When calculating the cavity factor, it is possible for the relative
distance of all traversed connected vertices to be zero. This results in
a division by zero which does not get clamped correctly to the expected
[0.0, 1.0] bounds. Prior to 4.2, this would have had no effect, as the
processing of this vertex would be have been skipped entirely. Due to
changes during the brush refactor, this flaw in the existing code was
exposed.
Pull Request: https://projects.blender.org/blender/blender/pulls/140168
Prior to this commit, the sculpt scene and object was only initialized
once, at the very beginning of the test. This has the downside of slowly
changing the performance characteristics as more and more strokes are
performed, as the number of vertices within a given stroke may
eventually approach 1. To ensure a consistent measurement each time,
rebuild the scene between each brush stroke.
Pull Request: https://projects.blender.org/blender/blender/pulls/139960
This commit lowers the size of the mesh from approximately 2 million
verts to 250 thousand verts. This brings the graph more in line with the
mesh and multires usecase, and is more helpful in measuring and
detecting real performance issues.
While the upper end of dyntopo can certainly be massively improved,
strokes that take in the seconds to complete are already unusable from a
user perspective,
Usually, this enum should only ever be compared with `==`.
Confusingly, however, to check if a strip is an effect, one must
'bitwise-and' it instead. This can backfire if e.g. one tries to do
(strip->type & STRIP_TYPE_IMAGE) which doesn't work.
Make sure we only ever 'and' against the type enum when checking if
it is an effect. There was only this one case that didn't adhere.
Prior to this commit, the `object.subdivision_set` would prevent
actually applying a subdivision to a multires modifier when the relative
option was set. This is commonly accessed via Alt-1 / Alt-2 or D / Shift
D in Sculpt mode.
However, when the multires modifier did not already exist, pressing
these keybinds would still create the modifier and further subdivide the
mesh.
To fix this, this commit adds a hidden property to the operator:
`ensure_modifier` to indicate if the keybind should create the modifier
or not.
Pull Request: https://projects.blender.org/blender/blender/pulls/130254
Partial revert of 5102880f51. That commit decreased pen tablet drag
threshold as pressure is added. Unfortunately, although this allowed a
quicker response to start of drag, this also increased the difficultly
of initiating a press when both actions are available. With numerical
inputs for example, although dragging is more responsive clicking into
the input to edit was more difficult.
Pull Request: https://projects.blender.org/blender/blender/pulls/140066
Introduced with 23951e1b12
The multiplane scrape brush uses two separate distances, one in world
space and the other in local brush space. Both need to be filtered on
for determining the brush strength.
Pull Request: https://projects.blender.org/blender/blender/pulls/140143
Similar to the recent fix 457cccd964.
There was another code path which could pass in the wrong time into USD
for certain Mesh Sequence Cache scenarios.
I was not able to craft a faulty scenario by hand to observe a real
problem though. The scenario begins by importing a USD file needing a
Mesh Sequence Cache modifier and then attaching a particle system (like
Hair) to the object. This will trigger the specific check calling into
`can_use_mesh_for_orco_evaluation` with the wrong time.
This makes the code path more explicit and passes in the correct time to
USD regardless now e.g. frame 28 vs time 1.166666666666667 (24fps)
Pull Request: https://projects.blender.org/blender/blender/pulls/140092
Prevent race conditions caused by calling `GPUWorker::wake_up` when the
worker is not waiting.
Found to be an issue in #139627, since `wake_up` is likely to be called
before the thread has fully started.
Pull Request: https://projects.blender.org/blender/blender/pulls/139842
When pressing Ctrl to snap the playhead while scrubbing,
Blender could crash while trying to snap to the first key.
This would happen if the current frame was higher than the
left keyframe, but the difference was less than the `BEZT_BINARYSEARCH_THRESH`.
Pull Request: https://projects.blender.org/blender/blender/pulls/140122
It is generally expected that items within the same section in the Node
Add menu appear in alphabetical order.
However, certain handful of places in the different node editor menus
don't seem to follow this convention. This patch simply re-arranges the
items in some places to follow alphabetical order.
Pull Request: https://projects.blender.org/blender/blender/pulls/140070
In Grease Pencil, when using the tools in Sculpt Mode and Vertex Paint
mode, the check on editable layers wasn't entirely accurate. There was
a strict check on an editable _active_ layer, but since the tools work
on _all_ editable layers, the check should be wider: if there is _any_
editable layer, the tool can work.
That is fixed in this PR. Now the tools can be used when, for example,
a layer group is active or when the active layer is hidden, but there
are other editable layers present.
In Vertex Paint mode there was an additional issue: with Auto Keying
enabled, a new keyframe was created for the active layer only. A new
keyframe should be created for _every_ editable layer. As is the case
now with this PR.
Pull Request: https://projects.blender.org/blender/blender/pulls/140119
By definition, Bounds for single points (size zero) are empty (this
matches BLI_rct behavior), so doing an intersect will actually fail.
So to resolve, use the existing `is_point_inside_bounds` for single-
point-curves.
Pull Request: https://projects.blender.org/blender/blender/pulls/140124
e934792169 introduced a workaround for
NVIDIA windows where NVIDIA drivers fail to allocate swapchain images
of minimized windows. The fix was to clamp it to 1,1. With this clamping
AMD driver seems to tell Blender that the created swapchain is suboptimal,
and needs to be recreated. This results in over and over creation of
swapchains as they are all considered sub-optimal.
This PR limits the clamping to NVIDIA drivers only.
Pull Request: https://projects.blender.org/blender/blender/pulls/140112
The `Eraser Hard` and `Eraser Point` were almost identical in behavior.
* `Eraser Hard` was set to `Disslove` making the edges soft, not hard.
Now this is set to `Point` mode to delete the points and make actual
hard edges.
* The `Eraser Point` is obsolete and was removed.
Code selecting objects in collections that have been instanced was only
handling objects directly owned by the collection, not the objects in
the hierarchy of children collections.
The objects from the whole hierarchy need to be hanlded here, since
children collections will not be processed (as they are already
instanced by their parent one).
Pull Request: https://projects.blender.org/blender/blender/pulls/140068
Previous implementation allows to move VKBuffers, but didn't do a
proper std::move. On second thought it is a bad idea to be able to move
GPU resources. This PR removes the ability to move the buffer and
replace the usages with a unique ptr.
Pull Request: https://projects.blender.org/blender/blender/pulls/140103
The color picking reads 3 channels from a 4 channel texture. In the
common case the data conversion is required. Data conversion happens on
all 4 channels, but the backed memory only has 3 channels.
This is a workaround as the actual solution needs to work in many
different situations. Previous solution had to many issues that we
reverted the solution. This PR adds a local workaround specific to color
picking.
In Blender 5.0 we want to add test cases for all situations and
implement a solution that works better.
Pull Request: https://projects.blender.org/blender/blender/pulls/140101
Currently, we are forced to use `reinterpret_cast` (or c-style-casts) when
converting between `ID` and the actual data-block types like `Object` and
`Mesh`. While `reinterpret_cast` does preserve constness, it does nothing to
make sure that a cast is actually valid. This patch adds `blender::id_cast`
which can be used as a drop-in replacement of `reinterpret_cast` in supported
cases.
It supports a couple of cases (using `Object` as example for all data-block
types):
* Convert `ID *` to `Object *`. This asserts that the conversion is valid at
run-time if the pointer is not null.
* Convert `ID &` to `Object &`. This always asserts that the conversion is valid
at run-time.
* Convert `Object *` to `ID *`. This is preferred over `&object->id` even if the
result is the same because the latter results in ASAN warnings if `object` is
null.
* Convert `Object &` to `ID &`. Added for good measure. There isn't really much
of a benefit of this over using `object.id` except for maybe in generic code
where the source type can be an `Object &` or `ID &`.
* Identity cast identical id types (e.g. `Object *` to `Object *`). Not really
necessary to add a cast in such a case but may be useful in generic code where
the input can be an `Object *` or `ID *`.
Some additional notes:
* Const-correctness is preserved with this cast.
* When attempting to do a non-id-cast, the function triggers a `static_assert`.
* The data-block types must not be just forward declarations because the actual
type information is necessary.
* Casting to and from `void *` is not supported because that can't provide any
additional type-safety over `static_cast`.
* Casting between different data-block types like `Object` and `Mesh` is not
allowed.
* It does not support casting double-pointer types currently (e.g. `ID **` to
`Object **`).
The new cast is added in `DNA_ID.h`. As part of this patch, I also sprinkled in
a couple of `id_cast` in various places for testing (it's surprisingly tricky to
have the type checks working for all the different combinations of
pointer/referenced/constness/etc.). Replacing more existing `reinterpret_cast`
can be done separately afterwards.
It's nice that `blender::id_cast` has exactly the same length as
`reinterpret_cast`. So formatting should never change when using it in global
scope. Using just `id_cast` when inside of the `blender::` namespace is
preferable of course.
Pull Request: https://projects.blender.org/blender/blender/pulls/139820
Only the collection clearing was removed, adding an item
and assigning values to it was kept - breaking the preset.
Detect collections and removed all lines which add & manipulate items.
Based on the fix from !139309 with fewer changes.
Also include FBX operators when cleaning up presets.
Co-authored-by: Guillermo Venegas <guillermovcra@gmail.com>
Presets would write a property multiple times in the case both the class
and it's parent defined a property.
While not a bug, it looks like an error. Only write values once.
The Ellipse/Box Mask nodes recently got their size clamped by their soft
limit, which was not previously the case, breaking compatibility and
limiting the node's use in some setups. The same goes for their
position. We fix this by allowing any size and position.
There have been a number of commits that have introduced regressions
in Sculpt mode where NaN begins to be propagated. While it is visually
very obvious that this is happening, this commit adds an assert so that
any automated testing with asserts on will also catch this issue.
Related to 23951e1b12
Pull Request: https://projects.blender.org/blender/blender/pulls/133992
Introduced with 23951e1b12
The paint brush uses either absolute or local distance values, mapping
this distance to a brush strength requires using different values for
`BKE_brush_calc_curve_factors` and cannot just use
`calc_brush_strength_factors`. To fix this, use the more generic method
instead to allow passing the correct radius.
Pull Request: https://projects.blender.org/blender/blender/pulls/140084
Introduced with 23951e1b12.
When using brushes that use a cube distance, some distances are set to
be float::max(). This breaks assumptions made inside the previously
linked commit, as the operations inside `BKE_brush_calc_curve_factors`
can easily cause this distance value to become infinity, which when
multiplied with the factor value of 0 results in NaN propagation in the
mesh.
To fix this, set the max distance in `calc_cube_distance` to 1.0f.
Pull Request: https://projects.blender.org/blender/blender/pulls/139965
On the Windows platform if you select "Automatic" for language, it
won't actually select correctly for the Chinese languages. This is
because our Windows code returns a locale like ""zh_CN" or "zh_TW",
not the "zh_HANS" and "zh_HANT" expected. This PR checks for language
code "zh" and will then will append "_HANT" if your 2-letter region
is TW (Taiwan), HK (Hong Kong), or MO (Macau), and "_HANS" for other
regions (like China, Malaysia, Singapore, etc).
Pull Request: https://projects.blender.org/blender/blender/pulls/140027
Depending on things like order or processing etc., appending
liboverrides data could generate error message in the console regarding
invalid override hierarchy data.
While this could probably be solved, it feels fairly brittle and risky
to directly make linked liboverride hierarchies locale anyway, so just
alsways make local copies of these on append.
Currently, there is the general rule that outputs must be above inputs in any
given panel. Since internally a panel toggle input is part of the panel that it
toggles and has to be the first item in a panel, it was not possible to have a
panel toggle for a panel with outputs.
This patch relaxes the rule for ordering of inputs and outputs for the special
case of panel toggles.
Pull Request: https://projects.blender.org/blender/blender/pulls/140054
Not only depend on the camera position but also on the other camera parameters.
This is important because otherwise Geometry Nodes won't be updated when e.g.
the focal length changes which is important when implementing e.g. camera
culling.
This already works when depending on a specific camera directly, but not when
depending on the active camera.
Pull Request: https://projects.blender.org/blender/blender/pulls/140046
To make this work, I had to add a new rna callback to get the default value for
an enum property at run-time. The same exists for other property types like
float and bool already.
Pull Request: https://projects.blender.org/blender/blender/pulls/140050
This fix make it so that the brush cursor is always the same size as
the drawn strokes.
The code logic is modified from `pixel_radius_to_world_space_radius`
in `grease_pencil_utils.cc`
Co-authored-by: Falk David <falk@blender.org>
Pull Request: https://projects.blender.org/blender/blender/pulls/139964
This allows to reduce the waiting time caused by
shader compilation on some GPU-driver combo.
A new settings in the User Preferences make it
possible to override the default amount of worker
threads and optionally use subprocesses.
We still use only one worker thread in cases where
there is no benefit with adding more workers
(like AMD pro driver and Intel windows).
It doesn't scale as much as subprocesses for material
shader compilation but that is for other reasons
explained in #139818.
Add some heuristic to avoid too much memory usage
and / or too many stalls.
Also add some heuristic to the default number of subprocess for
the platform that shows scalling.
Historically, multithreaded compilation was prevented by the
need of context per thread inside `DRWShader` module.
Also there was no good scaling at that time. But
nowadays numbers shows different results with
good scaling with reasonable amount of threads on many
platforms.
Even if we are going for vulkan in the next release
most of the legacy hardware will still use OpenGL for
a few other releases. So it is relevant to make this
easy improvement.
See pull request for measurements.
Pull Request: https://projects.blender.org/blender/blender/pulls/139821
Baking and storing simulation state within loops or closures is not supported.
Previously, attempting to use the bake node or simulation zone in such a zone
would just silently fail. Now there is an error on the node and the bake
settings are grayed out.
Pull Request: https://projects.blender.org/blender/blender/pulls/140041
