**Problem Description**
Blender's current mesh data layout often lacks spatial coherence,
causing performance bottlenecks during BVH construction for sculpting
and painting operations. Each time a BVH is built, the system must
recompute spatial partitioning and vertex groupings from scratch,
leading to redundant calculations and suboptimal memory access patterns.
**Proposed Solution**
This patch implements pre-computed spatial organization of mesh data
through a new `mesh_apply_spatial_organization()` function that:
- Reorders vertices and faces based on spatial locality using recursive
spatial partitioning.
- Stores pre-computed MeshGroup hierarchies in MeshRuntime for reuse.
- Enables the BVH system to bypass expensive spatial computation when
pre-organized data is available.
This approach separates the expensive spatial computation from more
frequent BVH rebuilds, providing sustained performance improvements
across multiple sculpting operations.
**Limitations**
- Requires manual invocation (occurs automatically only during remesh
operations).
- Additional memory overhead for storing MeshGroup metadata.
- One-time computational cost during initial organization.
- Spatial group references are not yet stored in files.
**User Interface**
The feature is accessible via a new "Reorder Mesh Spatially" operator in
the Mesh Data Properties panel under the Geometry Data section. Users
can invoke it manually when needed, or it will be applied automatically
during quadriflow and voxel remesh operations. The operator provides
feedback confirming successful spatial reordering.
Pull Request: https://projects.blender.org/blender/blender/pulls/139536
Add a "dumb vector" storage option for custom normals, with the
"custom_normal" attribute. Adjust the mesh normals caching to
provide this attribute if it's available, and add a geometry node to
store custom normals.
## Free Normals
They're called "free" in the sense that they're just direction vectors
in the object's local space, rather than the existing "smooth corner
fan space" storage. They're also "free" in that they make further
normals calculation very inexpensive, since we just use the custom
normals instead. That's a big improvement from the existing custom
normals storage, which usually significantly decreases
viewport performance. For example, in a simple test file just storing
the vertex normals on a UV sphere, using free normals gives 25 times
better playback performance and 10% lower memory usage.
Free normals are adjusted when applying a transformation to the entire
mesh or when realizing instances, but in general they're not updated for
vertex deformations.
## Set Mesh Normal Node
The new geometry node allows storing free custom normals as well as
the existing corner fan space normals. When free normals are chosen,
free normals can be stored on vertices, faces, or face corners. Using
the face corner domain is necessary to bake existing mixed sharp and
smooth edges into the custom normal vectors.
The node also has a mode for storing edge and mesh sharpness, meant
as a "soft" replacement to the "Set Shade Smooth" node that's a bit
more convenient.
## Normal Input Node
The normal node outputs free custom normals mixed to whatever domain is
requested. A "true normal" output that ignores custom normals and
sharpness is added as well.
Across Blender, custom normals are generally accessed via face and
vertex normals, when "true normals" are not requested explicitly.
In many cases that means they are mixed from the face corner domain.
## Future Work
1. There are many places where propagation of free normals could be
improved. They should probably be normalized after mixing, and it
may be useful to not just use 0 vectors for new elements. To keep
the scope of this change smaller, that sort of thing generally isn't
handled here. Searching `CD_NORMAL` gives a hint of where better
propagation could be useful.
2. Free normals are displayed properly in edit mode, but the existing
custom normal editing operators don't work with free normals yet.
This will hopefully be fairly straightforward since custom normals
are usually converted to `float3` for editing anyway. Edit mode
changes aren't included here because they're unnecessary for the
procedural custom normals use cases.
3. Most importers can probably switch to using free normals instead,
or at least provide an option for it. That will give a significant
import performance improvement, and an improvement of Blender's
FPS for imported scenes too.
Pull Request: https://projects.blender.org/blender/blender/pulls/132583
The core issue was that the geometry batch cache (e.g. `MeshBatchCache` or
`PointCloudBatchCache`) was dependent on the object. This is problematic when
the the same geometry is used with multiple different objects because the cache
can't be consistent with all of them.
Fortunately, the only thing that was retrieved from the object was the number of
material slots, so if that can be avoided we should be fine. We can't just use
the number of material slots stored on the geometry because that may have no
material slots but still has material indices which are overridden on the object
level.
The solution is to take make the number of materials for a geometry only
dependent on the actual `material_index` attribute and not on the number of
available slots. More specifically, we find the maximal referenced material
index and handle that many materials. This number does not depend on how many
material slots there are on the object, but it still allows the object to
override materials slots that the mesh references.
A downside is that the maximum material index has to be computed which often
requires an iteration over the mesh. Fortunately, we can cache that quite easily
and the computation can be done in parallel. Also we are probably able to
eagerly update the material index in many cases when it's set instead of
computing it lazily. That is not implemented in this patch though.
The largest part of the patch is making the maximal material index easily
available on all the geometry types. Besides that, the material API is slightly
replaced and the drawing code now makes use of the updated API.
Pull Request: https://projects.blender.org/blender/blender/pulls/133498
Avoid rebuilding BVH trees when meshes are copied.
Similar to the other uses of the shared cache system,
this can arbitrarily improve performance when meshes
are copied but not deformed and BVH building is the
main bottleneck. In a simple test file I got a 6x speedup.
The amount of code is also reduced and the system is
much simpler overall-- built out of common threading
patterns like `SharedCache` with its double-checked lock.
RAII is used in a few places to simplify memory management
too.
The downside is storing more `SharedCache` items in the
mesh runtime struct. That has a slight cost when copying
a small mesh many times, but we have ideas to improve that
in the future anyway (#104327).
Pull Request: https://projects.blender.org/blender/blender/pulls/130865
Addresses #121240.
Instead of allocating a new array and recalculating mesh triangulation
every time the user enters sculpt mode, reuse the mesh's existing cache.
Currently in order to avoid recalculating triangulation on every brush update
(which would typically be necessary because the triangulation direction
depends on vertex positions), add a mechanism to "freeze" the cache to
skip recalculations until the user exits sculpt mode. That even avoids
recalculation if vertex positions aren't affected. This is necessary because
we can't use the cache in a dirty state; tagging the cache dirty frees the
triangulation array.
Removing the duplicate triangles array reduces memory usage by 384 MB
in a 16 million vertex sculpt, and makes entering sculpt mode 125ms faster
(tested on a Ryzen 7840u).
In the long term, I hope we find a different solution that's a bit more
transparent and hopefully more integrated with the caching system
in general. In the meantime, this is a relatively safe low impact change
that helps document the needs for such a system anyway.
Pull Request: https://projects.blender.org/blender/blender/pulls/123638
BKE_mesh_validate is called by mesh.validate() Python API, as well as optionally
when doing file imports. This PR speeds it up a bit:
- Faster face data sorting by using parallel sort instead of qsort,
- One allocation for all face vertex indices, instead of a separate allocation
for each face,
- (more like a fix) Validation no longer adds a MDeformVert layer when there was
none in the input mesh.
- Small cleanups (more const inputs, etc.)
On my Windows/VS2022/Ryzen5950X machine, import time in seconds (validation on
before this PR -> validation on with this PR, validation off):
- USD (Intel Moore Lane): 9.1 -> 6.7, 4.8.
- OBJ (Blender 3.0 splash): 22.7 -> 18.6, 16.5.
Pull Request: https://projects.blender.org/blender/blender/pulls/121413
The main motivation for this is that it's part of a fix for #113377,
where I want to propagate the edit mesh pointers through copied
meshes in modifiers and geometry nodes, instead of just setting the
edit mesh pointer at the end of the modifier stack. That would have
two main benefits:
1. We avoid the need to write to the evaluated mesh, after evaluation
which means it can be shared directly among evaluated objects.
2. When an object's mesh is completely replaced by the mesh from another
object during evaluation (with the object info node), the final edit
mesh pointer will not be "wrong", allowing us to skip index-mapped
GPU data extraction.
Beyond that, using a shared pointer just makes things more automatic.
Handling of edit mesh data is already complicated enough, this way some
of the worry and complexity can be handled by RAII.
One thing to keep in mind is that the edit mesh's BMesh is still freed
manually with `EDBM_mesh_free_data` when leaving edit mode. I figured
that was a more conservative approach for now. Maybe eventually that
could be handled automatically with RAII too.
Pull Request: https://projects.blender.org/blender/blender/pulls/120276
This is mainly to make the computation threadsafe, to allow computing
the cache on a const mesh, and also to decrease the cost of copying
meshes. Computing caches on const meshes generally makes it easier
to avoid copying meshes unnecessarily in other ways, which would be
useful for some pending fixes and cleanups to modifier evaluation.
With this patch, materials are kept intact in simulation zones and bake nodes
without any additional user action.
This implements the design proposed in #108410 to support referencing
data-blocks (only materials for now) in the baked data. The task also describes
why this is not a trivial issue. A previous attempt was implemented in #109703
but it didn't work well-enough.
The solution is to have an explicit `name (+ library name) -> data-block`
mapping that is stored in the modifier for each bake node and simulation zone.
The `library name` is necessary for it to be unique within a .blend file. Note
that this refers to the name of the `Library` data-block and not a file path.
The baked data only contains the names of the used data-blocks. When the baked
data is loaded, the correct material data-block is looked up from the mapping.
### Automatic Mapping Generation
The most tricky aspect of this approach is to make it feel mostly automatic.
From the user point-of-view, it should just work. Therefore, we don't want the
user to have to create the mapping manually in the majority of cases. Creating
the mapping automatically is difficult because the data-blocks that should
become part of the mapping are only known during depsgraph evaluation. So we
somehow have to gather the missing data blocks during evaluation and then write
the new mappings back to the original data.
While writing back to original data is something we do in some cases already,
the situation here is different, because we are actually creating new relations
between data-blocks. This also means that we'll have to do user-counting. Since
user counts in data-blocks are *not* atomic, we can't do that from multiple
threads at the same time. Also, under some circumstances, it may be necessary to
trigger depsgraph evaluation again after the write-back because it actually
affects the result.
To solve this, a small new API is added in `DEG_depsgraph_writeback_sync.hh`. It
allows gathering tasks which write back to original data in a synchronous way
which may also require a reevaluation.
### Accessing the Mapping
A new `BakeDataBlockMap` is passed to geometry nodes evaluation by the modifier.
This map allows getting the `ID` pointer that should be used for a specific
data-block name that is stored in baked data. It's also used to gather all the
missing data mappings during evaluation.
### Weak ID References
The baked/cached geometries may have references to other data-blocks (currently
only materials, but in the future also e.g. instanced objects/collections).
However, the pointers of these data-blocks are not stable over time. That is
especially true when storing/loading the data from disk, but also just when
playing back the animation. Therefore, the used data-blocks have to referenced
in a different way at run-time.
This is solved by adding `std::unique_ptr<bake::BakeMaterialsList>` to the
run-time data of various geometry data-blocks. If the data-block is cached over
a longer period of time (such that material pointers can't be used directly), it
stores the material name (+ library name) used by each material slot. When the
geometry is used again, the material pointers are restored using these weak name
references and the `BakeDataBlockMap`.
### Manual Mapping Management
There is a new `Data-Blocks` panel in the bake settings in the node editor
sidebar that allows inspecting and modifying the data-blocks that are used when
baking. The user can change what data-block a specific name is mapped to.
Pull Request: https://projects.blender.org/blender/blender/pulls/117043
Except for vertex groups and a few older color types, these
are generally replaced by newer generic attribute types.
Also remove some includes of DNA_mesh_types.h, since it's
included indirectly by BKE_mesh.hh currently.
Use the standard "elements_num" naming, and use the "corner" name rather
than the old "loop" name: `verts_num`, `edges_num`, and `corners_num`.
This matches the existing `faces_num` field which was already renamed.
Pull Request: https://projects.blender.org/blender/blender/pulls/116350
Make the naming consistent with the recent change from "loop" to
"corner". Avoid the need for a special type for these triangles by
conveying the semantics in the naming instead.
- `looptris` -> `corner_tris`
- `lt` -> `tri` (or `corner_tri` when there is less context)
- `looptri_index` -> `tri_index` (or `corner_tri_index`)
- `lt->tri[0]` -> `tri[0]`
- `Span<MLoopTri>` -> `Span<int3>`
- `looptri_faces` -> `tri_faces` (or `corner_tri_faces`)
If we followed the naming pattern of "corner_verts" and "edge_verts"
exactly, we'd probably use "tri_corners" instead. But that sounds much
worse and less intuitive to me.
I've found that by using standard vector types for this sort of data,
the commonalities with other areas become much clearer, and code ends
up being naturally more data oriented. Besides that, the consistency
is nice, and we get to mostly remove use of `DNA_meshdata_types.h`.
Pull Request: https://projects.blender.org/blender/blender/pulls/116238
The term `looptri` was used ambiguously for both single & arrays.
The term `tri` was also used, causing `tri->tri`.
Use terms:
- `looptris` for an array or when dealing with multiple items.
- `looptri` is used when dealing with a single item.
- `lt` for a single MLoopTri variables & arguments.
This was already a convention but not followed closely.
These shouldn't be called in hot loops at all, and mostly aren't
anymore anyway. Definining them outside of a header allows removing
the `BKE_customdata.hh` include from `BKE_mesh.hh`.
"mesh" reads much better than "me" since "me" is a different word.
There's no reason to avoid using two more characters here. Replacing
all of these at once is better than encountering it repeatedly and
doing the same change bit by bit.
Convert shrinkwrap data arrays to use C++ arrays and BitVector,
use references in "EditMeshData" code, and store both structs
with `std::unique_ptr` instead of a raw allocation.
Automatic memory management and clearer ownership! Requires
removing `MEM_CXX_CLASS_ALLOC_FUNCS` from `MeshRuntime`,
but that's used very inconsistently anyway, and `MeshRuntime` isn't
that large.
For improving performance in the triangulate node (#112264), it's
helpful to know whether the mesh has edges or faces that overlap
each other. If that is known to be false, the node can skip edge de-
duplication.This might apply to the future "Replace Faces" node too.
This information is stored as a flag on the mesh and set in various
places that create "clean" new meshes. It isn't calculated lazily unlike
other other areas, because the point is to improve performance, and
the calculation probably isn't faster than the duplication check it's
meant to replace.
Pull Request: https://projects.blender.org/blender/blender/pulls/113205
Design task: #93551
This PR replaces the auto smooth option with a geometry nodes modifier
that sets the sharp edge attribute. This solves a fair number of long-
standing problems related to auto smooth, simplifies the process of
normal computation, and allows Blender to automatically choose between
face, vertex, and face corner normals based on the sharp edge and face
attributes.
Versioning adds a geometry node group to objects with meshes that had
auto-smooth enabled. The modifier can be applied, which also improves
performance.
Auto smooth is now unnecessary to get a combination of sharp and smooth
edges. In general workflows are changed a bit. Separate procedural and
destructive workflows are available. Custom normals can be used
immediately without turning on the removed auto smooth option.
**Procedural**
The node group asset "Smooth by Angle" is the main way to set sharp
normals based on the edge angle. It can be accessed directly in the add
modifier menu. Of course the modifier can be reordered, muted, or
applied like any other, or changed internally like any geometry nodes
modifier.
**Destructive**
Often the sharp edges don't need to be dynamic. This can give better
performance since edge angles don't need to be recalculated. In edit
mode the two operators "Select Sharp Edges" and "Mark Sharp" can be
used. In other modes, the "Shade Smooth by Angle" controls the edge
sharpness directly.
### Breaking API Changes
- `use_auto_smooth` is removed. Face corner normals are now used
automatically if there are mixed smooth vs. not smooth tags. Meshes
now always use custom normals if they exist.
- In Cycles, the lack of the separate auto smooth state makes normals look
triangulated when all faces are shaded smooth.
- `auto_smooth_angle` is removed. Replaced by a modifier (or operator)
controlling the sharp edge attribute. This means the mesh itself
(without an object) doesn't know anything about automatically smoothing
by angle anymore.
- `create_normals_split`, `calc_normals_split`, and `free_normals_split`
are removed, and are replaced by the simpler `Mesh.corner_normals`
collection property. Since it gives access to the normals cache, it
is automatically updated when relevant data changes.
Addons are updated here: https://projects.blender.org/blender/blender-addons/pulls/104609
### Tests
- `geo_node_curves_test_deform_curves_on_surface` has slightly different
results because face corner normals are used instead of interpolated
vertex normals.
- `bf_wavefront_obj_tests` has different export results for one file
which mixed sharp and smooth faces without turning on auto smooth.
- `cycles_mesh_cpu` has one object which is completely flat shaded.
Previously every edge was split before rendering, now it looks triangulated.
Pull Request: https://projects.blender.org/blender/blender/pulls/108014
Add three cached topology maps to `Mesh`, to avoid computations when
mesh data isn't changed. Choosing the right maps to cache is a bit
arbitrary, but generally we have to start somewhere. The limiting
factor is memory usage (all the new caches combined have a
comparable footprint to a UV map).
For now, the caches added are:
- Vertex to face corner
- Vertex to face
- Face corner to face
These caches are used in quite a few places already;
- Face corner normal calculation
- UV value merging
- Setting sharp edges from face angles
- Data transfer modifier
- Voxel remesh attribute remapping
- Sculpt mode painting
- Sculpt mode normal calculation
- Vertex paint mode
- Split edges geometry node
- Mesh topology geometry nodes
Caching topology maps means they don't have to be rebuilt every time
they're used. Meshes copied but without topology changes can share
the cache, further reducing re-computations. For example, FPS with a
large mesh using the "Corners of Vertex" node went from 1.8 to 2.3.
Entering sculpt mode is slightly faster too.
There is some obvious work for future commits:
- Use caches in attribute domain interpolation
- More multithreading of second phase of map building
- Update/build caches eagerly in some geometry nodes
Pull Request: https://projects.blender.org/blender/blender/pulls/107816
Since the normals are stored in a shared cache, tagging them dirty
recreated the cache from scratch when it was shared. Instead,
add a function that updates the cache in the same call as tagging
it dirty. This keeps the old state of the cache around even if it was
shared, and reflects the way that it's really the PBVH and sculpt
mode managing the dirty status of normals while sculpt mode
is active.
One consequence is that the BVH cache and the triangulation
cache need to be tagged dirty manually. I'd like to avoid abstracting
this more than necessary, because I'm hoping in the long term
different caching abstractions like a more global cache that takes
implicit sharing versions into account will make this complexity
unnecessary.
Fixes#111628, #111563
Pull Request: https://projects.blender.org/blender/blender/pulls/111641
Since vertex and face normals can be calculated separately, it simplifies
things to further separate the two caches. This makes it easier to use
`SharedCache` to avoid recalculating normals when copying meshes.
Sharing vertex normal caches with meshes with the same positions and
topology allows completely skipping recomputation as meshes are
copied. The effects are similar to e8f4010611, but normals are much
more expensive, so the benefit is larger.
In a simple test changing a large grid's generic attribute with geometry
nodes, I observed a performance improvement from 12 to 17 FPS.
Most real world situations will have smaller changes though.
Completely splitting face and vertex calculation is slightly slower
when face normals aren't already calculated, so I kept the option
to recalculate them together as well.
This simplifies investigating the changes in #105920 which resolve
non-determinism in the vertex normal calculation. If we can make the
topology map creation fast enough, that might allow simplifying this
code more in the future.
Pull Request: https://projects.blender.org/blender/blender/pulls/110479
Listing the "Blender Foundation" as copyright holder implied the Blender
Foundation holds copyright to files which may include work from many
developers.
While keeping copyright on headers makes sense for isolated libraries,
Blender's own code may be refactored or moved between files in a way
that makes the per file copyright holders less meaningful.
Copyright references to the "Blender Foundation" have been replaced with
"Blender Authors", with the exception of `./extern/` since these this
contains libraries which are more isolated, any changed to license
headers there can be handled on a case-by-case basis.
Some directories in `./intern/` have also been excluded:
- `./intern/cycles/` it's own `AUTHORS` file is planned.
- `./intern/opensubdiv/`.
An "AUTHORS" file has been added, using the chromium projects authors
file as a template.
Design task: #110784
Ref !110783.
Implements the rest of #101689, after 5e9ea9243b.
- `vdata` -> `vert_data`
- `edata` -> `edge_data`
- `pdata` -> `face_data`
- `ldata` -> `loop_data`
A deeper rename of `loop` to `corner` will be proposed as a next
step, and renaming `totvert` and `totedge` can be done separately.
Pull Request: https://projects.blender.org/blender/blender/pulls/110432
Implements part of #101689.
The "poly" name was chosen to distinguish the `MLoop` + `MPoly`
combination from the `MFace` struct it replaced. Those two structures
persisted together for a long time, but nowadays `MPoly` is gone, and
`MFace` is only used in some legacy code like the particle system.
To avoid unnecessarily using a different term, increase consistency
with the UI and with BMesh, and generally make code a bit easier to
read, this commit replaces the `poly` term with `poly`. Most variables
that use the term are renamed too. `Mesh.totface` and `Mesh.fdata` now
have a `_legacy` suffix to reduce confusion. In a next step, `pdata`
can be renamed to `face_data` as well.
Pull Request: https://projects.blender.org/blender/blender/pulls/109819
