rely on the new data
format when it exists.
A candidate for 3.5.1 release.
Pull Request: blender/blender#106811
The shader was compiled without the right define, disabling the world
volume lighting.
This had nothing to do with the light path node as the lighting
was totally disabled.
Pull Request: blender/blender#106787
The Exit Group menu entry was mistakenly only whown for nodes who are
themselves also group nodes. It should be show for any node inside a
group.
Pull Request: blender/blender#106643
Issue caused by inconsistency in GPUFramebuffer viewport state
between Metal and OpenGL. The MTLFramebuffer code has been
modified such that framebuffer viewport/scissor state is retained
and only updated if attachments are modified during bind.
This is consistent with OpenGL. Previously, other updates to the
framebuffer in Metal would reset the viewport region, especially
if attachments were temporarily removed. This caused the color
picker selection to be misaligned.
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#106619
Resolve small indexing issue in Metal implementation
of motion path line rendering.
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#106595
Fixes a bug introduced in b0b9e746fa.
The volume transformation matrix is multiplied in the wrong order
which means the grid scale is applied on the translation.
Regression caused by 98bfa8d458
The `XRAY_ENABLED(v3d)` check was lost during a modification to
`transform_snap_context_project_view3d_mixed_impl` function.
This caused vertex snapping in solid shading to jump between visible
and occluded faces.
Pull Request: blender/blender#106495
This was the case with hidden faces and `Sync Selection` turned ON.
Caused by 8f543a73ab.
Since 8f543a73ab, the UV element map
respects the hidden state of geometry, but stitching [which also
respected this on its own even prior to the culprit commit in its
calculation of connectivity] did this differently [it only skipped
hidden geo when UV_SYNC_SELECTION was OFF -- even though UVs would not
be visible which is probably the real error here, I believe there is
this principle that we "dont act on stuff we dont see"].
To fix this, also skip hidden geo (even with UV_SYNC_SELECTION = ON) in
the stitch calculation of connectivity, just as
`BM_uv_element_map_create` does it.
Should go into 3.3 LTS as well.
Pull Request: blender/blender#106493
`.asset_generate_preview()` internally calls `UI_icon_render_id` as a
job -- as opposed to `.preview_ensure()` [which internally also calls
`UI_icon_render_id`, but not as a job] leading to crashes in background
mode.
This might be due to the fact that OpenGL context is not set up
correctly (so there might be other ways to fix this), but there seems to
be other places/comments indicating that icon handling is only for main
thread (see e.g. 13beeb5892).
And while this does not fully explain why doing this with jobs works fine
from the UI, the patch certainly fixes the crashes in background mode for
now (by not using jobs).
Pull Request: blender/blender#106046
Found together with a fix for #106043.
Edge panning (in Node Editors, Outliner and VSE) does unnecessary
updates when the view has not changed at all. This includes adding
`MOUSEMOVE` events (even if you dont move the mouse at all).
Adding `MOUSEMOVE` events results in the transform system constantly running (even if you dont move the mouse) which we certainly want to avoid.
Rectify this by only calling these updates when the view changes.
Pull Request: blender/blender#106301
The fix for #105363 (6d3ce8273a) made the transform system respect
(Frame) nodes `offsetx`/`offsety`.
Now Node Editors run the transform system even if the mouse is still (due to edge panning -- unnecessarily, see PR #106301 for fixing this). And due to the way `frame_node_prepare_for_draw` recalculates these offsets (based on
on updated positions and `node.runtime->totr` [which in turn gets rounded in `node_update_basis` -- so subpixel precision is lost there, see [1]]), this can lead to slight imprecisions/noise/jitter during transform (if we use float offsets, see the PR for more info).
So to counter this, use rounded offsets now [which will keep the whole circle stable].
NOTE: PR #106301 would fix this already for having the cursor still, but this patch still improves slight jitter when moving, so will commit separately.
[1] comment from `node_update_basis`
> /* Round the node origin because text contents are always pixel-aligned. */
Pull Request: blender/blender#106096
When retrieving the evaluated mesh to paint on, it can be reevalauted,
where a smaller set of attributes are requested than for regular
evaluated meshes. That reevaluation should be completely removed
(see #106186), but a simple fix in the meantime is to manually
request that data.
Splitting workload dependency chains such that they
only exist within the context of a single frame.
Dependencies are required to ensure sequential
command buffer submissions execute in order,
but the additional dependencies between frames
could incur GPU timeouts, if a signal was delayed.
This could be triggered by both CPU/GPU cycles
texture updates and Viewport Compositor operations.
Should also resolve#106401
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#106443
Some functions changed slightly for this non beta release.
No functional changes though as we didn't use what was removed.
Pull Request: blender/blender#106861
The assets are required to build proper Blender release, so they can not be
skipped from packing.
The packing ignores the `working` directory as it seems to be big and sounds
that it is not needed for the release.
The assets are bundled under the `release/datafiles/assets` folder in the
blender sources. This is where they will reside after switch to the Git LFS.
Pull Request: blender/blender#106536
When accessing UVmaps from python in BMesh, the UVmap name/index was ignored
and the active UVmap always used. This fixes this by passing the layer index
to the underlying CustomData function.
Pull Request: blender/blender#106537
Caused by the changes in the submodules configuration.
Can no longer rely on the `git submodule` command to
show list of external references to be packed.
Needs to be backported to 3.5.1.
Pull Request: blender/blender#106502
This change aimed to solve the following issues:
- Possible threading issue of two tests writing to the same
file, depending on how the ctest is invoked
- Test using the release directory, and potentially leaving
temp file behind on test failure, breaking code sign on
macOS.
Pull Request: blender/blender#106311
The command buffer fails to execute, the cause is unknown. It does not
appear to be related to the binary archive cache as disabling that does
not prevent the issue.
Pull Request: blender/blender#106328
The UI needs persistent pointers to catalog paths that it can pass
around via context. It was trying to manage this in own storage, but
this didn't work. Not entirely sure why - the menus get redrawn
continuously while assets load, and the root menu rebuilds the mentioned
storage every time. Maybe the submenus redraw a bit later, or the
keyboard navigation handling still accessed data from a previous redraw
somehow.
Either way, instead of managing own catalog path storage, point into the
asset system, which already has persistent storage for the catalog
paths.
Pull Request: blender/blender#106237
Don't install libraries in blender.shared and don't generate manifest
as it's not working and not important to have a clean directory here.
Always detect MSVC_REDIST_DIR as it is used by oneAPI. The way
InstallRequiredSystemLibraries is called it already only sets some
variables and doesn't install anything by itself.
Pull Request: blender/blender#105999
This is a workaround fix for Open PGL 0.4.1 when the first volume
samples are collected in a later training iteration.
The problem is fixed in Open PGL > 0.5.0 and the workaround
can be removed after upgrading Open PGL.
The bug was found by ASAN when running the file from #105953. It's not
entirely clear what the root issue for the division by zero is, but
at least now the result is more predictable.
Pull Request: blender/blender#106165
Previous fix/hack a60f651502 made it so that all 'Fake User' linked
IDs are considered as directly linked, so that some IDs (like linked
Text) references are not lost on file save becaus they are not
effectively used.
However, this created unwanted references to all linked assets data,
because an asset always has its fake user set. This ended up keeping
lots of references to sub-assets when appending from the new Essentials
node groups.
NOTE: This is adding another hack to the collection, the whole handling
of 'which unused linked data reference to keep on file save' needs to be
reworked for Blender 3.6.
This was an issue with the `operator_context`. It's a bit confusing because it
looks like every `uiLayout` has its own, but in `uiLayoutSetOperatorContext`
you can see that the `operator_context` is always set on the root layout for
some reason.
The fix is to set `INVOKE_REGION_WIN` as the operator context after it has been
set to something else (I found this to be the default by printing the operator
context before it was changed).
Pull Request: blender/blender#106213
Currently assigning values to IDProperties clears their UI data.
That is fixed by #106161, but that is too risky for 3.5. Instead,
work around the issue by triggering `MOD_nodes_update_interface`
function which recreates the UI data of the modifier properties.
Pull Request: blender/blender#106190
UV attribute refactor in 6c774feb has changed the logic from "UV data does not exist" to "there's no active UV layer". The repro mesh has a UV layer, but not UV data due to the mesh being only a point cloud.
Pull Request: blender/blender#106185
Regression introduced at 19b63b932d
The crash happened in `gizmogroup_xform_find` because
`t->region->gizmo_map` was `nullptr`.
The `T_NO_GIZMO` flag was supposed to prevent the
`gizmogroup_xform_find` function from being called, but it was being
removed with `t->flag = t->flag & ~T_PROP_EDIT_ALL;`.
This is because 19b63b932d missed updating the maximum enum value in
`ENUM_OPERATORS`.
The fix is simply to set the correct maximum value in `ENUM_OPERATORS`.
But for more safety, `gizmogroup_xform_find` now returns early when the
region does not have `gizmo_map`.
Co-authored-by: Germano Cavalcante <germano.costa@ig.com.br>
Pull Request: blender/blender#106174
Vertex paint's filter operators were not properly going through the SculptSession API.
Undo didn't work; sometimes the operators would do nothing or produce corrupted colors.
The `transform_active_color` helper function now handles all of this. It ensures a valid
SculptSession exists and executes an undo push.
Reviewed By: Hans Goudey
Pull Request: blender/blender#106136
Bug introduced by c2c6707919
Fixing it by undoing the changes to the if statements
at the start of the FCurve drawing functions.
This keeps the intended behavior of the previous path, while fixing the drawing
Pull Request: blender/blender#106100
When using the slider operators in the Graph Editor
the code would try to access `FCurve.bezt` without checking that exists.
When the curve is baked that is a null pointer.
Pull Request: blender/blender#106102
Nodes inside of frames where pasted with an offset from the cursor.
Since the location of nodes is in parent space, child nodes don't need
to be offset separately.
Pull Request: blender/blender#106099
Workaround gnome-shell including uninitialized memory when pasting
from the clipboard. Where `read` would not write data into the range
return by the length.
Reading from the pipe into a power-of-two buffer
works around the problem.
It's not clear why this only impacts gnome-shell - as there is no
significant down-side to changing the buffer size, apply a workaround.
Ref !106091.
BLI_array_store still performed poorly for boolean arrays or any arrays
where many memory chunks had identical contents since the temporary hash
had many collisions, making lookups slow.
Resolve by ensuring duplicate chunks aren't added to the hash table.
Also increase the memory chunk size for edit-mesh undo to 64kb
which performs well with high poly meshes as it reduces the overhead of
having to manage many small memory chunks.
Notes:
- Before this change performance was quite bad (10-20x worse than v3.3).
- Performance from the test in #105046 is roughly the same as before.
- Performance of #105205 compared with v3.3 is close, even faster at
times but varies much more (likely caused by threading).
* Fix SDL not finding Xlib.h
* Link to clangSupport library for newer LLVM versions
* Add FreeBSD essential symbols to symbols_unix.map
Pull Request: blender/blender#105892
* With the move to gitea, .gitea doesn't need to be copied into installs.
* Add .github and .gitignore in more places for future proofing.
* Deduplicate ìnstall_dir macro.
Co-authored-by: Brecht Van Lommel <brecht@blender.org>
Pull Request: blender/blender#105980
Fix#105955: Cycles crash loading multiple PNG, TIFF and other image textures
concurrently. This OpenImageIO version includes a fix for this problem.
Ref #99618
Pull Request: blender/blender#105958
Workaround for compiler issue on AMD
platforms resulting in the erroneous
discarding of valid rays in fragment
raytracing.
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#105967
Use XCURSOR_THEME & XCURSOR_SIZE environment variables for Wayland.
While this isn't an official part of the spec, many Wayland compositors
& applications use these variables.
Fixed a bug where the active color wasn't being set
on imported meshes, resulting in no colors displaying
in the viewport.
This bug has been in the code for a long time. However,
the colors have been displaying correctly until recently,
so this issue wasn't previously apparent.
Also, changed custom color data name from "displayColors"
to "displayColor", to match the actual USD primvar name.
(This was a typo in the original code.)
Note that pull request
blender/blender#104542
addresses other issues in the color import code (e.g.,
converting all color primvars and not just "displayColor",
avoiding hard-coding of attribute names, handling all
iterpolation types, etc.).
However, the current commit is meant as a short term fix
to a regression, where the "displayColor" attribute does
not render in the viewport at all, until the above pull
can be merged.
133dde41bb changed how 'fake user' flag is handled with linked data.
Previous behavior was a bug/inconsistency, in that the 'directly linked'
tag would be 'over-set' and never cleared, forcing saving references to
a lot of unused linked data.
Note that ideally, 'Fake user' flag should be ignored, and the only way
to decide whether to keep or not a linked ID should be whether it's
actually used by some local data.
However, #103867 and #105687 show that this is causing issues in some cases,
where users wrongly relied on the linked data's pre-defined 'Fake user' flag
to keep their linked data in their production files, even if said data had no
real user.
While not ideal, for now we should consider 'fake user' flag for linked data
as a real usage case. A better handling of this edge-case is related to
wider designs aboud handling of 'non used' data on file save, whether
linked IDs should keep track of being explicitly or implicitly linked by
the user, etc.
The NLA's behaviour of always forcing a full set of keys (XYZW) is now
done in both 'Combine' and 'Replace' mode (previously only the former).
This fixes the bug where quaternions could not be properly keyed in 'full
stack tweak mode' under certain conditions.
Pull Request: blender/blender#105604
The `NODE_OT_link_viewer` operator has some complicated logic that was
intended to stop cycling through geometry nodes to make it work better
with nodes that have multiple geometry and data outputs (group nodes).
This logic failed when geometry was not the first socket and it would
just stop connecting data sockets altogether after the geometry socket.
The reason is that the `determine_socket_to_view` function stops looking
for already-viewed outputs after the first geometry, and then simply
reconnects that geometry.
This patch changes the behavior slightly so that, if a geometry output
is found, the node skips over all subsequent geometry outputs, but still
considers data sockets for viewing. Viewing a node with multiple geometry
outputs will connect the first geometry output it encounters and then
keep cycling through data outputs on repeated execution.
Pull Request: #105836
In a previous step of the mesh writing process, the loop `totlayer` was
replaced and a temporary vector is used for layer data instead. Use this
temporary vector to find the UV sub-layers, since the vector and the
CustomData struct in the mesh don't necessarily match. Also use a
local variable for mesh loop count to make it clearer that the mesh
isn't used except at the end.
Thanks to Campbell for finding the source of this error.
There were logic errors and use-after-free errors with the attribute
removal function. Because the custom data layers are reallocated,
we can't reuse the name pointer after removing an attribute. And
we can't return early on the first domain to fail for the edit mode
implementation, because another domain might have the attribute.
Also reorganize some of the code to make the logic clearer: only remove
sub-attribuutes and change attribute names after actually removing the
attribute,and assert if the attribute isn't removed after it is found.
The crash is fixed by reverting 87fd798ae3 and
some follow up commits. While it would generally be nice to move to a more
SoA format for these things, we are not there yet and this is causing more
trouble than it's worth currently. The main difficulty is that the socket
indices are changed by many different operations which invalidates the array
too often and led to many follow up bugs.
Pull Request: blender/blender#105877
Follow connections when reading the varname attribute of a primvar
reader, and support both string and TfToken types for the varname.
A unit test is also provided.
Authored by Apple: Matt McLin
Pull Request: blender/blender#105508
This was lost in the refactor to store material indices in a generic attribute.
The attribute API still allows this, but that will be handled separately
since it's a more complex task. The existing API that already clamped
input values should still do that.
Other areas like blenkernel and Cycles clamp the material indices to be
positive so this should be consistent with them. There is still discussion
if material indices should be made impossible, but this at least avoids
crashing for the 3.5 release.
There was also an inconsistency in how sculpt mode handles material index
higher than the number of slots.
Ref #105577
Simple logic: when the repository is not found under the relative
path to the main repository's origin, enforce clone of the upstream
repository instead.
Pull Request: blender/blender#105858
This flag is moved to a different variable but the default value is still placed on the wrong variable.
This fixes the default value assignment but due to the old flag bits are in conflict with used bits in the new flag variables, versioning changes are not included.
Pull Request: blender/blender#105852
When multiple nodes (Frame nodes included in the selection) are scaled/
rotated, the TransData location and center can get "wrong" due to the
fact that Frame nodes dont only use `locx`/`locy` for their
representation while drawing, but also `offsetx`/`offsety`.
So in order to use the "real" top-left corner in the transform system,
we have to respect `offsetx`/`offsety` when creating/flushing transform
data.
In addition to the file in the report, this patch was also tested to work
well with nested Frame nodes.
Pull Request: blender/blender#105400
Viewer node paths usually start with the modifier, but in pinned
node editors the tree may not be used by the object in context.
In that case the modifier part of the path should be ignored.
The viewer node is always disabled in that case.
Pull Request: blender/blender#105826
Resizing an image via the operator did not mark it dirty
(`IB_BITMAPDIRTY` is needed to pick this up as being modified, if this is
not set, no warning/option is shown on file close).
Note that using RNA would already do this correctly (since it uses
`BKE_image_scale` -- which already calls `BKE_image_mark_dirty`
internally).
Pull Request: blender/blender#105851
While **marking** an asset would update the Outliner immediately (this
due to the fact that `ED_asset_generate_preview` indirectly took care of
a refresh), **clearing** an asset would not do this.
Now be explicit about this in the Outliner listener and consider asset
notifiers there.
Pull Request: blender/blender#105287
Preview render depsgraphs are put in the depsgraph registry
concurrently with other threads. This was lacking a mutex lock
and a map value that remains unchanged when other elements of
the map are updated.
Pull Request: blender/blender#105839
When an object has no UV layers and an anonymous UV layer is created,
the anonymous layer gets set as the default (render) layer. This is
very confusing because it then uses a hidden anonmous layer. This patch
suppresses the usage of anonymous layers for rendering.
Pull Request: blender/blender#105192
`scissor_test_set` wasn't able to parse the arguments that were
passed correctly, due to incorrect control data during functino
registration.
This patch uses the correct control data during registration and
is able to parse arguments.
Ref: #104911
Pull Request: blender/blender#105850
Auto-depth is no longer reset during consecutive touch-pad motion.
Details:
- Add wmEvent::flag, WM_EVENT_IS_CONSECUTIVE to detect consecutive
track-pad & NDOF motion events. Expose via RNA as Event.is_consecutive.
- Consecutive events are broken by button/key presses and mouse motion.
- Add `WM_event_consecutive_data_*` functions, so operators can store
data between consecutive events.
- Add `ED_view3d_autodist_last_*` functions to access the last autodist
pivot point for view operators to use.
the path for clang_format_paths.py changed when the submodules
moved but format.cmd was never updated for that.
the work previously done by check_submodules.cmd is now done by
make_update.py so this file can be removed.
This is an issue revealed by the recent optimization in 4d3bfb3f41 to have
CPU and GPU subdivision topology both cached.
BKE_subsurf_modifier_subdiv_descriptor_ensure is what (re)creates the
topology refiner when needed. Invalidating the topology refiner on changes
must be done before it, otherwise we end up with an incomplete Subdiv that
either does not draw or draws incorrectly.
Pull Request: blender/blender#105844
Various UI code would store the `AssetHandle` in a way that turns out to
be unsafe. The file-data is part of the file browser caching system that
releases file-data when a certain maximum of items is in the cache. So
even while just iterating over the assets, earlier iterated asset
handles may become invalid. Now asset handles are really treated as
volatile, short lived objects.
For the asset-view, the fix was more involved. There we need an RNA
collection of asset-handles, because the UI list code requires that. So
we create a dummy collection and get the asset handles as needed by
index. This again meant that I had to keep the index of the collection
and the asset-list in sync, so all filtering had to be moved to the UI
list.
I tried duplicating the file-data out of the cache instead, but that
caused problems with managing the memory/ownership of the preview
images.
`AssetHandle` should be removed and replaced by `AssetRepresentation`,
but this would be an even more disruptive change (breaking API
compatibility too).
Fixes#104305, #105535.
Pull Request: #105773
When linking a texture directly to the material output, it's likely being
done for the purpose of previewing. In that case, bias the heuristic towards
not building a light tree, as it's likely not needed and slow on dense meshes.
Fix race condition if several competing threads are inserting Metal
buffers into the MTLSafeFreeList simultaneously while a new list
chunk is being created.
Also raise the limit for an MTLSafeFreeListChunk size to optimize
for interactivity when releasing lots of memory simultaneously.
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#105254
Optimized node graphs do not get cached and were
not correctly freed once their reference count reached
zero, due to being excluded from the GPUPass garbage
collection.
Also suppress Metal shader warnings, which are prevalent
during material optimization.
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#105795
Metal buffer binding validation would trigger an error
when a given shader had an empty PushConstantBlock.
This patch removes the default uniform code gen if
no uniforms are present, to avoid any possible issues
with buffers being bound to a shader where the destination
data block is size zero.
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#105796
Crash when selecting objects on AMD platforms running
Metal. This was caused by shader compilation warnings
being treated as errors in macOS 10.15. Wrapping
compilation failure with success check resolves error.
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#105739
immDrawPixels performs significantly slower in Metal
than OpenGL. This was caused by two main factors. Firstly,
the additional overhead of tiled texture update, where all
memory needed to be kept in flight for each update, but
caused update to take a slow path. Avoiding tile update
with Metal is more efficient for both memory pressure
and GPU pipelining.
Secondly, on AMD platforms, the staging buffer used
for temporary texture data was page-faulting when
several texture updates would occur within one frame.
This is due to limitations of allocating one large contiguous
memory chunk. Using the Metal buffer pool for staging
data is more efficient.
Authored by Apple: Michael Parkin-White
Pull Request: blender/blender#105794
Maximum distance of lines in screen space is limited. This limit seems
reasonable for FPS higher than 1, but UI allows to set 0.01 FPS with
soft. even lower values are possible.
This patch allows for normal operation within soft limits and labels are
still visible and quite usable within hard limits.
Pull Request: blender/blender#104849
- Show/hide mask layers: the tooltip was confusing from a user's
perspective, because they should not be expected to know what a hide
flag is.
- Active Spline -> Active Point: likely a copy and paste error.
- Geo Nodes face is planar node: forgotten article.
- Axes, plural instead of axis for the viewport preferences. Here
there are several usages of axis or axes. Since they refer to
coordinate axes I believe they should be plural, even though the
property `mini_axis_type` is still wrong.
-----
Pull Request: blender/blender#105814
The bug only happened in wireframe mode.
Originally the `const bool use_depth` indicated when projection should
be done on invoke.
Later it started to indicate when to snap on invoke.
But with keymap snap detection, using this `use_depth` is no longer
correct.
When creating the bounding box mesh for the viewport replacement,
copy the material from the original mesh. I'm not sure if Cycles is
meant to load materials from the original mesh or the Alembic file
itself, but either way, this should be a harmless change and fixes the
issue in the report.
Pull Request: blender/blender#105798
... or carnidal interpolation with only 2 color stops.
This was triggering an incorrect path due to missing optimisation cases.
Just fall back to the unoptimized case fixes the assert.
This was likely caused by a bad merge when doing the uniform name
patching for Metal.
This also fix the same issue for active UDIM tile overlay which wasn't
reported.
eaa87101cd made evaluated metaballs become meshes
for render engines. This exposed the normals calculated by the metaball
tessellation directly, whereas before they were recalculated later.
There are two fixes: not using the metaball tessellation normals at
all and calculating them with the standard Mesh code path instead,
or switching a define in the metaball code to use a more mesh-like
vertex normal accumulation. Since the results are very similar
(see attached renders), use the second solution, which should
be less expensive.
Pull Request: blender/blender#105799
Move the deform curves with surface node group to last so it affects the
final generated curves consistently (though this will have a negative
performance impact), and make the generate and interpolate modifiers use
the surface mesh's rest positions.
My PR blender/blender#104535
was committed as 88f9c55f7f and the logic was changed
while adding support for face sets, making the logic incorrect and
the warning system disfunctional.
Restore the logic from the original PR with added support for face sets,
fix const correctness issues, improve variable naming, and remove a
check for empty names, since all attribute-type layers should have
names in a valid mesh.
Fixes#105780
The switch from `stroke_start` to `vertex_start` in 8c6d4aa103 forgot
to also update `iter->stroke_index_last` which also needs to use
`vertex_start`.
Pull Request: blender/blender#105797
Skip explicit binding location for samplers in OpenGL when not needed, since drivers can usually handle more sampler declarations this way (as long as they're not actually used by the shader).
Pull Request: blender/blender#105770
The loading for the "All" asset library would include the "Current File"
library as if it were a regular asset libray on disk. Instead make sure
the latter is loaded properly first and is skipped when recursively
reading on disk libraries.
This PR fixes an error that is given when Metal API validation is enabled. The compute grid can exceed 65536 threads so `ushort` is not sufficient for `metal_grid_id [[threadgroup_position_in_grid]]`.
This PR also fixes OS version warnings ([Cycles Metal: Unguarded access to newer macOS features #105630](blender/blender#105630))
Pull Request: blender/blender#105763
The transformations were applied two times and the old fix was wrong because it needs to use the evaluated point, not the original one. Also I did a small code cleanup.
Pull Request: blender/blender#105202
Fix support for Wireframe and parametric nodes by resolving
compilation failures surrounding barycentric coordinates.
A final missing part of the Metal implementation for barycentric
coordinates was missing.
Feedback also addressed to move barycentric calculation out
of code-gen and into surface_lib.
Authored by Apple: Michael Parkin-White
This also resolves#103606.
Ref #96261
Pull Request: blender/blender#105740
Now only dynamic function parameters that use ParameterDynAlloc support
dynamically sized parameters arrays.
Add tests for both dynamic arrays that don't support resizing
(Image.pixels) and dynamic sized arguments using
(VertexGroup.add(index=[..])).
Regression in [0] which extended support for dynamic sized function
arguments.
[0]: dfb8c5974e
- DriverVariable.name update function passed DriverVar to
BKE_driver_invalidate_expression as a ChannelDriver.
- DriverTarget.name update function passed DriverTarget to
BKE_driver_invalidate_expression as a ChannelDriver.
- DriverVariable.type update function DriverVar accessed ChannelDriver,
clearing a flag.
This was exposed by [0] however this issue existed beforehand.
[0]: c26566ad27
This is a non-recent regression that strangely went unreported.
It is expected that when snapping, only visible elements are considered
which does not include faces in wireframe mode.
This works like this before, and this change doesn't appear to have
been intentional.
Ref #105664
One of the advantages of separating this enum member from the others is
because mixing several members in a single one hinders debugging since
in this case the IDE does not define which enums were set.
Also separating this item makes it more readable as `SCE_SNAP_MODE_GEOM`
is not a snap mode but a combination of modes.
The active frame is pasted always, so if multiframe is
copying the strokes don't need copy the active frame again.
Pull Request: blender/blender#105605
Issue was that the clamping parameters were not in
the correct order. This leads to undefined behavior
and also lead to small artifacts on other platforms.
Pull Request: blender/blender#105735
In most places where it appears in a menu, the operator would already
apply to all selected F-Curves. Now it is done consistently and explicitly
from all menu items. The default of the operator is now also set to 'all
selected', so that it also behaves like that when called from the operator
search menu.
- The Math node lost the headers of its operation type menu in
ee985fa925 , because a translation context was assigned to the RNA
property, but the headers declaration was not updated to extract the
messages with matching contexts.
- The message "Group Input" had a trailing space, which can be added
after translation.
Fix a regression that allowed to create several links between an
output socket and a multi input socket either by inserting
links or using the link swap feature.
This regression was caused by the link swapping feature
introduced in commit 89aae4ac82.
Pull Request: blender/blender#105631
Multiple user actions performed quickly could be blocked by undo
compacting memory - if the background compacting task was not complete
when the next undo step was pushed.
Notes:
- This and recent improvements to BLI_array_Store gives over ~2x speedup
compared with 3.3x, over 10x compared with 3.4x.
A sub-surfaced cube with the modifier applied was used for testing
(~1.5 million polys), both randomized & non-randomized verts/edge/faces
were used to avoid the sub-surface memory layout biasing the results.
Tested transforming ~1/3rd of the mesh and inverting selection.
- Without compacting mesh-data in parallel, the optimizations to
BLI_array_store can give similar performance to 3.3x, however there
are still cases where performance isn't quite as good - so compact the
arrays in parallel to ensure performance is at least as good as 3.3x.
Resolves#105046.
The method of accumulating values to create a hash for each chunk has
been improved for ~16% better distribution of the resulting hashes.
Improve performance of array de-duplication, see: #105046.
Accumulating hashes with a byte/boolean array didn't include enough
information for a useful hash, creating hashes with many collisions.
This is the root cause of a performance regression since 3.3 where
mesh data (used for storing edit-mesh undo steps) was changed to store
selection in a boolean array, creating a bottleneck de-duplicating
chunks of that array for edit-mesh undo's custom-data de-duplication.
Resolve by increasing hash accumulation for arrays with smaller elements,
so each chunk of memory (a candidate for de-duplication) isn't as likely
to have hash collisions.
`char` (single byte) arrays now accumulate 22 values instead of 7, it's
taking more values into account was necessary as these are effectively
bits in the case of boolean arrays, 2-byte values accumulate 32 bytes,
4-byte elements accumulate 44 bytes, larger structs accumulate
`sizeof(type) * 7` bytes (as before).
Also ensure the accumulation read-ahead never exceeds the chunk size -
technically a fix although this would only happen when passing a small
`chunk_count` to BLI_array_store_create (in the range of 1-16) so this
didn't happen in practice.
Improve performance of array de-duplication, see: #105046.
Use uint32_t since it's sufficient for hashing, using an int64_t was
especially inefficient when allocating an int64_t for every boolean
(when compacting an array of booleans).
Improve performance of array de-duplication, see: #105046.
This is a fix for the previous commit d7c023eb25.
Before, every time the lambda was called, a copy of the BitVector was
made. This was very inefficient.
Now this has been fixed by passing the BitVector by reference (&) in
the lambda function.
In very specific cases, during intersection testing, `intersect` can
add polygons already checked as duplicates in the buffer that
corresponds to the rest of polygons that can form groups of duplicates.
As the buffer cannot have repeated indices, re-adding, even
temporarily, these duplicates can cause a buffer overflow.
While this may have some impact on performance, it's difficult to
predict these cases and thus add a buffer pad.
So the solution is to check if they are already duplicated.
Don't expand view layer elements if element is previously used
This is similar as done for other display mode (eg. scene, library override)
Pass view_layer ID instead of scene as argument in `outliner_add_element`
Old Differential Revision: https://archive.blender.org/developer/D16661
Pull Request: blender/blender#104862
This patch enables edge panning for the operators where there was
a mismatch in behaviour between using the shortcut `G` and the
"Move" operator exposed via the menu and operator search.
Commit 19da434e9c introduced a `view2d_edge_pan` property which
was activated for the translate operators invoked when pressing G in
the default keymap.
The property is not enabled by default, to avoid it getting in the
way in some cases, so the "Move" operators exposed via the menus
didn't have edge panning. This lead to inconsistent behavior between
moving via the shortcut or via the menu.
Pull Request: blender/blender#104482
The use of wordexp(3) permits arbitrary code execution from manually-crafted
glTF files. See https://github.com/syoyo/tinygltf/issues/368 for more details.
In practice this shouldn't be an issue for Blender since the GlTF data isn't
manually crafted but from the OpenXR runtime (a bit like a driver). But
updating the library to include the fix is not a big deal anyway.
Note that the warning that required the local modification is no longer present upstream since
0bfcb4f49e
Pull Request: blender/blender#105536
Workaround for a crash when `addComputePipelineFunctionsWithDescriptor` is called *after* `newComputePipelineStateWithDescriptor` with linked functions (i.e. with MetalRT enabled). Ideally we would like to call `newComputePipelineStateWithDescriptor` (async) first so we can bail out if needed, but we can stop the crash by flipping the order when there are linked functions. However when addComputePipelineFunctionsWithDescriptor is called first it will block while it builds the pipeline, offering no way of bailing out.
Note that this only has an impact when the "MetalRT (Experimental)" option is checked.
Pull Request: blender/blender#105629
Archlinux faced an issue with AoT compilation: https://bugs.archlinux.org/task/77554
These were due to compiler flag strings handling going wrong when cmake
calls the custom command. It ended up moving quotes around, leading flags to
get passed to the main compiler instead of the graphics compiler and trigger
the following error: "Unrecognized build options: -ze-opt-regular-grf-kernel".
Pull Request: blender/blender#105615
The count wasn't clamped above zero in some newly optimized code.
Instead of adding it there, move the clamping to the field network,
similar to some other nodes. That makes it so the rest of the code
doesn't have to deal with the clamping, and should be faster in the
single-value case.
When subsampling was introduced in VSE it was disabled during
editing. Only when rendering it was enabled. This lead to
cache being different between editing and rendering. Leading
to confusion.
This PR will enabled the subsampling when editing. This way
it is more consistent.
Pull Request: blender/blender#105612
Some checks here are really critical and should assert, but that one is
more an indication that something is not going right, though data itself
should still be mostly valid, so better warn the user with a LOG
warning, than be silent in release builds, and crash in debug ones.
The traversable handle of a BLAS may be zero when the relevant geometry
is empty (no triangles/curves/points/...), as no BLAS is built in such cases.
It is not correct to attach a zero handle to a TLAS, so filter out such instances.
Invalid nodes are not added to the lazy-function graph. Therefore, their
outgoing links are also not added, which implies that the targets need
some default value.
The correction bbc6bb3468 was still wrong because there it was
disregarded that `vert_ctx_len` does not necessarily indicate merges in
the same polygon.
Therefore, it is not safe to rely on `vert_ctx_len` to count possible
new polygons.
NOTE: It might be worth preempting part of the
`weld_poly_split_recursive` logic to identify what the new polygons are
in advance. But this can be left for a future refactor.
Node tree updates can crash if the tree contains a node group that points at an "undefined" tree type.
This can happen if the tree is linked from a library and the path is lost,
or if a custom (python) tree is used and the script is not run.
The fix is to check if the node group type is valid ("registered") and return an empty list otherwise.
Pull Request: blender/blender#105564
With the move to Gitea the URI for line numbers changed from `$1234` to `#L1234`.
The change also removes a superfluous space character from the generated RST syntax.
When trying to use the Fill Range by Selection operator, in the
Distance from Camera Freestyle modifier, the operator fails if no
camera is active in the scene.
Before getting the data from the camera, check that it is defined, and
return otherwise.
Pull Request: blender/blender#105475
Pick select is only meant to change a single element from a single
data-block. However, the operator worked on each object individually
rather than first finding the closest point, then processing the
selection. Change the operator to find the closest point across all
objects, then deselect if necessary, then select the closest point.
Pull Request: blender/blender#105495
When running unit tests or other fast completing renders, forced crashes can occur if there are any slow, outstanding PSO compilation requests (due to the `std::terminate` fall-back case in `~ShaderCache`).
This patch eliminates the need for this shutdown hack by using of the async version of `newComputePipelineStateWithDescriptor` when creating a PSO for the first time. In doing so, we are able to explicitly respond to app shutdown instead of waiting for the pipeline to finish compiling (..and then timing out and force-crashing). We still use the blocking version of `newComputePipelineStateWithDescriptor` when loading from an archive, as this can handle loading from a corrupted archive gracefully. Finally, we move `addComputePipelineFunctionsWithDescriptor` to *after* the PSO is built (as this will trigger a full blocking compile if the PSO has not yet been built, which would bring back the original issue).
Pull Request: blender/blender#105506
MTLFramebuffer's viewport was not correctly updated when
updating attachments. Behaviour modified to be consistent
with OpenGL.
Authored by Apple: Michael Parkin-White
Ref #96261
Pull Request: blender/blender#105529
After ed870f87b9, panels headers displayed inside panels had their
label duplicated when translations were enabled. This is because a
string comparison was made against the original message, instead of
the translated message.
Pull Request: blender/blender#105151
- "Lines" in the sense of number of lines
- "Number" can mean "amount, count" or "index, offset"
- "Second" can be an ordinal number or a unit
- "Root": add the brush curve to the "square root falloff" sense
- "Strip" can be a sequence or a type of hair rendering
- "Constant" in the sense of a value, for the Geometry Nodes add
submenu (#105447).
Additionally, extract:
- "Press a key" from the Keymap preferences.
- "MaskLayer", upon new mask layer creation
Ref #43295, #105447
This was caused by an incorrect assumption in the solver:
It tries to solve both collision and length constraints simultaneously,
using the projected movement of a point as a slide direction along the surface.
This only works if the distance of the previous curve point to the surface
is less than the allowed segment length. Otherwise the segment will
exceed the allowed length even with zero slide and NaN values are computed.
The case of larger surface distance can occur if the previous segment
solve was already stretching the current segment and then the point
moves further away. In this case we can simply clamp the segment length
without violating the contact constraint.
Pull Request #105499
Some UI functions have a "translate" argument, which if set to False
specifies that the message is not to be translated. This sometimes
means that it was already translated beforehands.
But many messages were still getting extracted, sometimes twice in
different contexts. Some featured errors because the arguments of
various functions would be concatenated, such as:
```
col.label(text=iface_("Branch: %s") % bpy.app.build_branch.decode('utf-8', 'replace'), translate=False)
```
which would get extracted as:
```
msgid "Branch: %sutf-8replace"
```
Pull Request #105417
3e5ce23c99 introduced a regression in case the freed Main was part of a
list, and was supposed to be removed from it, since calling
`BLI_remlink` does _not_ clear the `prev`/`next` pointers of the removed
link.
This commit also contains a few more tweaks to recent related b3f42d8e98
commit.
Pull Request #105485
While this behavior can be useful in some cases, it can also create
issues (as in one of own recent commits, 3e5ce23c99), since it
implicetly keeps the removed linknode 'linked' to the listbase.
At least warn about it in the documentation of `BLI_remlink`.
This commit introduces a new Main boolean flag that marks is as invalid.
Higher-level file reading code does checks on this flag to abort reading
process if needed.
This is an implementation of the #105083 design task.
Given the extense of the change, I do not think this should be
considered for 3.5 and previous LTS releases.
This has the effect that the message is cut off at the end of the
first line. I copied the solution from other similar docstrings
elsewhere in the code.
As far as my regex-fu can tell, there are no other occurrences of this
in the codebase.
Issue reported by Joan Pujolar in #43295.
Pull Request #105474
When render is triggered from python and the render result is displayed
it isn't being updated as it wasn't tagged as being invalid.
Pull Request #105480
If the texture image path in the MTL is a "quoted" absolute path, the importer will fail to find the
file. It was only attempting to un-quote the path for the relative case. Now we attempt to un-quote
in all cases.
Pull Request #105478
msgfmt has a TBB dependency though bf_blenlib, now for a release build
The MSVC linker is smart enough to realize none of the TBB code is
actually used and discards it. In debug mode the linker is a bit more
conservative and doesn't, leaving msgfmt with a runtime dependency
on TBB. The problem here is, we only copy the runtime dlls during
the install phase, and msgfmt runs long long before that.
For this reason when we run msgfmt we should make sure any runtime
needs it could have are met in the path, there already is a handy
variable for that since oslc has similar requirements.
Pull Request #105048
The issue was that when using the `HD_ALIGNED` handle type,
Blender would not automatically move the keyframe handles with the key.
Instead one handle would get stuck in place.
To remedy that manually move the keyframe handles in case the type is `HD_ALIGNED`
This makes it consistent with moving a key with G
Pull Request #105401
Consistently use edge draw flag instead of original index to determine if an
edge should be drawn or not.
In GPU subdivision the edge original index was used for both edge optimal
display and selection mapping to coarse edges, but they are not the same.
Now match the CPU subdivision logic and use a separate edge draw flag VBO.
For cage display, match Blender 3.3 behavior more in showing/hiding of edges
in wireframe mode. That is edges without a mapping to an original edge are
always hidden when there is no distinct cage, and drawn otherwise. This is
not ideal for e.g. the bevel modifier where it will always show some edges on
corners despite all edges being hidden by the user. But we currently have
no good information to decide if these should be hidden or not, so err on
the side of showing too much as it did before.
Fie #103706: bevel modifier edges not drawn correctly
Fix#103700: optimal display can't be turned of with GPU subdivision
Fix wrong edge display with GPU subdivision preceded by other modifiers
Pull Request #105384
Timer management code often loops over the list of timers, calling
independant callbacks that end up freeing other timers in the list. That
would result in potentail access-after-free errors, as reported in #105160.
The typical identified scenario is wmTimer calling wmJob code, which
calls some of the job's callbacks (`update` or `end` e.g.), which call
`WM_report`, which removes and add another timer.
To address this issue on a general level, the deletion of timers is now
deferred, with the public API `WM_event_remove_timer` only marking the
timer for deletion, and the private new function
`wm_window_delete_removed_timers` effectively removing and deleting all
marked timers.
This implements design task #105369.
Pull Request #105380
Window activation events on Windows-10 don't seem to be reliable as it's
possible for Alt-Tab to trigger WM_ACTIVATE on a window when switching
away from it. As detecting the keys which are held relies on a valid
active state - this meant Alt could become stuck when using Alt-Tab
to switch between windows.
Disable reading modifiers on activation for WIN32, activating the window
now clears modifiers on WIN32. This isn't ideal as held modifiers wont
be detected, re-introducing the error reported in #40059.
Previously, UBO bind locations were linearly incremented and
relied on the correct uniform location being queried. This fix
is a future requirement for EEVEE next, however, pulling forward
due to Issue #105280 highlighting a possible flaw with expected
uniform locations.
Authored by Apple: Michael Parkin-White
Ref #96261
Pull Request #105311
Intel GPUs exhibit a number of rendering artifacts.
The most substantial being incorrect resolve of reflections.
Splitting the reflections_resolve shader into two passes,
one for SSR and one for light probes ensures correct rendering
and optimal performance on this GPU.
Also resolves an artifact with ambient occlusion wherein
the pow(a, b) function causes excessive precision loss.
Using an alternative method for power calculation on these
platforms resolves the issues.
Authored by Apple: Michael Parkin-White
Ref T96261
Pull Request #105240
Metal backed requires HOST_READ texture usage flag
for irradiance grid. This was correctly in place for the
basic grid, but not for grid_prev.
Authored by Apple: Michael Parkin-White
Ref #96261
Pull Request #105312
Pick selection for the Curves object was not considering multi
object editing. Only the active object was considered.
This fix introduces pick selection for Curves for multi object editing.
Pull Request #105184
The toggle (for anything but float and int types) was swallowed in
bf948b2cef. Also seems ef68a37e5d discarded
`property_overridable_library_set` for bools.
Now treat the overridable toggle as a general property for all property
types.
Pull Request #105370
Grease Pencil (when not in object mode) implements its own selection
opertor. This operator (`gpencil.select`) returns
`OPERATOR_PASS_THROUGH`, then falls though to `view3d.select` which can
toggle object selection (when using shift-click picking).
Removing `OPERATOR_PASS_THROUGH` would fix the object toggling, but this
was added in 62c73db734 with good reason (the tweak tool would not
work then).
Now prevent `view3d.select` from acting on Grease Pencil (when not in
object mode).
NOTE: longer term we could have grease pencil use view3d.select to avoid having to add these awkward exceptions
Pull Request #105342
Windows 11 has strange behavior with Alt-Tab.
In some cases an Alt-Press event is sent to the window immediately
after it is de-activated (both Left & Right Alt keys for some reason
even when only one is held).
This meant that:
- Modifiers could be enabled for de-activated windows
(so we can't assume de-activated windows have modifiers released).
- Releasing the modifier key would not be sent to the inactive window
causing the modifier key to be stuck.
- Button events over an inactive window are generated before activation,
so even though activation reads the correct modifier state,
the button event uses the "stuck" modifier state.
Now button & drop events on inactive windows always read the modifier
state first instead of relying on the modifier state to be cleared.
This has some advantages:
- If modifiers are held, they will be used as part of the click action.
- While modifier keys on inactive windows should be rare,
in the case this does happen - stuck keys are avoided.
So it makes sense to apply these changes for all platforms.
Also remove USE_WIN_ACTIVATE & USE_WIN_DEACTIVATE defines as they
were only added when changes to modifier handling failed on WIN32.
This logic has now been tested to work on all platforms.
1. Changes the subdivision function to not fill in time but add 0 to fix bug #104824
2. Fixes a bug in sanitization function noticed while fixing this bug.
Pull Request #105306
The material indices from the caps were only copied if the base mesh had
a material index attribute. Fix that by copying them manually if the cap
has the attribute.
Nodes are sorted based on the selection. In some cases (even depending
on processor speed, nodes can be selected and reordered, and another
operation can run before the next redraw). That gives a window where
operators mapped to the same input as selection can run with invalid
socket locations (which aren't updated after the nodes are reordered,
since they are stored in a separate array).
To fix this, move the socket locations from the node editor runtime
data to the node tree, tag them as invalid when the nodes are
reordered, and check for that status in a few more places.
A better longer term solution is not reordering nodes based on
UI status and instead storing the UI drawing order separately.
Pull Request #104420
This patch fixes hanging unit tests when MetalRT is enabled. It simplifies and fixes the kernel selection logic by baking the MetalRT-specific options into `kernels_md5` rather than expanding out and testing MetalRT bit flags explicitly.
Pull Request #105270
When an RNA type was registered, any existing dynamic types with the
same name were unregistered. This didn't account for built-in types
which cannot be unregistered in (most cases) allowing duplicate
operators for e.g. to be registered with the same name as existing
built-in operators (asserting with debug builds with GHash duplicates).
In practice the newly registered operator would be called unless
adding operators caused the GHash to be resized which could cause the
original operator to be used.
As registered RNA type identifiers are meant to be unique, don't allow
duplicate names in the first place. Now all dynamically registerable
RNA types prevent this situation, raising an error when scripts
attempt to register a type with an identifier matching the name of a
type that could not be removed.
Shortened in [0], increase to 1024 with the following changes:
- Use BKE_ST_MAXNAME for translation_context.
- Use OP_MAX_TYPENAME for wmOperatorType.name.
Both these limits were already used elsewhere but happened to use
RNA_DYN_DESCR_MAX (incorrectly).
[0]: f403d9a2b1
Exiting curves edit mode (going to object mode) would not update the
screen.
The fix adds a case to `ED_object_editmode_load_free_ex` for CURVES
to make sure the function returns properly. This then correctly adds the
notifier in `ED_object_editmode_exit_ex` to update the screen.
Pull Request #105252
- "Value" in the sense of color lightness is not the same word in
Japanese as other usages. See #105113.
- "Double" as a data type vs. a value.
Also extract "Custom Color Presets" in the tracking UI.
Run clang-format as well.
Pull Request #105187
This adds some simple null checks to avoid the crash. It might still
be good to improve the error message, but also does not seem as
important as avoiding the crash. Typically, users should not run into
this issue because the assets are shipped with Blender.
When using multiple compositor output nodes, compositing would fail
, showing a completely black output as it doesn't respect the active
node.
This patch will equalize the implementation with the viewer nodes.
Patch created by @OmarEmaraDevFixes: #86836
Pull Request #105235
Similar to recent issues with gl_shader_interface,
ShaderInput lists need to be sorted to ensure correct
and efficient uniform lookup by name.
Authored by Apple: Michael Parkin-White
Ref #96261
Pull Request #105239
This commit adds the default .usdz export capability.
The importer already supports usdz so no change is required other than updating the text in menu to match the updated exporter text.
On export, a .usd/a/c file has to be created first, and then converted to .usdz. A weird change of directory is required because of a quirk with the USD conversion to usdz. If an absolute filepath is passed into the `UsdUtilsCreateNewUsdzPackage` function, the usd files inside the usdz archive will have the same directory structure, i.e. if one tries to create a file at `C:\code\BlenderProjects\file.usdz`, when the usdz file is created, inside it will have the structure `\code\BlenderProjects\file.usdc`.
This is counteracted by setting the current working directory to the temporary session directory where both the usdc and usdz files are created, and just passing the file name to `UsdUtilsCreateNewUsdzPackage` without any filepath. Once the usdz file is created it is moved to the intended directory.
There is a separate `UsdUtilsCreateNewARKitUsdzPackage` capability for exporting usdz for iOS devices that will be implemented in a follow up patch as it will require some more small UI changes.
Co-authored-by: Charles Wardlaw (@CharlesWardlaw)
Co-authored-by: Sonny Campbell (@SonnyCampbell_Unity)
Co-authored-by: Bastien Montagne (@mont29)
Pull Request #105185, based on #104556.
Pull Request #105223
While some implementations of `getcwd` may return an allocated string
instead of the given char buffer in some cases, this is not the expected
behavior of the BLI wrapper. Not to mention the danger of returning a
pointer to a static char buffer...
Improve `ChangeWorkingDirectoryTest` to be more 'full check' regarding
behavior of both `BLI_current_working_dir` and `BLI_change_working_dir`.
Also move call to `BLI_threadapi_init` into proper `SetUp` method (to
have correct symmetry with the call to `BLI_threadapi_exit` in the
`TearDown` one).
Based on investigation by Charles Wardlaw (@CharlesWardlaw).
Pull Request #105220
Add `BLI_change_working_dir(path)` to change the current working directory.
This change is required for adding USDZ support to Blender. When exporting to that format, we are required to do a weird change of directory because of a quirk with the USD library's USDZ functionality. If an absolute filepath is passed into the `UsdUtilsCreateNewUsdzPackage` function, the USDZ archive will store that full path.
macOS uses `NSFileManager` through some new Mac-only wrapper functions.
Ref #99807
Pull Request #104525
Apparently `git checkout -t` is only allowed to happen for new branches.
Added a code which checks whether the branch already exists and it so
uses the `git checkout <branch>`.
Pull Request #105234
e8f4010611 unified the bounds computation for the new curves
object type and the rest of the curves system used by geometry nodes.
In the process, it made bounds affected by the control point radius.
In theory that makes sense; the bounds are supposed to be the extents
of the visible geometry. But in practice the change wasn't expected,
for a few reasons:
- The radius has never affected the bounds for the legacy curve type
- The default radius of legacy curve objects is absurdly large at 1.0m
- Only the new curve object has visible radius, and only in "strip"
mode or when rendering with Cycles
Currently the bounds are only used for the "Bounding Box" geometry node
and the panel in the 3D viewport sidebar, so there isn't any incentive
to choose less intuitive behavior yet.
Long term, the correct behavior is probably to include the radius in
the bounds, but this commit postpones that change to when it works
better with the rest of the curves system.
Pull Request #105154
Writing to a bitmap from multiple threads causes races when writing to
bits within the same integer. Instead, write to a separate boolean
array while subdividing, then move that to the final mesh bit vector.
Notes:
- The final copy to the bit vector could be replaced by a generic
`copy_from(Span<bool>)` call in the future.
- Theoretically we could entirely replace the `BitVector` with an
`Array<bool>`, but 1/8 the memory use for edges is likely worth it.
Pull Request #105156
GPencil 3D stroke rendering uses a geometry shader.
This is unsupported by the Metal backend, so implement
fix for this failing compilation by shifting geometry shader
logic into the Vertex shader for Metal backend.
Authored by Apple: Michael Parkin-White
Ref #96261
Pull Request #105143
Metal LineLoop emulation path does not correctly apply
when using SSBO vertex fetch mode alongside 3D line
rendering.
Patch moves line emulation above SSBO
vertex fetch setup to ensure the correct emulation
parameters are passed to the shader.
Authored by Apple: Michael Parkin-White
Ref #96261
Pull Request #105142
Resolves issue with nearest filtering on UI Icons. Note that as
Metal does not support LOD bias as a parameter on a sampler
object, the original code has been modified to perform LOD
biasing at the shader level.
As GPU_SAMPLER_ICON is not widely used, it is more
efficient to apply directly to the affected shaders, rather
than workaround passing in the sampler LOD bias as a
separate value e.g. uniform or push constant.
Original PR feedback addressed to also refactor ICON
shaders to use consistent style for single and multi
Icon rendering.
Authored by Apple: Michael Parkin-White
Ref #96261
Pull Request #105145
After 7eda9d8dda, crash occurs when operation search is called and no
active tree element is present. Add missing null check in new poll
function to fix the crash.
Pull Request #105004
After 49ad91b5ab, an extra visibility
toggle is being drawn for grease pencil layer in dopesheet. In this PR,
a condition is added to skip drawing of visibility toggle for GPencil
layer. Also, Grease pencil does not support pinning so restore the offset
added to channel for drawing the pin icon.
Pull Request #105080
This is a workaround for [issue #104087](blender/blender#104087). We encounter crashes when using shader binary archives on AMD, so this disables them while we investigate a proper fix. Kernels will still be cached automatically by the OS file system cache. This cache may occasionally be purged due to external factors, in which case kernels will get compiled again.
Pull Request #105186
If repository has multiple remotes with the same name of branch
checking out to the branch using simple `git checkout branch` exists
with an error: this is because there is ambiguity w.r.t which remote
to track.
Now the code explicitly provides remote to track, preferring to use
"origin" first (which is to be used for Blender style of workflow,
and Github style workflow when there is a fork available), and use
"upstream" if there is no origin.
Pull Request #105176
The preset names were translated before building the UI. This meant
that the translated name was passed to the operator responsible for
deleting the preset file, instead of the original name.
Pull Request #105155
This was actually caused by two problems. The first is that the
code actually deliberately detects if nothing happened and pushed
an undo step, a bugfix for a problem that no longer exists.
The second was that SCULPT_test_location sometimes does a closest
point search instead of ray casting and didn't check the result
against the brush radius.
This might've been a merge error, the result of color mixing
was being overwritten by a simple copy of source to destination
inside of layerCopyValue_propcol.
In 6514bb05ea I misinterpreted the function's intended
behavior when there was already an existing active layer. The data from
the active layer is just meat to be copied, the function should always
add a new attribute.
When using ShaderCreateInfo with builtin uniform(blocks) there are
cases where the current implementation could not find an existing
block. The reason is that it uses name matching and name matching
requires that the shader inputs are sorted based on the name hash.
This change fixes this by first for the sorting of the shader
inputs before resolving the builtins.
Pull Request #105127
Cycles fallback display shader previously did not use viewport.
This would crash or cause the display not to show when using
GPU backends other than OpenGL, if another display shader
was unavailable.
Now use ShaderCreateInfo for Cycles fallback display.
Authored by Apple: Michael Parkin-White
Ref #96261
Pull Request #104987
If the resolution attribute exists, it needs to be set to a non-zero
value for the newly added curves. While it might also make sense
to interpolate the value from neighbor curves, for now it's
simplest to just set it to the default value of 12.
Pull Request #105094
In curves edit mode, the "Move Texture Space" and "Scale Texture Space"
operators where shown in the "Curves > Transform" menu.
Since these are not implemented, the fix removes these operators
from the menu.
The code below has to handle the case when `sample_length` is greater
or equal to the total length already anyway, so we can just make that
a valid input. Currently, the snake hook brush for curves also makes
use of passing in larger sample lengths which current results in crashes
in debug builds.
When trying to rotate in curves edit mode using proportional editing
and connected only mode, Blender would crash.
This was because the `TransData` structs for disconnected elements
did not have the location field populated, resulting in a null pointer
access.
The fix skips these elements entirely using `TD_SKIP`.
This fixes issue [#105100](blender/blender#105100) where multi-pass renders can be incorrect due to kernels using stale specialisation constants (e.g. when rendering Pokedstudio).
This patch adds a new group of md5 hashes (`global_defines_md5`) to track whether the injected block of #defines is stale and regenerate the source string as appropriate. It also renames the existing group of md5 hashes from `source_md5` to `kernels_md5` to clarify that these refer to a specific kernel set rather than just the source (which might build an arbitrarily large number of kernel sets).
Pull Request #105103
The translation contexts module wasn't imported after c7611d61e8,
which caused an error and prevented some constraints' UI from being
displayed.
Pull Request #105095
The stoke shader of grease pencil uses a geometry shader stage. Apple
devices don't support shaders with geometry shader stage. In the
OpenGL driver there was a pass-through implemented so it didn't fail.
When using the metal backend this needs to be solved more explicitly.
This change patches the grease pencil shader to support both the
backends supporting a geometry stage and those without.
Fixes#105059
Pull Request #105116
Complex EEVEE nodegraphs, particularly those combining
multiple principledBSDF shader nodes have a tendancy
to require a large number of simultaneous live registers
due to function call depth. In some instances, this
causes substantial performance drop and corruption if
the stack gets too large.
To mitigate this, splitting calls to closure_eval such
that only a single individual closure is evaluated in each
call reduces the number of live registers required. This
is preferred over using compound closure evaluation
functions which require a large amount of in-flight data.
Note that this is generally not more optimal, if the stack
does not spill, as there is an increased instruction count.
The specific trade-off depends on the exact architecture
in question. Hence, this is limited to AMD GPUs.
Authored by Apple: Michael Parkin-White
Ref #96261
Pull Request #104985
66dda2b902 made an incorrect change to account for the special
case for NURBS. Instead, make the step that turns the lengths into
parameters more explicit, and pass the correct total length for each
curve, even in the cyclic case.
Pull Request #105079
This resulted in disappearing NURBS curves when joining them with other
curve types in some cases. The attribute has to be handles similarly to
the radius and resolution attributes rather than as simple generic data.
The attribute was retrieved when converting to Bezier curves when it
wasn't necessary. Instead copy it like a regular attribute if any of the
result curves are NURBS. Also fix a check for an empty span for the
attribute which was never false.
Invalid node trees (e.g. when nodes are linked in a cycle) can not be
evaluated and the viewer is not available in them. This commit just adds
some null checks to handle this case more gracefully.
Do the domain check directly in the field input class to avoid the need
for another function to do it elsewhere. Also move one function to
be closer to a similar one, rename some functions, and avoid the need
for two intermediate span variables.
Erroneous cache warming case where the generated material is
identical to default material and cached shader is re-used,
resulting in case where the parent shader is identical to the
source.
Authored by Apple: Michael Parkin-White
Ref #96261
Originally caused by 6514bb05ea
More cases where the active/default color attributes were not set
correctly:
[1] Using the old Python vertex_colors API (vertex_colors.new)
[2] OBJ importer
[3] Collada importer
[4] Data Transfer layout (both standalone operator and "Generate Data
Layers" from the modifier)
Similar to 101d04f41f.
Brought over from https://archive.blender.org/developer/D16977, see
discussion there why some of the code for data transfer is not for the
genereal attribute API.
Pull Request #105020
Drawing code `paint_draw_smooth_cursor` would be called correctly, it
was just the color not being initialized.
This is usually done with `BKE_paint_init`, but in the case of curves
sculpting brushes this would create an additional (unnamed) brush which
should be avoided since the workspace toolsystem creates the "right"
brush anyways.
So this patch just does the minimal work to get the Stabilize Stroke
indicator line to draw (which is initializing the color).
Brought over from https://archive.blender.org/developer/D16793
(cherry picked from commit 612a4382c443bcd02e0bb5ffd1b1fdbb251f6e7b)
Pull Request #105021
Write RNA properties backed by IDProperties with the `IDP_INT` type like
before ef68a37e5d. That keeps forward compatibility, because
earlier versions don't support the boolean type. At some point in a future
release we can remove the forward compatibility. 4.0 is a good time to
do that because forward compatibility will be broken in other ways.
This commit also adds support for reading integer arrays for boolean
property defaults, which was only half-way supported before.
Pull Request #104995
The light tree itself is disabled on the AMD GPUs due to a compiler issue.
There are couple of places where this was not fully checked:
- The `light_sample` function in the kernel.
- The light threshold during synchronization
The former one is solved as easy as just adding an ifdef block.
The latter one is solved by delaying the threshold assignment for
later on.
Pull Request #105022
This commit implements described in the #104573.
The goal is to fix the confusion of the submodule hashes change, which are not
ideal for any of the supported git-module configuration (they are either always
visible causing confusion, or silently staged and committed, also causing
confusion).
This commit replaces submodules with a checkout of addons and addons_contrib,
covered by the .gitignore, and locale and developer tools are moved to the
main repository.
This also changes the paths:
- /release/scripts are moved to the /scripts
- /source/tools are moved to the /tools
- /release/datafiles/locale is moved to /locale
This is done to avoid conflicts when using bisect, and also allow buildbot to
automatically "recover" wgen building older or newer branches/patches.
Running `make update` will initialize the local checkout to the changed
repository configuration.
Another aspect of the change is that the make update will support Github style
of remote organization (origin remote pointing to thy fork, upstream remote
pointing to the upstream blender/blender.git).
Pull Request #104755
Alternatively this could use `ASSET_LIBRARY_ALL` like
`add_node_search.cc`, but then it would need a different
method for skipping duplicate local assets.
Fixes the `-Winconsistent-missing-override` warning.
In theory the `virtual` is redundant in such case, but this is how
it is done in may other areas of USD code.
Pull Request #104977
In the outliner, the icons for modifiers are tinted blue. This didn't
work for the geometry nodes modifier icon.
Defining the icon with the macro `DEF_ICON_MODIFIER` also
defines the appropriate theme color so it's now tinted blue
when drawn in the outliner like the other modifier icons.
Pull Request #104957
The check was triggering the 'this' pointer cannot be null in
well-defined C++ code
We do not check for this pointer in any other areas. If it is
needed due to possible opaque pointer cast to the check prior
to the cast.
Pull Request #104974
After rBdb87e2a638f9, two contexts were missing:
- VirtualReality, to be used in an add-on
- Constraint, used in the constraints UI
The latter was actually used without being added, which caused errors.
It was reverted in rB31a640027982.
The operator's description used a format string like:
```
BLI_sprintfN(TIP_("%s %s vertex groups of the active object"),
action_str, target_str)
```
which is almost guaranteed to be impossible to properly localize to
some languages -- I know there are a couple of issues for French
already.
So instead of hoping formatting works, write all possible strings
explicitly, even if it looks stupidly verbose.
Simplify modifier sample mode didn't transfer UV parameters, now fixed.
(This is the corrected commit, previous one has multiple other commits merged.)
Pull Request #104964
This reverts commit 19222627c6.
Something went wrong here, seems like this commit merged the main branch
into the release branch, which should never be done.
Currently Metal is more stable then the OpenGL backend on apple
devices. Also the Metal backend supports more features then the
OpenGL backend. For example the viewport compositor and rendering
of production files.
This has been validated with users and studios.
This patch will default to the Metal backend when starting
Blender 3.5 for the first time or when loading factory startup. It
is still possible to switch to OpenGL via the user preferences.
It will not automatically select the Metal backend when there is
already user preferences available for Blender 3.5.
This is a missing part of the Metal backend. Metal backend doesn't
support the bgl commands, but it was not possible to use the gpu
module to use scissor testing.
Without this change addon developers would not be able to migrate
their addons to support the Metal backend.
This patch adds:
- `gpu.state.scissor_set`
- `gpu.state.scissor_get`
- `gpu.state.scissor_test_set`
Fix#104911
Texture clamping is by default off, but for VDM textures this leads
to several artifacts.
- Clamping is done when the image is sampled and when the brush strenght
is applied.
- VDM textures have negative values that are lost due to clamping
when clamping during sampling.
This patch fixes this by changing the default clamping of newly
created textures to not clamp anymore.
Textures are mostly being used by sculpt brushes, where clamping is also
implemented on brush level.
Fix#104747
Regression in [0] which caused interacting with 2D gizmos not to
update 3D gizmos once the gizmo finished it's modal interaction.
This caused the cameras lens gizmo not to update when navigating using
the viewport navigation buttons.
Resolve by detecting this case and flagging other draw steps to be
updated.
[0]: fb27a9bb98
In the first loop is safe to remove the current element.
The second loop can remove any element, potentially the next.
This triggers a read after freed.
Pull Request #104897
When the new UV to legacy format conversion happens, the mesh is in
the middle of being written and is an at best "complicated", at worst
invalid state. The attribute API looks at other domains and is a bit
less forgiving in that respect, and shouldn't really be used here. Use
the CustomData API instead. Also sort the layers the same way as
b642dc7bc7.
Note that the node group has its sockets names translated, while the
built-in nodes don't. So we need to use data_ for the built-in nodes
names, and the sockets of the created node groups.
Pull Request #104889
This reverts commit 68181c2560.
I merged 3.6 into 3.5 by mistake. Basically I had a PR against main,
then changed it in the last minute to be against 3.5 via the
web-interface unaware that I shouldn't do it without updating the
patch.
Original Pull Request: #104889
Note that the node group has its sockets names
translated, while the built-in nodes don't.
So we need to use data_ for the built-in nodes names,
and the sockets of the created node groups.
Pull Request #104889
When updating a mesh, the GPU Subdivision code makes calls to
`GPU_indexbuf_bind_as_ssbo()`.
This may cause the current VAO index buffer to change due to calls from
`glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, ibo_id_)` in
`GPU_indexbuf_bind_as_ssbo()`.
The solution is to unbind the VAO (by calling `glBindVertexArray(0)`)
before creating the index buffer IBO.
Co-authored-by: Germano Cavalcante <grmncv@gmail.com>
Pull Request #104873
Apparently, the 65bit Intel architecture is presented differently
on Linux and Windows.
Allow both variants for the command line, so that semantically the
command line argument can be seen as a lower case platform.machine.
Possible values are x86_64 and arm64.
Allows to use make_update.py in a cross-compile environment, like
building x86_64 macOS Blender from Apple Silicon machine.
Pull Request #104863
Fix a crash when using the Euler Filter from the Graph Editor on baked curves.
The crash happened because baked curves have no bezt array.
Skipping any curves where that was missing fixes the issue.
Co-authored-by: Christoph Lendenfeld <chris.lenden@gmail.com>
Pull Request #104858
According to the report it is a regression since 3.2, but it is tricky
to pin-point which exact commit caused it.
The root of the issue is that under certain circumstances frame might
be read and processed twice, depending on the order in which panels and
the main area is drawn: the footage information panel skips cache, so
it it is drawn prior to the main area it leads to 2 frame reads. Opening
the Metadata panel triggers code path which forces frame to be put to
the cache, solving the double frame read.
Solution is simple: do not skip cache when acquiring image buffer for
the footage information: the same frame will be needed for the main
area as well.
Pull Request #104860
On a user level there are no expected changes, other than being able
to update submodules and libraries from a main repository at a detached
HEAD situation (which did not work before).
On the infrastructure level of things this moves us closer to ability
to use the main make_update.py for the buildbot update-code stage, and
to remove the update-code section from the pipeline_config.yaml.
The initial idea of switching make_update to the pipeline config did
not really work, causing duplicated work done on blender side and the
buildbot side. Additionally, it is not easy to switch make_update.py
to use pipeline_config.yaml because the YAML parser is not included
into default package of Python.
There will be few more steps of updates to this script before we can
actually clean-up the pipeline_config: the changes needs to be also
applied on the buildbot side to switch it to the actual make_update.
Switching buildbot to the official make_update.py allows to much more
easily apply the submodules change as per #104573.
This was broken even before 0649e63716 and was always expanding the
`Image`, not the movie clip (even if the source was set to
`CAM_BGIMG_SOURCE_MOVIE`)
Now the rule here seems to be to always expand unconditionally, so
remove checking the source and always expand image and movie clip.
Co-authored-by: Philipp Oeser <philipp@blender.org>
Pull Request #104815
Cycles uses the "split faces" mesh function to support sharp edges
and auto-smooth. However, 75ad8da1ea updated that
function to ignore the edges that are explicitly tagged as sharp and
only use the edge angle. Fix by taking the attribute into account too.
The custom data layer mappings from dfacaf4f40 were created
*before* the BMesh shape key layers were added, invalidating the BMesh
data offsets they stored. Fix by creating the mappings after all layers
have been created.
Clean up logic to make it more clear and formalize the way to choose
fixed node data type based on operation. This make possible to more
easily fix wrong node data type for color type and less than ops.
Pull Request #104617
When appending assets it often isn't expected for the asset tags and
meta-data to be included. Add an option to the append operator to
disable appending the asset data, exposing existing internal options.
When the users click the "Mark as Asset" with the mouse hover the fake
user button, the button was not refreshed. In fact, the areas are not
listening to the "NC_ID NA_EDITED", which is the signal emitted after
an asset is marked/unmarked. Because of this, the areas aren't redrawn
(especially the ID buttons).
This little patch adds the event listening for the areas where this
problem is happening node editor and properties editor.
Pull Request #104694
* Fix#92539: Hard to read the breadcrumbs.
* Fix View Item active, hover, and text color (e.g. count numbers in the
Spreadsheet were almost unreadable).
* Fix mismatching node type colors with the default theme.
Blender Light is meant to be simply a brighter version of the default,
so screenshots and tutorials can be followed with both themes.
* Use the same outline color for widgets, so they match when aligned in a row.
* Make panels standout (not fully transparent), like in the default theme.
The active tools in `_defs_curves_sculpt` don't use names that are
exactly the same as the corresponding brush name with "builtin_brush."
at the beginning, instead they use more standard identifiers without
capitals or spaces.
The "brush_select" utility operator assumed the names matched though.
That can be fixed by manually mapping the brushes to the active tools.
Pull Request #104792
The keymap name in `WM_keymap_guess_from_context` didn't match the
name of the keymap in the Blender default keymap (`km_sculpt_curves`).
Fix by changing the utility function to match the keymap name.
Before right clicking on any tool in curves sculpt mode gave an assert,
now it shows a context menu.
Pull Request #104791
Lookup cache was not invalidated, to update attached effects position, a handle
of a input strip is touched.
To update attached effects, currently the code only does that when strip
position is changed. This is, because effect strip updating is done internally
in sequencer module code and ideally shouldn't be done at all. A TODO comment
with further explanation is added.
Avoid running out of attributes when multiple material slots use the same one.
Cleanup:
Removes the return value from drw_attributes_add_request since it shouldn't be modified afterward and it's never used.
Avoid making copies of DRW_AttributeRequest in drw_attributes_has_request.
Co-authored-by: Miguel Pozo <pragma37@gmail.com>
Pull Request #104709
The evaluated positions cache can live longer than a specific
`CurvesGeometry`, but for only-poly curves, it pointed to the positions,
which are freed when the curves are. Instead, use the same pattern
as the evaluated offsets and don't store the positions span, just return
it when retrieving evaluated positions.
Since the menu doesn't automatically align the labels like other menus
and pulldowns in Blender, I manually made them align using the blank
icon. However the menu button would also include this blank icon now.
This is a specific fix for the 3.5 release. In the main branch I will
replace it with proper support for automatically aligning labels in such
menus.
* BLENDER_VERSION_CYCLE set to beta
* Update pipeline_config.yaml to point to 3.5 branches and svn tags
* Update and uncomment BLENDER_VERSION in download.cmake
Usually when a menu item displays an icon, we indent all other items
with an empty icon so items align nicely. Now with more built-in asset
libraries (the new "Essentials" library), this inconsistency becomes
more apparent.
Also add a separator line between the "All" asset library and the
others, makes the menu look more organized.
The Preferences for asset libraries are becoming more than a simple name
+ path. E.g. there is now an Import Method options, and we previously
also considered a Relative Paths option (which we may still want to
add). The previous UI, while consistent with the Auto Run Python Scripts
UI isn't well suited for less than trivial cases. Using UI lists makes
the UI more scalable and follows usual list UI patterns more. There is
also more space for the path button now.
Part of #104686.
The default import method for an asset library can now be determined in
the Preferences. The Asset Browser has a new "Follow Preferences" option
for the importing. The essentials asset library still only uses "Append
(Reuse Data)".
This is part of #104686, which aims at improving the import method
selection, especially for the introduction of the new essentials library
(which doesn't support certain import methods). Further changes are
coming to improve the UI, see #104686.
Pull Request: #104688
The quick fur operator now uses the new hair system. It adds a new
curves object for every selected mesh, and adds geometry nodes
modifiers from the essentials assets that generate curves. A few
settings are exposed in the redo panel, including an option for whether
to apply the modifier to generate the initial curves so that there is
original editable data.
The point of the operator is to give people a sense of how to use the
node groups and to give a very fast way to build a basic setup for
further tweaking.
Pull Request #104764
Use the right hand side selection, as it fits the typical workflow
the best.
Arguably, the same would need to be done for the k-shortcut, but
that is another issue to be tackled. As well as making the selection
active.
Pull Request #104777
We (Dalai, Hans, Falk, Simon and me) decided that the curves edit mode
is useful enough to justify moving it out of experimental now. So far it
supports the following features:
* Various selection tools. The selections are synced with sculpt mode.
* Transform tools.
* Delete curves/points.
More functionality of the old curve edit mode will be ported over in
future releases.
This implements the delete operator in curves edit mode. The behavior
is similar to the delete operator in the edit mode of legacy curves,
i.e. it's actually dissolving and doesn't split curves. This is also
the behavior that we generally want for the hair use case.
The operator is added to the `Curves` menu and can be accessed via
the keyboard using `X` or `Del`.
Pull Request #104765
This adds a new `Curve Falloff` popover to the comb brush tool settings.
The curve control allows changing the brush weight along the curve to
e.g. affect the tip more than the root. This is a relative way to get
something like stiffness for short hair.
This functionality could potentially be added to some other brushes,
but the comb brush is the most important one, so that is added first.
I did add the buttons add the buttons to choose a curve map preset.
However, I did not add the preset dropdown, because that just adds
some unnecessary complexity in the code now and is redundant.
Pull Request #104589
This patch adds a simple operator to set values of the active
attribute for the selected element. The aim is to give simple control
over attribute values in edit mode rather than to provide the fastest
workflow for most cases. Eventually this operator might be less
important compared to more advanced attribute editing tools, but for
now, exposing a little bit of functionality is low hanging fruit and
will help to see the possibilities.
The implementation mostly consists of boilerplate to register the
necessary property types for the operator and draw their UI.
Beyond that, we just loop over selected elements and set a value.
Pull Request #104426
Caused by strips being flagged for removal, but the flag was never
cleared. As far as I can tell, this issue is not reproducible anymore,
but there may be files with this flag still set.
overlay_uniform_color_clipped was inheriting from overlay_depth_only, which doesn't
make much sense.
I've changed it to inherit from overlay_uniform_color instead, which is consistent
with other \*\_clipped variants of shaders.
Pull Request #104761
Certain material node graphs can be very expensive to run. This feature aims to produce secondary GPUPass shaders within a GPUMaterial which provide optimal runtime performance. Such optimizations include baking constant data into the shader source directly, allowing the compiler to propogate constants and perform aggressive optimization upfront.
As optimizations can result in reduction of shader editor and animation interactivity, optimized pass generation and compilation is deferred until all outstanding compilations have completed. Optimization is also delayed util a material has remained unmodified for a set period of time, to reduce excessive compilation. The original variant of the material shader is kept to maintain interactivity.
Also adding a new concept to gpu::Shader allowing assignment of a parent shader from which a shader can pull PSO descriptors and any required metadata for asynchronous shader cache warming. This enables fully asynchronous shader optimization, without runtime hitching, while also reducing runtime hitching for standard materials, by using PSO descriptors from default materials, ahead of rendering.
Further shader graph optimizations are likely also possible with this architecture. Certain scenes, such as Wanderer benefit significantly. Viewport performance for this scene is 2-3x faster on Apple-silicon based GPUs.
Authored by Apple: Michael Parkin-White
Ref T96261
Pull Request #104536
The logic for looping over imported OBJ faces and checking whether any
of them are "invalid" (duplicate vertices) was wrongly skipping
validation of the next face right after some invalid face. It
was the previously invalid face, moving the last into its place,
but then the loop was incrementing the face index and that just-moved
face was not properly validated.
Fixes#104593 - importing attached obj file (which contains some faces
that have duplicate indices). Added test coverage with a much smaller
obj file.
When the attribute doesn't exist, the node should give the default
of 12, as defined by the accessor method for `bke::CurvesGeometry`.
Pull Request #104674
This better aligns with OSX/Linux warnings.
Although `__pragma(warning(suppress:4100))` is not the same as
`__attribute__((__unused__))` in gcc (which only affects the attribute
instead of the line), it still seems to be better to use it than to
hide the warning entirely.
As described in #104171, add an operator that creates a new node group
that contain the current node group and named attribute nodes to deal
with the outputs. This saves manual work when moving a high-level
modifier to the node editor for better procedural control.
Pull Request #104546
This adds a new overlay for curves sculpt mode that displays the curves that the
user currently edits. Those may be different from the evaluated/original curves
when procedural deformations or child curves are used.
The overlay can clash with the evaluated curves when they are exactly on top of
each other. There is not much we can do about that currently. The user will have
to decide whether the overlay should be shown or not on a case-by-case basis.
Pull Request #104467
This renames `data` and `color` to `selection`. This is better because
it's actually what the corresponding buffers contain. Using this
more correct name makes sharing vertex buffers between different
gpu batches for different shaders easier.
The "current file" mode is only useful when creating new assets.
However, the far more common use case and the one that should require
fewer steps is to use existing assets.
There is a risk that this causes freezing if the file browser preview
caching does not work properly. So we'll have to keep an eye on the bug
tracker to see if this is an issue in practice.
Pull Request #104749
This patch adds an "Essentials" asset library that is bundled with Blender.
Also see #103620. At build time, the `lib/assets/publish` folder is copied
to `datafiles/assets` in the build directory.
In the UI, the "Essentials" library can be accessed like other custom asset
libraries with the exception that assets from that library cannot be linked.
The immediate impact of this is that Blender now comes with some geometry
node groups for procedural hair grooming.
Pull Request #104474
The OBJ spec (page B1-17) allows "l" entries to specify
polylines with more than 2 vertices, optionally with texture
coordinates.
Previously, only the first 2 vertices of each polyline
were read and added as loose edges, failing when texture
coordinates were present.
This adds support for proper polylines, reading but ignoring
texture coordinates.
Pull Request #104503
In 161908157d we moved all warnings
coming out of the library folder to /W0 as many of them do not follow
our code-style nor can we force them to.
When i made this change, i took `/external:templates-` to mean
"and that goes for you too, templates" and it decisively does the
opposite leading to /W3 warnings coming out of openvdb
This change removes the flag as it should have never have been added
in the first place.
Sculpt: Added vector displacement for the sculpting draw brush (area plane mapping only for now)
Vector displacement maps (VDM) provide a way to create complex displacements that can have overhangs in one brush dab.
This is unlike standard displacement with height maps that only displace in the normal direction.
Forms like ears, curled horns, etc can be created in one click if VMDs are used.
There is a checkbox on the draw brush in the texture settings "Vector Displacement" that enables/disables this feature.
Technical description: The RGB channels of a texture in a brush stroke are read and interpreted as individual vectors, that are used to offset vertices.
As of now, this is only working for the draw brush using the area plane mapping. Symmetry and radial symmetry are working.
A few things to consider when making VD-Maps:
* UVs need to stay intact for the bake mesh (e.g. voxel remeshing can't be used to create VD Meshes)
* When exporting a VD Map it should be in the file format OpenEXR (for positive and negative floating point values).
* Export resolution can be 512x512 or lower (EXR files can get very large, but VDM brushes don't need a high resolution)
And when using them:
* Inside Blender clamping needs to be unchecked on the texture
* The brush falloff should be set to constant (or nearly constant)
This patch was inspired by this [right-click-select proposal](https://blender.community/c/rightclickselect/WqWx/) Thanks for the post!
(Moved [this patch](https://archive.blender.org/developer/D17080) to here.)
Co-authored-by: Robin Hohnsbeen <robin@hohnsbeen.de>
Pull Request #104481
Curve type counts are updated eagerly but it was missing in this
node leading to a crash further down the line where the counts
were expected to be correct.
`MEM_delete()` is designed for type safe destruction and freeing, void
pointers make that impossible.
Was reviewing a patch that was trying to free a C-style custom data
pointer this way. Apparently MSVC compiles this just fine, other
compilers error out. Make sure this is a build error on all platforms
with a useful message.
The Volume property of the Maintain Volume constraint was marked as a
distance, which made it confusing--especially with non-metric units.
The volume can actually be understood as a factor of the initial
volume, so it should be dimensionless.
Additionally, the volume had a range of 0.001 to 100.0. This is wide
enough in most cases, but sometimes you may need to go orders of
magnitude higher or lower to consider vast or thin objects, and there
should be no drawback to extending the limits, provided they stay
positive.
Pull Request #104489
The proper fix (bb9eb262d4) caused compilation problems with HIP, so we're
delaying it until 3.6.
To fix the original bug report (#104586), this is a quick workaround that'll
hopefully not upset the compiler.
Pull Request #104723
A properly authored USD file will have the extent attribute authored on all prims conforming to UsdGeomBoundable.
This cached extent information is useful because it allows the 3D range of prims to be quickly understood without reading potentially large arrays of data. Note that because the shape of prims may change over time, extent attributes are always evaluated for a given timecode.
This patch introduces support for authoring extents on meshes and volumes during export to USD.
Because extents are common to multiple kinds of geometries, the main support for authoring extents has been placed in USDAbstractWriter, whose new author_extent method can operate on any prim conforming to pxr::UsdGeomBoundable. The USD library already provides us the code necessary to compute the bounds for a given prim, in pxr::UsdGeomBBoxCache::ComputeLocalBound.
Note that not all prims that are imageable are boundable, such as transforms and cameras.
For more details on extents, see https://graphics.pixar.com/usd/release/api/class_usd_geom_boundable.html#details.
Note that when new types of geometries are introduced, such as curves in https://developer.blender.org/D16545, we will need to update the USD writer for that geometry such that it calls this->author_extent.
Update on Feb 2: This patch has been updated to include a unit test to ensure authored extents are valid. This test requires new test assets that will need to be submitted via svn. The test assets are attached in the d16837_usd_test_assets.zip file. To use, unzip and merge the contents of this zip into the lib/tests/usd folder.
This unit test also addresses #104269 by validating compliance of exported USD via UsdUtils.ComplianceChecker.
Pull Request #104676
Use a MAKE wrapper for 'make deps' on Linux that ensures dependencies
are built one at a time. This is preferable because building many
dependencies at once made troubleshooting impractical and had the
downside that large deps such as LLVM would bottleneck on a single core.
This may be used for macOS, so far it's only tested on Linux.
- Correct broken link for undocumented modules.
Point to the contributing page, it seems #51062 was lost with the
GITEA task migration.
- Correct Blender Version link to the SHA1.
Matching the RNA id's to the search pattern is slow because
of the function `fnmatchcase`. This patch first checks the string
prefix without any special characters used by fnmatch,
if the `startswith` check fails, there is no need to check `fnmatchcase`.
Before the optimization, an online manual lookup took about 400ms
which is quite noticeable, with this patch applied it's under 10ms.
Ref !104581.
The internal compiler error appears to be gone. Unclear why it appeared in the
first place and why it's gone now. Just random kernel code changes causing it.
Pull Request #104719
- Rename roughness variables for more clarity - before, the SVM/OSL code would
set s and v to the linear roughness values, and the setup function would over-
write them with the distribution parameters. This actually caused a bug in the
albedo code, since it intended to use the linear roughness value, but ended up
getting the remapped value.
- Deduplicate the evaluation and sample functions. Most of their code is the
same, only the middle part is different.
- Changed albedo computation to return the sum of the intensities of the four
BSDF lobes. Previously, the code applied the inverse of the color->sigma
mapping from the paper - this returns the color specified in the node, but
for very dark hair (e.g. when using the Melanin controls) the result is
extremely low (e.g. 0.000001) despite the hair still reflecting a significant
amount of light (since the R lobe is independent of sigma). This causes issues
with the light component passes, so this change fixes#104586.
- There's quite a few computations at the start of the evaluation function that
are needed for sampling, evaluation and albedo computation, but only depend on
the view direction. Therefore, just precompute them - we still have space in
PrincipledHairExtra after all.
- Fix a tiny bug - the direction sampling code did not account for the R lobe
roughness modifier.
Pull Request #104669
In order to experiment with different storage types for `DRW_Attributes`
and for general cleanup (see #103343). Also move a curves header to C++.
Pull Request #104716
The existing logic to copy `BMesh` custom data layers to `Mesh`
attribute arrays was quite complicated, and incorrect in some cases
when the source and destinations didn't have the same layers.
The functions leave a lot to be desired in general, since they have
a lot of redundant complexity that ends up doing the same thing for
every element.
The problem in #104154 was that the "rest_position" attribute overwrote
the mesh positions since it has the same type and the positions weren't
copied. This same problem has shown up in boolean attribute conversion
in the past. Other changes fixed some specific cases but I think a
larger change is the only proper solution.
This patch adds preprocessing before looping over all elements to
find the basic information for copying the relevant layers, taking
layer names into account. The preprocessing makes the hot loops
simpler.
In a simple file with a 1 million vertex grid, I observed a 6%
improvement animation playback framerate in edit mode with a simple
geometry nodes modifier, from 5 to 5.3 FPS.
Fixes#104154, #104348
Pull Request #104421
Don't crash on nonexisting uv selection layers. Add an assert
because for now it is a bug if they don't exist. But when converting
back to Mesh it is preferable to accept in release mode, as opposed to
crashing.
Pull Request #104600
When generating a Mesh from a BMesh the uv map bool layers are not
copied if all elements are false. To suppress the copying the flag
CD_FLAG_NOCOPY is set in the layer flags. However these layers *do*
need to be copied to other BMeshes (for example undo steps). So we
need to clear them afterwards.
This commit adds the ability to import USD Shape primitives (Gprims).
They are imported as Blender Meshes using the USD API to convert, so
that they appear the same as they would in other applications. USD
Shapes are important in many workflows, particularly in gaming, where
they are used for stand-in geometry or for collision primitives.
Pull Request #104707
The issue was caused by rather recent refactor in 7dea18b3aa.
The root of the issue lies within the fact that the optical center was updated
on the Blender side after the solution was run. There was a mistake in the code
which double-corrected for the pixel aspect ratio.
Added a comment in the code about this, so that it does not look suspicious.
Pull Request #104711
`CTX_wm_operator_poll_msg_set()` is covered by the translation script
and always translates these strings. Checked with Bastien, he prefers
not having the redundancy here.
## Cleanup: Refactor NLATrack / NLAStrip Remove
This PR adds 3 new methods:
* BKE_nlatrack_remove_strip
* BKE_nlastrip_remove
* BKE_nlastrip_remove_and_free
These named BKE methods are really just replacements for BLI_remlink, but with some added checks, and enhanced readability.
Co-authored-by: Nate Rupsis <nrupsis@gmail.com>
Pull Request #104437
Send a notification and tag for an update even if the selection doesn't
exist, which is still necessary for drawing that depends on the
selection domain.
As a side effect of this change, more resolution divisions are now available.
Before this patch the possible resolution divisions were all powers of two.
Now the possible resolution divisions are the multiples of pixel_size.
This increase in possible resolution divisions is the same idea proposed in https://archive.blender.org/developer/D13590.
In that patch there were concerns that this will increase the time between a user navigating
and seeing the 1:1 render. To my knowledge this is a non-issue and there should be
little to no increase in time between those two events.
Pull Request #104450
This is needed to be able to query asset library information from an
asset. This again is relevant especially for the "All" asset library,
where you can't just directly access the library itself, which is
different for different assets.
The current design is that an asset representation is owned by exactly
one asset library, so having this pointer is perfectly compatible with
the design.
Reviewed by: Julian Eisel
This replaces `GPU_SHADER_3D_POINT_FIXED_SIZE_VARYING_COLOR` by
GPU_SHADER_2D_POINT_UNIFORM_SIZE_UNIFORM_COLOR_OUTLINE_AA`.
None of the usage made sense to not use the AA shader.
Scale the point size to account for the rounded shape.
Ref #104280
The scaling of area light / spot light blend was wrong because it is
calculated for pivot at the edges. The new implementation in theory
works for all `abs(pivot) <= 0.5f`, although we only have -0.5, 0, and
0.5.
- Axis constraint for box cage was only applied when there is translate
flag, now the same logic is applied regardless of the translate flag,
this means when dragging the edge, the scaling in the other axis stays
the same; when dragging the corners, it applies free-form scaling.
- Due to the existence of margin, `data->orig_mouse` does not lie
exactly on the boundary. Using that value to compute the scaling causes
the error to accumulate over distance. The new implementation uses the
original dimension of the object instead, and only uses
`data->orig_mouse` to determine the side of the original cursor relative
to the pivot.
- For circular gizmo with unsigned scaling, the gizmo only follow the
cursor exactly when the cursor stays in the original quadrant, otherwise
it's hard to handle the logic when we should clamp the scaling.
The buttons of enum context menus are of type `UI_BUT_ROW`. They
are part of the set of buttons we create underline shortcuts for in
`ui_menu_block_set_keyaccels`.
But since they weren't handled in `ui_handle_button_activate_by_type`,
pressing the underline shortcuts didn't do anyting in those cases.
Co-authored-by: Leon Schittek <leon.schittek@gmx.net>
Co-authored-by: Brecht Van Lommel <brecht@noreply.localhost>
Pull Request #104433
Currently the passepartout color is hardcoded to black. While a
sensible default for cinema, it may make less sense for other media,
whether video, print, web, etc. It greatly affects viewing conditions
of the image and should be user selectable, much like painting
programs allow.
Pull Request #104486
Don't create caps when using cyclic profile splines with two or fewer
points.
This case wasn't handled, yet, leading to invalid meshes or crashes.
Co-authored-by: Leon Schittek <leon.schittek@gmx.net>
Pull Request #104594
Vulkan has a pluggable memory allocation feature, which allows internal
driver allocations to be done by the client application provided
allocator. Vulkan uses this for more client application allocations
done inside the driver, but can also do it for more internal oriented
allocations.
VK_ALLOCATION_CALLBACKS initializes allocation callbacks for host allocations.
The macro creates a local static variable with the name vk_allocation_callbacks
that can be passed to vulkan API functions that expect
const VkAllocationCallbacks *pAllocator.
When WITH_VULKAN_GUARDEDALLOC=Off the memory allocation implemented
in the vulkan device driver is used for both internal and application
oriented memory operations.
For now this would help during the development of Vulkan backend to
detect hidden memory leaks that are hidden inside the driver part
of the stack. In a later stage we need to measure the overhead and
if this should become the default behavior.
Pull Request #104434
The GPU module has 2 different styles when reading back data from
GPU buffers. The SSBOs used a memcpy to copy the data to a
pre-allocated buffer. IndexBuf/VertBuf gave back a driver/platform
controlled pointer to the memory.
Readback is done for test cases returning mapped pointers is not safe.
For this reason we settled on using the same approach as the SSBO.
Copy the data to a caller pre-allocated buffer.
Reason why this API is currently changed is that the Vulkan API is more
strict on mapping/unmapping buffers that can lead to potential issues
down the road.
Pull Request #104571
Keep using the 3 evaluations dF_branch method for the Displacement output.
The optimized 2 evaluations method used by node_bump is now on its own macro (dF_branch_incomplete).
displacement_bump modifies the normal that nodetree_exec uses, so even with a refactor it wouldn’t be possible to re-use the computation anyway.
Using larger integer types allows for more efficient code, because we
can use the hardware better. Instead of working on individual bytes,
the code can now work on 8 bytes at a time. We don't really benefit
from this immediately but I'm planning to implement some more optimized
bit vector operations for #104629.
Pull Request #104658
Add a test to address the issue raised in #103913, where zero area
triangles could be created from polygons that have co-linear edges
but were not degenerate.
Since USD is no longer statically linked these linker tricks
are no longer needed.
Co-authored-by: Ray Molenkamp <github@lazydodo.com>
Pull Request #104627
When building a node group that's meant to be used directly in the
node editor as well as in the modifier, it's useful to be able to have
some inputs that are only meant for the node editor, like inputs that
only make sense when combined with other nodes.
In the future we might have the ability to only display certain assets
in the modifier and the node editor, but until then this simple solution
allows a bit more customization.
Pull Request #104517
During hair grooming in curves sculpt mode, it is very useful when hair strands
are prevented from intersecting with the surface mesh. Unfortunately, it also
decreases performance significantly so we don't want it to be turned on all the time.
The surface collision is used by the Comb, Pinch and Puff brushes currently.
It can be turned on or off on a per-geometry basis.
The intersection prevention quality of this patch is not perfect yet. This can
be improved over time using a better solver. Overall, perfect collision detection
at the cost of bad performance is not necessary for interactive sculpting,
because the user can fix small mistakes very quickly. Nevertheless, the quality
can probably still be improved significantly without too big slow-downs depending
on the use case. This can be done separately from this patch.
Pull Request #104469
Previously, the node used the "true" normal of every looptri. Now it uses the
"loop normals" which includes e.g. smooth faces and custom normals. The true
normal can still be used on the points by capturing it before the Distribute node.
We do intend to expose the smooth normals separately in geometry nodes as well,
but this is an important first step.
It's also necessary to generate child hair between guide hair strands that don't
have visible artifacts at face boundaries.
For perfect backward compatibility, the node still has a "Legacy Normal" option
in the side bar. Creating the exact same behavior with existing nodes isn't
really possible unfortunately because of the specifics of how the Distribute
node used to compute the normals using looptris.
Pull Request #104414
CustomData layer names should not be written except via the CusomData
api. Therefore use const char * instead of char * when referencing the
layer name.
Pull Request #104585
Avoid computing the non-derivative height twice.
The height is now computed as part of the main function, while the height at x and y offsets are still computed on a separate function.
The differentials are now computed directly at node_bump.
Co-authored-by: Miguel Pozo <pragma37@gmail.com>
Pull Request #104595
This adds a `select_lasso` and a `select_circle` function for the Curves object. It is used in the `view3d_lasso_select` and `view3d_circle_select` operator.
Co-authored-by: Falk David <falkdavid@gmx.de>
Pull Request #104560
This adds a new `select_linked` function that selects all the points
on a curve if there is at least one point already selected.
This also adds a keymap for the operator.
Co-authored-by: Falk David <falkdavid@gmx.de>
Pull Request #104569
Add `contains_group` method in python api for `NodeTree` type, cleanup
`ntreeHasTree` function, reuse `ntreeHasTree` in more place in code.
The algorithm has been changed to not recheck trees by using set.
Performance gains from avoiding already checked node trees:
Based on tests, can say that for large files with a huge number
of trees, the response speed of opening the search menu in the
node editor increased by ~200 times (for really large projects
with 16 individual groups in 6 levels of nesting). Group insert
operations are also accelerated, but this is different in some cases.
Pull Request #104465
This reverts commit aab707ab70.
A different solution to the submodule problem is being considered in #104573.
Revert to the previous behavior that developers are familiar with for now.
This just adds `threading::parallel_for` and `threading::parallel_invoke` in a few
places where it can be added trivially. The run time of the `separate_geometry`
function changes from 830 ms to 413 ms in my test file.
Pull Request #104563
Using callback functions didn't scale well as more arguments are added.
It got very confusing when to pass tehmarguments weren't always used.
Instead use a `FunctionRef` with indices for arguments. Also remove
unused edge arguments to topology mapping functions.
Adds an experimental option under "New Features" in preferences,
which enables visibility of the new Volume Nodes.
Right now this option does nothing but will be used during development.
See #103248
Pull Request #104552
Because of T95965, some attributes are stored as generic attributes
in Mesh but have special handling for the conversion to BMesh.
Expose a function to tell whether certain attribute names are handled
specially in the conversion, and refactor the error checking process
to use it. Also check for generic attributes on the face domain which
wasn't done before.
Author: Hans Goudey
Reviewed By: Joseph Eagar
Co-authored-by: Joseph Eagar <joeedh@gmail.com>
Pull Request #104567
When merging two gpencil layers, if the destination layer had a keyframe
where the source layer did not, strokes of the previous keyframe
in source layer were lost in that frame.
This happened because the merge operator was looping through
frames of the source layer and appending strokes in the
corresponding destination layer, but never completing
other frames than the ones existing in the source layer.
This patch fixes it by first adding in source layer
all frames that are in destination layer.
Co-authored-by: Amelie Fondevilla <amelie.fondevilla@les-fees-speciales.coop>
Pull Request #104558
When the sequencer is empty (i.e., there are no sequences),
we would have the deselect_all variable set to true called
ED_sequencer_deselect_all to select any existing sequences.
Ref !104453
Box-Selecting channels in the dope sheet with click-drag was no longer possible as of Blender 3.2
Due to the removal of tweak events the box select operator was always shadowed by the click operator.
Original Phabricator discussion here: https://archive.blender.org/developer/D17065
Use `WM_operator_flag_only_pass_through_on_press` on click operator to fix it
Co-authored-by: Christoph Lendenfeld <chris.lenden@gmail.com>
Pull Request #104505
(Follow on from D17043)
On AMD Navi2 devices the MetalRT checkbox was not hooked up properly and had no effect. This patch fixes it.
Co-authored-by: Michael Jones <michael_p_jones@apple.com>
Pull Request #104520
- Avoid flooding the output with every match that succeeds.
- Report patterns listed in the manual that don't match anything in
Blender.
- Disable external URL lookups, this is too slow.
Instead use a LOCAL_PREFIX (a local build of the manual)
or skip the test.
Without this, every access to "language" would warn that the enum
value didn't match a value in the enum items.
This made the bl_rna_manual_reference.py test output practically
unusable.
The generator now skips these with a warning, they will need to be
corrected in the user manual.
This caused tests/python/bl_rna_manual_reference.py to fail looking
up URL's.
Move the function for getting the language code associated with the
user manual into a utility function (from the generated
rna_manual_reference.py).
This allows other parts of Blender to create a manual URL based on the
current locale preferences and environment.
Ref !104494
Documented all functions, adding use case and side effects.
Also replace the use of shortened argument name by more meaningful ones.
Renamed `GPU_batch_instbuf_add_ex` and `GPU_batch_vertbuf_add_ex` to remove
the `ex` suffix as they are the main version used (removed the few usage
of the other version).
Renamed `GPU_batch_draw_instanced` to `GPU_batch_draw_instance_range` and
make it consistent with `GPU_batch_draw_range`.
When nodes are copied to the clipboard, they don't need their declaration.
For nodes with dynamic declaration that might depend on the node tree itself,
the declaration could not be build anyway, because the node-clipboard does
not have a node tree.
Pull Request #104432
Add a new node that groups faces inside of boundary edge regions.
This is the opposite action as the existing "Face Group Boundaries"
node. It's also the same as some of the "Initialize Face Sets"
options in sculpt mode.
Discussion in #102962 has favored "Group" for a name for these
sockets rather than "Set", so that is used here.
Pull Request #104428
Currently there's no way to assign a geometry node group from the asset
browser to an object as a modifier without first appending/linking it
manually. This patch adds a drag and drop operator that adds a new
modifier and assigns the dragged tree.
Pull Request #104430
As described in #95966, replace the `ME_EDGEDRAW` flag with a bit
vector in mesh runtime data. Currently the the flag is only ever set
to false for the "optimal display" feature of the subdivision surface
modifier. When creating an "original" mesh in the main data-base,
the flag is always supposed to be true.
The bit vector is now created by the modifier only as necessary, and
is cleared for topology-changing operations. This fixes incorrect
interpolation of the flag as noted in #104376. Generally it isn't
possible to interpolate it through topology-changing operations.
After this, only the seam status needs to be removed from edges before
we can replace them with the generic `int2` type (or something similar)
and reduce memory usage by 1/3.
Related:
- 10131a6f62
- 145839aa42
In the future `BM_ELEM_DRAW` could be removed as well. Currently it is
used and aliased by other defines in some non-obvious ways though.
Pull Request #104417
This adds a `select_box` function for the `Curves` object. It is used in the `view3d_box_select` operator.
It also adds the basic selection tools in the toolbar of Edit Mode.
Authored-by: Falk David <falkdavid@gmx.de>
Pull Request #104411
The required version numbers for various devices was hardcoded in the
UI messages. The result was that every time one of these versions was
bumped, every language team had to update the message in question.
Instead, the version numbers can be extracted, and injected into the
error messages using string formatting so that translation updates
need happen less frequently.
Pull Request #104488
`bAnimContext` had a float property called `yscale_fac` that was used to define the height of the keyframe channels.
However the property was never set, only read so there really is no need to have it in the struct.
Moreover it complicated getting the channel height because `bAnimContext` had to be passed in.
Speaking of getting the channel height. This was done with macros. I ripped them all out and replaced them with function calls.
Originally it was introduced in this patch: https://developer.blender.org/rB095c8dbe6919857ea322b213a1e240161cd7c843
Co-authored-by: Christoph Lendenfeld <chris.lenden@gmail.com>
Pull Request #104500
This is used for most Python release builds and has been reported to
give a modest 5-10% speedup (depending on the workload).
This could be enabled on macOS too but needs to be tested.
Blender was reporting that the GPU_TEXTURE_USAGE_HOST_READ wasn't set.
This is used to indicate that the textures needs to be read back to
CPU. Textures that don't need to be read back can be optimized by the
GPU backend.
Found during investigation of #104282.
- Use bpy.utils.execfile instead of importing then deleting from
sys.modules.
- Add a note for why keeping this cached in memory isn't necessary.
This has the advantage of not interfering with any scripts that import
`rna_manual_reference` as a module.
This allow to bypass all cost associated with shadow mapping.
This can be useful in certain situation, such as opening a scene on a
lower end system or just to gain performance in some situation (lookdev).
The merge with master updated the code to use the new matrix API. This
introduce some regressions.
For sunlights make sure there is enough tilemaps in orthographic mode
to cover the depth range and fix the level offset in perspective.
Implements virtual shadow mapping for EEVEE-Next primary shadow solution.
This technique aims to deliver really high precision shadowing for many
lights while keeping a relatively low cost.
The technique works by splitting each shadows in tiles that are only
allocated & updated on demand by visible surfaces and volumes.
Local lights use cubemap projection with mipmap level of detail to adapt
the resolution to the receiver distance.
Sun lights use clipmap distribution or cascade distribution (depending on
which is better) for selecting the level of detail with the distance to
the camera.
Current maximum shadow precision for local light is about 1 pixel per 0.01
degrees.
For sun light, the maximum resolution is based on the camera far clip
distance which sets the most coarse clipmap.
## Limitation:
Alpha Blended surfaces might not get correct shadowing in some corner
casses. This is to be fixed in another commit.
While resolution is greatly increase, it is still finite. It is virtually
equivalent to one 8K shadow per shadow cube face and per clipmap level.
There is no filtering present for now.
## Parameters:
Shadow Pool Size: In bytes, amount of GPU memory to dedicate to the
shadow pool (is allocated per viewport).
Shadow Scaling: Scale the shadow resolution. Base resolution should
target subpixel accuracy (within the limitation of the technique).
Related to #93220
Related to #104472
The previous change in the .gitmodules made it so the `make update`
rejects to do its thing because it now sees changes in the submodules
and rejected to update, thinking there are unstaged changes.
Ignore the submodule changes, bringing the old behavior closer to
what it was.
The meaning of the ignore option for submodules did change since our
initial Git setup was done: back then it was affecting both diff and
stage families of Git command. Unfortunately, the actual behavior did
violate what documentation was stating (the documentation was stating
that the option only affects diff family of commands). This got fixed
in Git some time after our initial setup and it was the behavior of the
commands changed, not the documentation. This lead to a situation when
we can no longer see that submodules are modified and staged, and it is
very easy to stage the submodules.
For the clarity: diff and status are both "status" family, show and
diff are "diff" family.
Hence this change: since there is no built-in zero-configuration way
of forbidding Git from staging submodules lets make it visible and
clear what the state of submodules is.
We still need to inform people to not stage submodules, for which
we can offer some configuration tips and scripts but doing so is
outside of the scope of this change at it requires some additional
research. Current goal is simple: make it visible and clear what is
going to be committed to Git.
This is a response to an increased frequency of incidents when the
submodules are getting modified and committed without authors even
noticing this (which is also a bit annoying to recover from).
Differential Revision: https://developer.blender.org/D13001
Subdivision surface efficiency relies on caching pre-computed topology
data for evaluation between frames. However, while eed45d2a23
introduced a second GPU subdiv evaluator type, it still only kept
one slot for caching this runtime data per mesh.
The result is that if the mesh is also needed on CPU, for instance
due to a modifier on a different object (e.g. shrinkwrap), the two
evaluators are used at the same time and fight over the single slot.
This causes the topology data to be discarded and recomputed twice
per frame.
Since avoiding duplicate evaluation is a complex task, this fix
simply adds a second separate cache slot for the GPU data, so that
the cost is simply running subdivision twice, not recomputing topology
twice.
To help diagnostics, I also add a message to show when GPU evaluation
is actually used to the modifier panel. Two frame counters are used
to suppress flicker in the UI panel.
Differential Revision: https://developer.blender.org/D17117
Pull Request #104441
The ear clipping method used by polyfill_2d only excluded concave ears
which meant ears exactly co-linear edges created zero area triangles
even when convex ears are available.
While polyfill_2d prioritizes performance over *pretty* results,
there is no need to pick degenerate triangles with other candidates
are available. As noted in code-comments, callers that require higher
quality tessellation should use BLI_polyfill_beautify.
Sockets after the geometry socket were ignored when cycling through
the node's output sockets. If there are multiple geometry sockets, the
behavior could still be refined probably, but this should at least make
basic non-geometry socket cycling work.
Now a single script to generate both links and release notes. It also includes
the issue ID for the LTS releases, so only the release version needs to be
specified.
Pull Request #104402
Minor change to [0], prefer calling em_setup_viewcontext,
even though there is no functional difference at the moment,
if this function ever performs additional operations than assigning
`ViewContext.em`, it would have to be manually in-lined in
`view3d_circle_select_recalc`.
[0]: 430cc9d7bf
Added missing documentation for `draw_cursor_add` and
`draw_cursor_remove` methods for `WindowManager`.
Differential Revision: https://developer.blender.org/D14860
Discard is not always treated as an explicit return and flow control can continue for required derivative calculations. This behaviour is different in Metal vs OpenGL. Adding return after discards ensures consistency in expectation as behaviour is well-defined.
Authored by Apple: Michael Parkin-White
Ref T96261
Reviewed By: fclem
Maniphest Tasks: T96261
Differential Revision: https://developer.blender.org/D17199
Host memory fallback in CUDA and HIP devices is almost identical.
We remove duplicated code and create a shared generic version that
other devices (oneAPI) will be able to use.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D17173
Straightforward port. I took the oportunity to remove some C vector
functions (ex: copy_v2_v2).
This makes some changes to DRWView to accomodate the alignement
requirements of the float4x4 type.
2023-02-06 21:25:45 +01:00
1962 changed files with 2100745 additions and 14308 deletions
option(WITH_CYCLES_DEVICE_HIP"Enable Cycles AMD HIP support"ON)
option(WITH_CYCLES_HIP_BINARIES"Build Cycles AMD HIP binaries"OFF)
set(CYCLES_HIP_BINARIES_ARCHgfx1010gfx1011gfx1012gfx1030gfx1031gfx1032gfx1034gfx1035gfx1100gfx1101gfx1102CACHESTRING"AMD HIP architectures to build binaries for")
set(CYCLES_HIP_BINARIES_ARCHgfx900gfx906gfx90cgfx902 gfx1010gfx1011gfx1012gfx1030gfx1031gfx1032gfx1034gfx1035gfx1100gfx1101gfx1102CACHESTRING"AMD HIP architectures to build binaries for")
mark_as_advanced(WITH_CYCLES_DEVICE_HIP)
mark_as_advanced(CYCLES_HIP_BINARIES_ARCH)
endif()
@@ -625,8 +625,10 @@ mark_as_advanced(
# Vulkan
option(WITH_VULKAN_BACKEND"Enable Vulkan as graphics backend (only for development)"OFF)
option(WITH_VULKAN_GUARDEDALLOC"Use guardedalloc for host allocations done inside Vulkan (development option)"OFF)
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.
