Age | Commit message (Collapse) | Author |
|
This commit replaces the `Mesh_Runtime` struct embedded in `Mesh`
with `blender::bke::MeshRuntime`. This has quite a few benefits:
- It's possible to use C++ types like `std::mutex`, `Array`,
`BitVector`, etc. more easily
- Meshes saved in files are slightly smaller
- Copying and writing meshes is a bit more obvious without
clearing of runtime data, etc.
The first is by far the most important. It will allows us to avoid a
bunch of manual memory management boilerplate that is error-prone and
annoying. It should also simplify future CoW improvements for runtime
data.
This patch doesn't change anything besides changing `mesh.runtime.data`
to `mesh.runtime->data`. The cleanups above will happen separately.
Differential Revision: https://developer.blender.org/D16180
|
|
|
|
This is the conventional way of dealing with unused arguments in C++,
since it works on all compilers.
Regex find and replace: `UNUSED\((\w+)\)` -> `/*$1*/`
|
|
Use function style casts in C++ headers & source.
|
|
|
|
|
|
To use function style cast '(unsigned char)x' can't be replaced by
'unsigned char(x)'.
|
|
Using the attribute name semantics from T97452, this patch moves the
selection status of mesh elements from the `SELECT` of vertices, and
edges, and the `ME_FACE_SEL` of faces to generic boolean attribute
Storing this data as generic attributes can significantly simplify and
improve code, as described in T95965.
The attributes are called `.select_vert`, `.select_edge`, and
`.select_poly`. The `.` prefix means they are "UI attributes",so they
still contain original data edited by users, but they aren't meant to
be accessed procedurally by the user in arbitrary situations. They are
also be hidden in the spreadsheet and the attribute list.
Until 4.0, the attributes are still written to and read from the mesh
in the old way, so neither forward nor backward compatibility are
affected. This means memory requirements will be increased by one byte
per element when selection is used. When the flags are removed
completely, requirements will decrease.
Further notes:
* The `MVert` flag is empty at runtime now, so it can be ignored.
* `BMesh` is unchanged, otherwise the change would be much larger.
* Many tests have slightly different results, since the selection
attribute uses more generic propagation. Previously you couldn't
really rely on edit mode selections being propagated procedurally.
Now it mostly works as expected.
Similar to 2480b55f216c
Ref T95965
Differential Revision: https://developer.blender.org/D15795
|
|
Smooth flag should come from the evaluated mesh, only selection and hidding
state should be mapped to the original bmesh.
Pre-existing issue revealed by refactor in b247588dc0f4.
|
|
Currently, when subdividing every single vertex on every loose edge,
Blender iterates over all other edges to find neighbors. This has
quadratic runtime and can be very slow. Instead, first create a
map of edges connected to each vertex.
With about 10000 edges, the performance goes from very slow to very
smooth in my tests. Because of the nature of quadratic runtime, the
improvement will depend massively on the number of elements.
The only downside to this is that the map will still be built when
there are only a couple loose edges, but that case is probably not
so common.
Differential Revision: https://developer.blender.org/D15923
|
|
Replace `mesh_attributes`, `mesh_attributes_for_write` and the point
cloud versions with methods on the `Mesh` and `PointCloud` types.
This makes them friendlier to use and improves readability.
Differential Revision: https://developer.blender.org/D15907
|
|
Follows existing naming for the most part, also use "num" as a suffix
in some instances (following our naming conventions).
|
|
Use `verts` instead of `vertices` and `polys` instead of `polygons`
in the API added in 05952aa94d33eeb50. This aligns better with
existing naming where the shorter names are much more common.
|
|
For copy-on-write, we want to share attribute arrays between meshes
where possible. Mutable pointers like `Mesh.mvert` make that difficult
by making ownership vague. They also make code more complex by adding
redundancy.
The simplest solution is just removing them and retrieving layers from
`CustomData` as needed. Similar changes have already been applied to
curves and point clouds (e9f82d3dc7ee, 410a6efb747f). Removing use of
the pointers generally makes code more obvious and more reusable.
Mesh data is now accessed with a C++ API (`Mesh::edges()` or
`Mesh::edges_for_write()`), and a C API (`BKE_mesh_edges(mesh)`).
The CoW changes this commit makes possible are described in T95845
and T95842, and started in D14139 and D14140. The change also simplifies
the ongoing mesh struct-of-array refactors from T95965.
**RNA/Python Access Performance**
Theoretically, accessing mesh elements with the RNA API may become
slower, since the layer needs to be found on every random access.
However, overhead is already high enough that this doesn't make a
noticible differenc, and performance is actually improved in some
cases. Random access can be up to 10% faster, but other situations
might be a bit slower. Generally using `foreach_get/set` are the best
way to improve performance. See the differential revision for more
discussion about Python performance.
Cycles has been updated to use raw pointers and the internal Blender
mesh types, mostly because there is no sense in having this overhead
when it's already compiled with Blender. In my tests this roughly
halves the Cycles mesh creation time (0.19s to 0.10s for a 1 million
face grid).
Differential Revision: https://developer.blender.org/D15488
|
|
This patch moves material indices from the mesh `MPoly` struct to a
generic integer attribute. The builtin material index was already
exposed in geometry nodes, but this makes it a "proper" attribute
accessible with Python and visible in the "Attributes" panel.
The goals of the refactor are code simplification and memory and
performance improvements, mainly because the attribute doesn't have
to be stored and processed if there are no materials. However, until
4.0, material indices will still be read and written in the old
format, meaning there may be a temporary increase in memory usage.
Further notes:
* Completely removing the `MPoly.mat_nr` after 4.0 may require
changes to DNA or introducing a new `MPoly` type.
* Geometry nodes regression tests didn't look at material indices,
so the change reveals a bug in the realize instances node that I fixed.
* Access to material indices from the RNA `MeshPolygon` type is slower
with this patch. The `material_index` attribute can be used instead.
* Cycles is changed to read from the attribute instead.
* BMesh isn't changed in this patch. Theoretically it could be though,
to save 2 bytes per face when less than two materials are used.
* Eventually we could use a 16 bit integer attribute type instead.
Ref T95967
Differential Revision: https://developer.blender.org/D15675
|
|
|
|
* Flip the logic to first detect if we are dealing with an unmodified mesh
in editmode. And then if not, detect if we need a mapping or not.
* runtime.is_original is only valid for the bmesh wrapper. Rename it to clarify
that and only check it when the mesh is a bmesh wrapper.
* Remove MR_EXTRACT_MAPPED and instead check only for the existence of the
origindex arrays. Previously it would sometimes access those arrays without
MR_EXTRACT_MAPPED set, which according to a comment means they are invalid.
Differential Revision: https://developer.blender.org/D15676
|
|
This commit moves the hide status of mesh vertices, edges, and faces
from the `ME_FLAG` to optional generic boolean attributes. Storing this
data as generic attributes can significantly simplify and improve code,
as described in T95965.
The attributes are called `.hide_vert`, `.hide_edge`, and `.hide_poly`,
using the attribute name semantics discussed in T97452. The `.` prefix
means they are "UI attributes", so they still contain original data
edited by users, but they aren't meant to be accessed procedurally by
the user in arbitrary situations. They are also be hidden in the
spreadsheet and the attribute list by default,
Until 4.0, the attributes are still written to and read from the mesh
in the old way, so neither forward nor backward compatibility are
affected. This means memory requirements will be increased by one byte
per element when the hide status is used. When the flags are removed
completely, requirements will decrease when hiding is unused.
Further notes:
* Some code can be further simplified to skip some processing when the
hide attributes don't exist.
* The data is still stored in flags for `BMesh`, necessitating some
complexity in the conversion to and from `Mesh`.
* Access to the "hide" property of mesh elements in RNA is slower.
The separate boolean arrays should be used where possible.
Ref T95965
Differential Revision: https://developer.blender.org/D14685
|
|
This loosens the current implementation a bit to only force optimal
display when editing on cage. It used to be any editing mode.
Brings GPU based subdivision closer to the CPU version.
|
|
openSubdiv_init() would detect available evaluators before any OpenGL context
exists, causing a crash with libepoxy. This test however is redundant as we
already check the requirements on the Blender side through the GPU API.
To simplify things, completely remove the device detection in the opensubdiv
module and reduce the evaluators to just CPU and GPU. The plan here is to move
to the GPU module abstraction over OpenGL/Metal/Vulkan and so all these
different backends no longer make sense.
This also removes the user preference for OpenSubdiv compute device, which was
not used for the new GPU subdivision implementation.
Ref D15291
Differential Revision: https://developer.blender.org/D15470
|
|
|
|
Pass `use_hide` to the compute shaders so that we can override the
hidden face flags, like CPU extraction is doing.
|
|
Addendum to previous fix, which was for point selection, this fixes the
face selection mode. The issue is caused by wrong flags used for paint
mode (the edit mode flag was always used). Also add back flag which was
accidentally removed in 16f5d51109bce849dff5379c60360f271622ac0f.
|
|
|
|
Instancing with geometry nodes uses just the evaluated Mesh, and ignores the
Object that it came from. That meant that it would try to look up the subsurf
modifier on the instancer object which does not have the subsurf modifier.
Instead of storing a session UUID and looking up the modifier data, store a
point to the subsurf modifier runtime data. Unlike the modifier data, this
runtime data is preserved across depsgraph CoW. It must be for the subdiv
descriptor contained in it to stay valid along with the draw cache.
As a bonus, this moves various Mesh_Runtime variables into the subsurf runtime
data, reducing memory usage for meshes not using subdivision surfaces.
Also fixes T98693, issues with subdivision level >= 8 due to integer overflow.
Differential Revision: https://developer.blender.org/D15184
|
|
The normals flags were not setup properly which made normals for all
elements (vertices, faces) to be drawn when using the normals overlay.
Also remove usage of uints for the flag in the APIs.
|
|
After this commit, all mesh data extraction and drawing code is in C++,
including headers, making it possible to use improved types for future
performance improvements and simplifications.
The only non-trivial changes are in `draw_cache_impl_mesh.cc`,
where use of certain features and macros in C necessitated larger
changes.
Differential Revision: https://developer.blender.org/D15088
|
|
|
|
|
|
Simple typo in rB55e3930b253e.
|
|
|
|
|
|
|
|
Although reusing the same patch coordinate for all corner pointing the
same vertex works for interpolation vertices, it does work for
interpolation face varying attributes. So we need to keep the original
patch coordinates around for face varying interpolation. This was caused
by the previous fix (a5dcae0c641604c033f852e41841f58460c40069).
|
|
|
|
Issues stems from the mesh not being watertight. This was caused by
floating point precision issues when evaluating patch coordinates at
patch boundaries (loops/corners in different patches pointing to the same
vertex). To fix this we ensure that all loops pointing to the same vertex
share the same patch coordinate. This keeps code simple, and does not
require to track precision issues in floating point math all over the
place.
|
|
|
|
The face dots normals may not be always requested, thus leading to a
crash by null pointer dereference.
|
|
|
|
Faces, edges, and vertices are still shown when GPU subdivision is
actived. This is because the related edit mode flags were ignored by the
subdivision code.
The flags are now passed to the various compute shaders mostly as part of
the extra coarse data, also used for e.g. selection. For loose edges, a
temporary buffer is created when extracting them. Loose vertices are
already taken into account as it reuses the routines for coarse mesh
extraction, although `MeshRenderData.use_hide` was not initialized,
which is fixed now.
|
|
This uses the recently introduced evaluator's vertex
data to smoothly interpolate original coordinates instead
of using linear interpolation.
The orcos are interpolated at the same time as positions
and as such, the specific subdivision routine for the
orco extractor has been removed. The patch evaluation
shader uses a definition to enable code specific to
orco evaluation.
Since the orco layer may not have been requested on first
render, and since orco data is now stored in the OpenSubDiv
evaluator, the evaluator needs to be recreated if an
orco layer is suddenly available. For this, a callback
to check if the evaluator has the data was added. This is
added to the evaluator as the `Subdiv` cache stored in the
modifier is invalidated less often than the Mesh batch cache
and so leads to fewer evaluator recreations.
Differential Revision: https://developer.blender.org/D14999
|
|
Knowing when layers are retrieved for write access will be essential
when adding proper copy-on-write support. This commit makes that
clearer by adding `const` where the retrieved data is not modified.
Ref T95842
|
|
The mesh drawing code used a different mesh to check whether or not to
draw face dots and to actually retrieve them. The fix is moving the
responsibility of determining whether to use subsurf face dots to the
creation of `MeshRenderData` where the mesh used for drawing is
known, rather than doing it at a higher level.
Differential Revision: https://developer.blender.org/D14855
|
|
The mesh drawing code used a different mesh to check whether or not to
draw face dots and to actually retrieve them. The fix is moving the
responsibility of determining whether to use subsurf face dots to the
creation of `MeshRenderData` where the mesh used for drawing is
known, rather than doing it at a higher level.
Differential Revision: https://developer.blender.org/D14855
|
|
The coarse polygon count was set to the one of the BMesh instead of
the the one of the mesh used for subdivision, which caused the
compute shaders to output wrong data.
|
|
The crash is caused as the data for the UV editor is requested before
the data for the mesh as a separate draw update. Since building the UV
stretch angle buffer requires the position buffer, the latter is not
created yet in this case.
To fix this, create a local position buffer from the subdivision data. An
alternate fix was considered to remove the dependency on the position
buffer by interpolating on the GPU the coarse stretch angle buffer but
this did work. Maybe this will be revisited.
|
|
Simple port with a few cosmetic changes:
- Attribute named "color" for indices VBO is now called "index"
- The indices VBO is now composed of `int`s instead of `uint`s (this simplifies the source)
Differential Revision: https://developer.blender.org/D14800
|
|
The subdivision is always recomputed on the CPU when displaying stats
if the mesh is animated which leads to bad performance.
This caches the subdivision topology counters from the draw code in the
mesh runtime and uses them for the viewport statistics.
Differential Revision: https://developer.blender.org/D14774
|
|
|
|
When more than one, consecutive, subdivision modifier is used on a Mesh,
the last subsurf modifier is used for GPU subdivision even though it
might be disabled. This is because retrieving the last subsurf modifier
in the draw code did not check whether the modifier was disabled or not.
To fix this, the session UUID of the modifier which delegated evaluation
to the GPU code is cached and used in the draw to select the right subsurf
modifier.
Differential Revision: https://developer.blender.org/D14488
|