This reverts commit 2ddecfffc3.
The correct fix is to compile with -fno-strict-aliasing on
release/relwithdebinfo builds also.
See the last commit/ {D9372}.
Maniphest Task T81077
This patch improves the single core performance of the lattice deform.
1. Prefetching deform vert during initialization. This data is constant for
each innerloop. This reduces the complexity of the inner loop what makes
more CPU resources free for other optimizations.
2. Prefetching the Lattice instance. It was constant. Although performance
wise this isn't noticeable it is always good to free some space in the
branch prediction tables.
3. Remove branching in all loops by not exiting when the effect of the loop
isn't there. The checks in the inner loops detected if this loop didn't
have any effect on the final result and then continue to the next loop.
This made the branch prediction unpredictable and a lot of mis
predictions were done. For smaller inner loops it is always better
to remove unpredictable if statements by using branchless code patterns.
4. Use SSE2 instruction when available.
This gives 50% performance increase measured on a
Intel(R) Core(TM) i7-6700 CPU @ 3.40GHz with GCC 9.3.
Also check other compilers.
Before:
```
performance_no_dvert_10000 (4 ms)
performance_no_dvert_100000 (30 ms)
performance_no_dvert_1000000 (268 ms)
performance_no_dvert_10000000 (2637 ms)
```
After:
```
performance_no_dvert_10000 (3 ms)
performance_no_dvert_100000 (21 ms)
performance_no_dvert_1000000 (180 ms)
performance_no_dvert_10000000 (1756 ms)
```
Reviewed By: Campbell Barton
Differential Revision: https://developer.blender.org/D9087
This looks like a optimizer bug where it makes wrong assumptions.
The code inside lib_id_delete:264 on rBafd13710b897cc1c11b
`for (id = last_remapped_id->next; id; id = id->next) {..}`
is not executed in release/relwithdebinfo builds.
This can be "fixed" by several ways:
- Adding a line that prints the `last_remapped_id->name` right before
the said for-loop starts.
- Turning off optimization for the whole function `id_delete`:
`#pragma clang optimize off/on` Ray Molenkamp
- Marking `last_remapped_id` volatile. Julian Eisel
- Marking `tagged_deleted_ids` volatile. But it adds a warning when
calling `BLI_addtail`: discards volatile qualifier. Discovered by
accident.
Fix T81077
Reviewed By: mont29
Maniphest Tasks: T81077
Differential Revision: https://developer.blender.org/D9315
The solo mode was skipping the layer creation data and the loop of masks expect to have all layers in the array or the loop crash.
The solution is just create the layer array data for the layer, but don't draw any stroke.
Selecting an F-Curve handle caused an assertion as well as treating
the key-frame as inactive.
Allow active the keyframe to be active when it's handle is selected,
as is done with bezier curves.
The new preset I made for 2.91 is way more controllable with lower
strength values and does not have the accumulate bug, but until the
brush management is in place to ship multiple versions of the brush,
probably most people expect something closer to the old version to be
the default.
Reviewed By: sergey
Maniphest Tasks: T81901
Differential Revision: https://developer.blender.org/D9289
Being able to adjust the distance between fluid and obstacles comes in handy when trying to achieve a fluid motion over inclined obstacles.
Depending on the slope of such obstacles, already small adjustments of this value can help when particles stick to obstacle surfaces (i.e. make particles not stick to obstacles).
In comments made by a tester on rB17b89f6dacba007bf, it seems that baking
of the spray maps would be useful. This commit adds that capability. Both
the spray map and its inverse are baked out in this change, for maximum
convenience and to avoid assuming what the user wants.
Differential Revision: https://developer.blender.org/D8470
The design for how we approach the "Everything Nodes" project
has changed. We will focus on a different part of the project initially.
While future me will likely refer back to some of the code I remove here,
there is no point in keeping this code around in master currently.
It would just confuse other developers working on the project.
This does not remove the simulation modifier and data block. Those are
just cleaned up, so that the boilerplate code can be reused in the future.
This changes how the simplify volumes setting works. Before, it only
affeted viewport rendering. This was an issue, because all internal
computations would still have to happen on the high resolution volumes.
With this patch, the simplify setting already affects file loading and
procedural generation of volumes.
Rendering does not have to care about the simplify option anymore,
it just gets the correct simplified version from the depsgraph.
Reviewers: brecht
Differential Revision: https://developer.blender.org/D9176
Corrects incorrect usages of the fragment 'apart of' when 'a part of' was required.
Differential Revision: https://developer.blender.org/D9245
Reviewed by Campbell Barton
Corrects incorrect usages of the word 'loose' when 'lose' was required.
Differential Revision: https://developer.blender.org/D9243
Reviewed by Campbell Barton
Corrects incorrect usage of contraction for 'it is', when possessive 'its' was required.
Differential Revision: https://developer.blender.org/D9250
Reviewed by Campbell Barton
This modifier is the opposite of the recently added Mesh to Volume modifier.
It converts the "surface" of a volume into a mesh. The "surface" is defined
by a threshold value. All voxels with a density higher than the threshold
are considered to be inside the volume, while all others will be outside.
By default, the resolution of the generated mesh depends on the voxel
size of the volume grid. The resolution can be customized. It should be
noted that a lower resolution might not make this modifier faster. This
is because we have to downsample the openvdb grid, which isn't a cheap
operation.
Converting a mesh to a volume and then back to a mesh is possible,
but it does require two separate mesh objects for now.
Reviewers: brecht
Differential Revision: https://developer.blender.org/D9141
Field time_base of video stream must be set for some containers,
otherwise avformat_write_header() will set it to default values.
Rendered file in such case won't be played at desired frame rate.
See init_muxer() in mux.c in ffpmeg sources.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D9213
A general refactor / fix commit that should clear out the issues that have been reported on external forces and moving effectors (e.g. T79537, T81660, T80088).
Code could call CustomData_get_layer_index_n with a negative index (if
no active and/or render UV layers are found). This would assert since
rBe86785c51445.
Spotted while looking into T81398.
Differential Revision: https://developer.blender.org/D9212
The new parameter made so that previously keyed Alpha values were lost
and instead the new "Emission Strength" was keyed.
Issue introduced with the original commit of Emission Strength: b248ec9776
Note: Files created since the issue (September 17) that keyframed the
Emission Strength will have to fix their files manually.
Differential Revision: https://developer.blender.org/D9221
Restore the old `correct_bezpart()` (pre-rBda95d1d851b4) function as
`BKE_curve_correct_bezpart()`, and use that where the old behaviour was
desired (that is, curve maps like used by the RGB Curves shader node).
The new (post-rBda95d1d851b4) function is also renamed to
`BKE_fcurve_correct_bezpart()` to avoid confusion.
Previously, all Face Set visibility logic was using mvert flags directly
to store the visibility state on the vertices while sculpting. As Face
Sets are a poly attribute, it is much simpler to use mpoly flags and let
BKE_mesh_flush_hidden_from_polys handle the vertex visibility, even for
Multires.
Now all operators that update the Face Set visibility state will always
copy the visibility to the mesh (using poly flags) and the grids, all
using the same code.
This should fix a lot of visibility glitches and bugs like the following:
- Sculpt visibility reset when changing multires levels.
- Multires visibility not updating in edit mode.
- Single face visibible when surrounded by visibile face set, even when
the face set was hidden.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D9175
ME_POLY_LOOP_NEXT and ME_POLY_LOOP_PREV expect the offset of
the loop in the poly as an argument, in other words, corner index of the poly.
This was violated in some places. It didn't cause issues when base mesh consists
of only quads due to the way how modulus worked inside of the macro. However,
if mesh had non-quad faces adjacency information was returning wrong vertex
indices. This was causing multiple brushes to work erratically, including brushes
like Face Set, Boundary automasking, mesh relax, and others.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D9173
This adds an operator property to use the paint cursor radius and
position for the depth of the trimming shape created by the trimming
tools.
When enabled, the shape is located in the surface point when the gesture
started and it will have the depth of the cursor radius. When the cursor
is not over the mesh, the shape will be positioned at the center of the
depth of the whole object from the viewport camera.
Reviewed By: dbystedt, sergey
Differential Revision: https://developer.blender.org/D9129
The voxel remesher was using the voxel size to limit the shrink-wrap
projection distance. Now that distance is increased to help preserving
more detail on hard surface edges.
Reviewed By: pablodp606
Differential Revision: https://developer.blender.org/D6204
Being in render 'context'was not taken into account in code evaluating
modifiers for meshes in Edit mode.
Reviewed By: #modeling, mont29
Differential Revision: https://developer.blender.org/D9217
When creating a particle system to display simulated particles, the phystype needs to be set to 'no physics' so that particle positions are just copied and not integrated.
This will make the "Reset to Default Value" operator in button right
click menus work for the fluid modifier. Before they always reset
the values to 0.
Differential Revision: https://developer.blender.org/D9206
With constructive + deform modifiers, loop-cut visualization
wasn't following the displayed mesh.
This now gets the coordinates from the cage when available.
