Not sure why this bug was only discovered by such an elaborate steps
and why it took so long to be discovered. The root of the issue is
that in the 956c539e59 the typical flow of tag+flush+evaluate was
violated and tagging for visibility change happened after flush.
Since rBb67fe05d4bea2d3c9efbd127e9d9dc3a897e89e6 collections
have a geometry component that depends on all the geometries
inside the collection. Contrary to what I originally thought
`ID_RECALC_COPY_ON_WRITE` does not trigger a collection geometry
update. This makes sense because a collection may change in ways
that do not require a geometry update.
Instead, we have to trigger the geometry update manually now by
passing `ID_RECALC_GEOMETRY` when appropriate.
Currently, colored links overlay only supports standard sockets defined
by Blender. Some add-ons like Animation Nodes defines custom sockets for
everything and hence doesn't get colored sockets.
This patch uses the draw color from the socket type info to draw links
in order to support custom sockets.
Differential Revision: https://developer.blender.org/D13044
Reviewed By: Hans Goudey
The geometry node port of voronoi_smooth_f1 function has a
division by zero when smoothness is set to zero.
Using a safe_divide within the function causes issues
and was noted in the original patch D12725.
Solution in this case is to clamp zero smoothness to FLT_EPSILON.
Reviewed By: JacquesLucke
Maniphest Tasks: T92736
Differential Revision: https://developer.blender.org/D13069
Need to make sure images needed for hair shaders are loaded
before running the shader.
The naming is a bit misleading, but this is an internal API
and we can change it easily. Submitting minimal patch needed
to fix logic in the code to make it safer to review for 3.0.
Differential Revision: https://developer.blender.org/D13067
Calling it with a None argument, or no arguments, or with a property
that is missing UI data for some reason would fail. There is no
particular reason why ensuring those things don't happen is helpful,
so just add null checks for safety.
Differential Revision: https://developer.blender.org/D13024
Some compositor tests (e.g. `compositor_color_test`) broke
because of rB0c3b215e7d5456878b155d13440864f49ad1f230.
The issue was a heap-use-after-free bug caused by a missing
call to `MEM_CacheLimiter_unmanage`.
Operators such as setting the object mode failed after calling
WM_OT_open_mainfile from Python.
Keep the window after loading a file outside the main event loop.
Previously, `ImageTile->ok` and `ImageUser->ok` were used to indicate
whether an image failed to load. There were three possible values
which (probably) had the following meanings:
* `0`: There was an error while loading the image. Don't try to load again.
* `1`: Default value. Try to load the image.
* `2`: The image was loaded successfully.
This image-wide flag did not make sense unfortunately, because loading
may work for some frames of an image sequence but not for others.
Remember than an image data block can also contain a movie.
The purpose of the `->ok` flag was to serve as an optimization to avoid
trying to load a file over and over again when there is an error (e.g. the
file does not exist or is invalid). To get the optimization back, the patch
is changing `MovieCache` so that it can also cache failed load attempts.
As a consequence, `ibuf` is allowed to be `NULL` in a few more places.
I added the appropriate null checks.
This also solves issues when image sequences are used with the
Image Texture node in Geometry nodes (also see D12827).
Differential Revision: https://developer.blender.org/D12957
Caused by own {rB5aa3167e48b2}.
Related commit: {rBebaa3fcedd23}.
For stereo renders, `BKE_image_is_multilayer` is true, however we seem
to get to down to `space_image_gpu_texture_get` [where this is called]
from `IMAGE_cache_init` with a NULL Image->RenderResult.
So what then happens is that `BKE_image_multilayer_index` is called and
even though it has an appropriate codepath for stereo, it earlies out
and does not set multi_index correctly.
Still a bit puzzled why RenderResult is NULL for a render, but since
other places also check for a valid RenderResult before going down the
_multilayer_ route (and doing _multiview_ instead), now do the same
thing, BKE_image_multiview_index is now called in these cases (and seems
to behave correctly, checked with layers and passes and all seems to
display correctly, either in stereo or choosing individual eyes).
thx @jbakker & @brecht for double-checking.
Maniphest Tasks: T92608
Differential Revision: https://developer.blender.org/D13063
Even though volumes can only have one material, it is still necessary to
allow assigning a material to a prodecurally created volume. This commit
adds volume support and a warning for when a non-constant selection is
used with a volume.
Fixes T92485
Differential Revision: https://developer.blender.org/D13037
