rBeed45d2a239a introduced a GPU backend for OpenSubDiv which lets us do
the subdivision at render time. However, some tools might still need to
have the subdivision data available on the CPU side. For this a
subdivision mesh wrapper was also introduced, and is computed whenever a
CPU side mesh is needed. The subdivision settings for this wrapper are
stored during modifier evaluation if GPU subdivision can be done.
The performance regression is due to the fact that although the
subdivision mesh was already computed on the CPU, and no subdivision
wrapper is generated, some checks for creating subdivision data in
`BKE_mesh_wrapper_ensure_subdivision` where still run, one of which is
very expensive.
To fix this we first check the runtime settings of the mesh to see if
subdivision is needed at all.
This evaluator is used in order to evaluate subdivision at render time, allowing for
faster renders of meshes with a subdivision surface modifier placed at the last
position in the modifier list.
When evaluating the subsurf modifier, we detect whether we can delegate evaluation
to the draw code. If so, the subdivision is first evaluated on the GPU using our own
custom evaluator (only the coarse data needs to be initially sent to the GPU), then,
buffers for the final `MeshBufferCache` are filled on the GPU using a set of
compute shaders. However, some buffers are still filled on the CPU side, if doing so
on the GPU is impractical (e.g. the line adjacency buffer used for x-ray, whose
logic is hardly GPU compatible).
This is done at the mesh buffer extraction level so that the result can be readily used
in the various OpenGL engines, without having to write custom geometry or tesselation
shaders.
We use our own subdivision evaluation shaders, instead of OpenSubDiv's vanilla one, in
order to control the data layout, and interpolation. For example, we store vertex colors
as compressed 16-bit integers, while OpenSubDiv's default evaluator only work for float
types.
In order to still access the modified geometry on the CPU side, for use in modifiers
or transform operators, a dedicated wrapper type is added `MESH_WRAPPER_TYPE_SUBD`.
Subdivision will be lazily evaluated via `BKE_object_get_evaluated_mesh` which will
create such a wrapper if possible. If the final subdivision surface is not needed on
the CPU side, `BKE_object_get_evaluated_mesh_no_subsurf` should be used.
Enabling or disabling GPU subdivision can be done through the user preferences (under
Viewport -> Subdivision).
See patch description for benchmarks.
Reviewed By: campbellbarton, jbakker, fclem, brecht, #eevee_viewport
Differential Revision: https://developer.blender.org/D12406
Adding a mirror modifier in edit-mode crashed.
Freeing meshes that hold a shallow copy happens in multiple places
while calculating modifiers, making it impractical to clear the
edit-mode pointer before freeing the mesh (as done in
BKE_editmesh_free_derived_caches).
Add a struct member to the edit-mesh struct so evaluated copies
don't free the edit-mesh contents.
- BKE_mesh_copy_parameters_for_eval to be used for evaluated meshes only
as it doesn't handle ID user-counts.
- BKE_mesh_copy_parameters is a general function for copying parameters
between meshes.
This uses mesh's runtime mutex for both `BKE_mesh_runtime_looptri_ensure`
(was using its own global RW mutex before), and `BKE_mesh_wrapper_ensure_mdata`
(was not protected at all before).
This is more like a band-aid than a proper fix, as mentioned in the report
proper fix would be for the modifier to request those data (the relevant
BVHTree, which would implicitely also call the tow others) through flags,
just like it does for regular CDData layers. But this is a much bigger
refactor to be done outside of bugfix scope.
Reviewed By: sergey
Maniphest Tasks: T78285
Differential Revision: https://developer.blender.org/D8415
This resolves a performance regression in 2.8x where every edit-mode
update performed an edit-mesh to mesh conversion.
Now the conversion will be lazily initialized if/when it's required.
New BKE_mesh_wrapper_* functions abstract over mesh data access.
Currently only edit-mesh and regular meshes are supported.
In the future sub-surface meshes may be supported too.
