Started happening with 422dd9404f that introduced multi-threaded
conversions of src->dst (usually RGBA->YUV) format before encoding
the frame with ffmpeg. But the issue itself is not related to
multi-threading, but rather with the fact that AVFrame objects
started to be backed by an AVBuffer object (as that is needed for
threaded swscale to work).
Turns out, if a frame is backed by AVBuffer object, said buffer
might become "non writable" because it got shared (non-1 refcount).
And that happens with some ffmpeg video codecs, particularly PNG one.
Make sure to make the AVFrame objects writable inside
generate_video_frame. This follows official ffmpeg example
(doc/examples/encode_video.c) that explains why that is needed:
"the codec may have kept a reference to the frame in its internal
structures, that makes the frame unwritable. av_frame_make_writable()
checks that and allocates a new buffer for the frame only if necessary"
Pull Request: https://projects.blender.org/blender/blender/pulls/126317
Just turn on the option for geometry nodes where attributes are
generally expected to be used. We could expose this and other
options at some point too but that requires a bit more thought.
This patch improves the isotropic Gabor noise UI controls such that
variations happen in both directions of the base orientation, as opposed
to being biased in the positive direction only.
Thanks to Charlie Jolly for suggesting this improvement.
This patch optimizes the Gabor noise standard deviation estimation by
computing the upper limit of the integral as the frequency approaches
infinity, since the integral is mostly constant for the relevant
frequency range. The limits are 0.25 for the 2D case and 1 / 4 * sqrt2
for the 3D case.
This also improves normalization for low frequencies, possibly due to
the effect of windowing.
Thanks to Charlie Jolly for spotting the optimization.
Technically the regression was caused by #124472 which made it so
duplicating ImBuf allocates the exact amount of memory needed to
hold the pixels, while before IMB_dupImBuf() would leave the float
buffer over-allocated for images that are less than 4 channels per
pixel.
At the same time IMB_scalefastImBuf() was hard-coded to use 4
channels per pixels, for both byte and float buffers. It did not
crash in Blender 4.1 as it was accessing memory that is over-allocated,
but it also did not generate proper preview.
This fix makes the IMB_scalefastImBuf() to operate on an arbitrary
number of channels in the float buffer.
Pull Request: https://projects.blender.org/blender/blender/pulls/126234
Optimize the Gabor noise texture code with an early exit for points that
are further away from the kernel center. This was already done for the
kernel, but is now being done earlier before computing the weight, so
its computation is now skipped.
Thanks to Charlie Jolly for the suggestion.
In `mesh_join_offset_face_sets_ID()`, `.sculpt_face_set` is
modified but the `finish()` call wasn't present on the
`SpanAttributeWriter`, leading to warnings and potentially broken
data. This is now fixed.
When restoring the orig_position values for certain brushes, we need to
apply similar logic and avoid updating the base mesh if we are sculpting
on a non-basis shape key.
Pull Request: https://projects.blender.org/blender/blender/pulls/126305
This simplifies the feedback shown while using the "Interactive Editor
Docking" experimental feature. It removes the "X" icon shown on areas
that are removed. It does not show an icon for the areas that are left
over and altered. It only shows an icon for the active target area, and
does so without the outline of the icon. The result is a calmer
experience and it is easier to see exactly where the area is going.
Pull Request: https://projects.blender.org/blender/blender/pulls/126304
Introduced in 45d9bfa734
We currently do not invalidate the `active_vert` variable when changing
between different mesh implementations. Therefore, it is incorrect to
assume the value can be turned into a valid concrete reference simply
based on the PBVH type.
Pull Request: https://projects.blender.org/blender/blender/pulls/126299
When the operator was introduced, the program specified the immediate
mode program GPU_SHADER_3D_UNIFORM_COLOR but did not bind a particular
value. This leads to the color used for this circle changing depending
on the last color bound when drawing.
This change makes the color of the circle white, as initially intended,
and adds a darker outline to improve readability.
Pull Request: https://projects.blender.org/blender/blender/pulls/126259
Even if/when we take visibility into account for multires calculations,
the per-grid-face visibility is stored in SubdivCCG::grid_hidden, not
the base mesh's face visibility.
d1049f6082 translated the logic from the `int8` to `float`
conversion incorrectly. Previously the data was extracted to an integer
array (multiplied by 254) first then translated to a float VBO (and
divided by 255). I think conceptually the special `1.0/255` value isn't
necessary because the "optimal display edges" area already taken into
account for the creation of the index buffer. But as a quick fix, just
fix the factor to be the same as before.
Pull Request: https://projects.blender.org/blender/blender/pulls/126290
Retrieve the relevant data directly from the original object's mesh
(or the evaluated SubdivCCG in that case) rather than using the
BVH tree's geometry points which we'd like to remove.
Pull Request: https://projects.blender.org/blender/blender/pulls/126284
No functional changes expected.
The function `BKE_action_get_item_transform_flags` was only ever used in a single place.
As such it could be a static function within the given file.
This reduction in scope will help with future refactors.
This is in service of fixing #126125 since the function in its current form
does not support layered actions.
Pull Request: https://projects.blender.org/blender/blender/pulls/126287
This avoids an issue in clang-cl on Windows ARM64 where an invalid pointer would be (re)used.
See #124182 for more details, but basically without this, when the vector of tiles inside changeset is resized, a reference to an invalid pointer (ie, the location in the old vector) is kept somewhere, which makes the test fail.
Pull Request: https://projects.blender.org/blender/blender/pulls/126083
Fixes missing intersections on straight 3D curves with the
Metal backend, with BVH2.
This issue could of manifested on other devices, but didn't seem to
in practice.
Pull Request: https://projects.blender.org/blender/blender/pulls/126197
The lazy-connect feature of node wrangler uses the built-in `connect_sockets`
function which automatically handles virtual sockets in group input and output
nodes already. However, it does not handle virtual sockets in other nodes.
The fix is to generalize this behavior. For that, a new `handle_dynamic_sockets`
boolean input is added to `tree.links.new`. When enabled, virtual sockets are
handled properly by internally calling the `bNodeType.insert_link` methods.
The new behavior is turned of by default for compatibility reasons.
Pull Request: https://projects.blender.org/blender/blender/pulls/126282
This patch adds support for CPU side buffers for the result class. A new
storage type member was added to identify the type of buffer storage,
and allocation will either allocate a GPU texture or a CPU buffer based
on the context's GPU usage.
This implements the `Apply Modifier` operator for GPv3.
Applies the modifier to the current frame. Drawings at the current frame will be updated.
If the result of the modifier creates new layers, the layers will be added and a drawing
at the current frame will be inserted.
Applying a modifier will run the following steps:
* Create a localized copy of the original Grease Pencil data and insert it into a geometry
set that owns it.
* Execute the `modifiy_geometry_set` function of the modifier with this geometry set.
* In the resulting Grease Pencil data loop over all layers:
* If the original data has a layer with the same name, overwrite its drawing data in the
current frame.
* Otherwise create a new layer in the original data and insert a resulting drawing at
the current frame.
* Remove all original layers that are not mapped to from a result layer.
* Remap the material indices for all drawings that have not been updated (e.g. the ones
that are not on the current frame).
* Copy all the layer attributes to the original Grease Pencil.
Pull Request: https://projects.blender.org/blender/blender/pulls/124543
The problem was the automatic instance deduplication. There were two
instance references before baking, both of which were referenced by instances.
When loading the bake, they were deduplicated, so there was only one geometry,
but the instances still referenced two.
The fix is to not do deduplication when loading instances from a bake.
No need to enable the GL_ARB_conservative_depth extension as it is
core in GLSL 4.20. Some drivers still complain that the
extension was explicitly enabled.
Detected on AMD 21.Q2.1 (27.20.21026.2006) legacy driver.
Pull Request: https://projects.blender.org/blender/blender/pulls/126223
The encoded material names used by cryptomatte were incorrect. Reason
was that it used the name including the object type.
NOTE: reference images needs to be updated
Pull Request: https://projects.blender.org/blender/blender/pulls/126267
