color attribute system.
This commit removes sculpt colors from experimental
status and unifies it with vertex colors. It
introduces the concept of "color attributes", which
are any attributes that represents colors. Color
attributes can be represented with byte or floating-point
numbers and can be stored in either vertices or
face corners.
Color attributes share a common namespace
(so you can no longer have a floating-point
sculpt color attribute and a byte vertex color
attribute with the same name).
Note: this commit does not include vertex paint mode,
which is a separate patch, see:
https://developer.blender.org/D14179
Differential Revision: https://developer.blender.org/D12587
Ref D12587
Instead of using `CurveEval` to draw the curve wire edges, use
the new `Curves` data-block, which is already built as part of
an object's evaluated geometry set whenever there is a
`CurveComponent`.
This means that we can remove `Curve`'s temporary ownership
of `CurveEval` for drawing (added in 9ec12c26f1),
which caused a memory leak as described in T96498.
In my testing this improved performance by around 1.5x during
viewport playback, back to the performance of 3.1 before the
curve data structure transition started.
The next step of using the GPU to do the final curve evaluation
for the viewport is described in T96455, but is unrelated.
Differential Revision: https://developer.blender.org/D14551
This commit furthers some of the changes that were started in
rBb9febb54a492 and subsequent commits by changing the way surface
objects are presented to render engines and other users of evaluated
objects in the same way. Instead of presenting evaluated surface objects
as an `OB_SURF` object with an evaluated mesh, `OB_SURF` objects
can now have an evaluated geometry set, which uses the same system
as other object types to deal with multi-type evaluated data.
This clarification makes it more obvious that lots of code that dealt
with the `DispList` type isn't used. It wasn't before either, now it's
just *by design*. Over 1100 lines can be removed. The legacy curve
draw cache code is much simpler now too. The idea behind the further
removal of `DispList` is that it's better to focus optimization efforts
on a single mesh data structure.
One expected functional change is that the evaluated mesh from surface
objects can now be used in geometry nodes with the object info node.
Cycles and the OBJ IO tests had to be tweaked to avoid using evaluated
surface objects instead of the newly exposed mesh objects.
Differential Revision: https://developer.blender.org/D14550
Also solves two warnings from the previous similar commit,
f688e3cc31. The change to the grease pencil
modifier is quite suspicious, but doesn't change the behavior,
which was already broken.
This function was not used for anything other than mat4. This
was because of a limitation of the DRW module/
This makes it cleaner for the GLSL and also less tempting to use
it for other unconventional purpose.
This was caused by the recent changes made to the way we handle matrix
copies. The matrix copy assumed that the uniform iteration was the same
as creation order. But this was far from true. The reality was that
the iterator was reverse for `unichunk` but not for `unichunk->uniforms`
so this was recreating wrong matrix.
I rewrote this part to always use reverse iteration and fix the
copy destination.
Also I simplified the code making the assumption this won't be used for
anything else than mat4.
- Metal uniform array compatibility in DRW module.
- Guard OpenGL-specific workarounds and flushes behind GPU_type_matches_ex API guard. Add further render boundaries for render paths called outside of the main loop.
Authored by Apple: Michael Parkin-White
Ref: T96261
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D14438
MSL does not have an implicit global scope, this is emulated via macro's adding an indirection for uniforms, attributes, shader stage inputs and outputs such as:
#define roughness shaderinst->roughness.
Variables in GLSL which exist within uniform blocks can be directly referenced via the global scope, unlike standard C++. This means that variable name pollution occurs if subsequent local variables in the code use the same name, resulting in compilation errors.
A number of these conflicting names have been renamed to ensure unique naming and no further scope pollution.
Ref T96261
Reviewed By: fclem
Maniphest Tasks: T96261
Differential Revision: https://developer.blender.org/D14452
The crash happens because the origindex layers created as part of
the modifier stack evaluation are not set in the `MeshRenderData` when
they should have been.
This is because when selecting in X-ray mode, a subdivision wrapper
is created to ensure that selection happens with a subdivided
geometry, and this replaces the `MDATA` wrapper which is also used to
setup the `MeshRenderData`.
As we do not seemingly have an `MDATA` wrapper, the draw code decides
that we can extract draw buffers directly from the BMesh, instead of
the mapped Mesh with origin indices layers.
To fix this, we should also consider to use mapped extraction if a
subdivision wrapper exists on the mesh.
Differential Revision: https://developer.blender.org/D14485
The lines paint mask IBO extraction was not implemented for GPU subdivision.
For it to work, we also now need to preserve the subdivision loop to
subdivision edge map, which until now was overwritten to store coarse edges
(the map to coarse edges is still preserved).
Also the paint flag stored in the 4th dimension of the loop normal buffer
was not properly set for flat shaded faces, leading to other kind of
artefacts and render issues.
Each time the user clicks the viewport 2 sets of engine views are
created. Each set is currently composed of 8 view objects, each of size
592 bytes.
Because space is not reserved in the vector that holds them, several
unnecessary re-allocation/copy cycles occur as the vector resizes and
the total allocation load is 8880 bytes. This happens twice.
Reduce to just the allocations necessary and with exactly 4736 bytes
allocated for each set
- Before: 8 allocations and 8 deallocations totaling 17760 bytes
- After: 2 allocations and 2 deallocations totaling 9472 bytes
Reviewed By: fclem, jbakker
Differential Revision: https://developer.blender.org/D13782
Rename "size" variables and functions to use "num" instead,
based on T85728 (though this doesn't apply to simple C++
containers, it applies here). Rename "range" to "points" in
some functions, so be more specific.
Differential Revision: https://developer.blender.org/D14431
Vertices were not drawn properly as the logic for mapped mesh was used
in the BMesh case.
Edge display would ignore subdivided edges which would come from coarse
edges when setting display flags.
Metal shading language follows the C++ 14 standard and in some cases requires a greater level of explicitness than GLSL. There are also some small language differences:
- Explicit type-casts (C++ requirements)
- Explicit constant values (C++ requirements, e.g. floating point values using 0.0 instead of 0).
- Metal/OpenGL compatibility paths
- GLSL Function prototypes
- Explicit accessors for vector types when sampling textures.
Authored by Apple: Michael Parkin-White
Ref T96261
Reviewed By: fclem
Maniphest Tasks: T96261
Differential Revision: https://developer.blender.org/D14378
This is quite a huge cleanup. Making use of the `common_gpencil_lib.glsl`
to share more codes and use more consistent codestyle.
The gpencil engine specifics are now out of the `gpencil_vertex()`
function making it easier to add more features.
There should be no regression as all workarounds are kept as is.
This library contains the needed functions to render GPencil object
geometry. Centralizing these will make it possible for other draw
engines (EEVEE, Overlay) to reuse the same vertex shader code and
possibly the same fragment rejection methods.
This adds a simple and more manageable temp texture behavior.
The texture is garanteed to be available only between the acquire/release pair.
This makes the same engine able to reuse the textures and even overlap the acquire
& release calls.
- Add name support to storage buffers
- Delete view functions for TextureFromPool
- Add support for different size acquire and assert on mulitple acquire
- Allow multiple release
This is supposed to hold the latest improvement from the EEVEE rewrite branch.
Note that a restart is necessary in order for the engine to appear.
The registration code is a bit convoluted as it needs to be after the WM_init.
