This repository has been archived on 2023-10-09. You can view files and clone it. You cannot open issues or pull requests or push a commit.
Files
blender-archive/source/blender/blenkernel/intern/subdiv_eval.cc
Hans Goudey 16fbadde36 Mesh: Replace MLoop struct with generic attributes
Implements #102359.

Split the `MLoop` struct into two separate integer arrays called
`corner_verts` and `corner_edges`, referring to the vertex each corner
is attached to and the next edge around the face at each corner. These
arrays can be sliced to give access to the edges or vertices in a face.
Then they are often referred to as "poly_verts" or "poly_edges".

The main benefits are halving the necessary memory bandwidth when only
one array is used and simplifications from using regular integer indices
instead of a special-purpose struct.

The commit also starts a renaming from "loop" to "corner" in mesh code.

Like the other mesh struct of array refactors, forward compatibility is
kept by writing files with the older format. This will be done until 4.0
to ease the transition process.

Looking at a small portion of the patch should give a good impression
for the rest of the changes. I tried to make the changes as small as
possible so it's easy to tell the correctness from the diff. Though I
found Blender developers have been very inventive over the last decade
when finding different ways to loop over the corners in a face.

For performance, nearly every piece of code that deals with `Mesh` is
slightly impacted. Any algorithm that is memory bottle-necked should
see an improvement. For example, here is a comparison of interpolating
a vertex float attribute to face corners (Ryzen 3700x):

**Before** (Average: 3.7 ms, Min: 3.4 ms)
```
threading::parallel_for(loops.index_range(), 4096, [&](IndexRange range) {
  for (const int64_t i : range) {
    dst[i] = src[loops[i].v];
  }
});
```

**After** (Average: 2.9 ms, Min: 2.6 ms)
```
array_utils::gather(src, corner_verts, dst);
```

That's an improvement of 28% to the average timings, and it's also a
simplification, since an index-based routine can be used instead.
For more examples using the new arrays, see the design task.

Pull Request: blender/blender#104424
2023-03-20 15:55:13 +01:00

385 lines
14 KiB
C++

/* SPDX-License-Identifier: GPL-2.0-or-later
* Copyright 2018 Blender Foundation. All rights reserved. */
/** \file
* \ingroup bke
*/
#include "BKE_subdiv_eval.h"
#include "DNA_mesh_types.h"
#include "DNA_meshdata_types.h"
#include "BLI_bitmap.h"
#include "BLI_math_vector.h"
#include "BLI_task.h"
#include "BLI_utildefines.h"
#include "BKE_customdata.h"
#include "BKE_mesh.hh"
#include "BKE_subdiv.h"
#include "MEM_guardedalloc.h"
#include "opensubdiv_evaluator_capi.h"
#include "opensubdiv_topology_refiner_capi.h"
/* --------------------------------------------------------------------
* Helper functions.
*/
static eOpenSubdivEvaluator opensubdiv_evalutor_from_subdiv_evaluator_type(
eSubdivEvaluatorType evaluator_type)
{
switch (evaluator_type) {
case SUBDIV_EVALUATOR_TYPE_CPU: {
return OPENSUBDIV_EVALUATOR_CPU;
}
case SUBDIV_EVALUATOR_TYPE_GPU: {
return OPENSUBDIV_EVALUATOR_GPU;
}
}
BLI_assert_msg(0, "Unknown evaluator type");
return OPENSUBDIV_EVALUATOR_CPU;
}
/* --------------------------------------------------------------------
* Main subdivision evaluation.
*/
bool BKE_subdiv_eval_begin(Subdiv *subdiv,
eSubdivEvaluatorType evaluator_type,
OpenSubdiv_EvaluatorCache *evaluator_cache,
const OpenSubdiv_EvaluatorSettings *settings)
{
BKE_subdiv_stats_reset(&subdiv->stats, SUBDIV_STATS_EVALUATOR_CREATE);
if (subdiv->topology_refiner == nullptr) {
/* Happens on input mesh with just loose geometry,
* or when OpenSubdiv is disabled */
return false;
}
if (subdiv->evaluator == nullptr) {
eOpenSubdivEvaluator opensubdiv_evaluator_type =
opensubdiv_evalutor_from_subdiv_evaluator_type(evaluator_type);
BKE_subdiv_stats_begin(&subdiv->stats, SUBDIV_STATS_EVALUATOR_CREATE);
subdiv->evaluator = openSubdiv_createEvaluatorFromTopologyRefiner(
subdiv->topology_refiner, opensubdiv_evaluator_type, evaluator_cache);
BKE_subdiv_stats_end(&subdiv->stats, SUBDIV_STATS_EVALUATOR_CREATE);
if (subdiv->evaluator == nullptr) {
return false;
}
}
else {
/* TODO(sergey): Check for topology change. */
}
subdiv->evaluator->setSettings(subdiv->evaluator, settings);
BKE_subdiv_eval_init_displacement(subdiv);
return true;
}
static void set_coarse_positions(Subdiv *subdiv,
const Mesh *mesh,
const float (*coarse_vertex_cos)[3])
{
const float(*positions)[3] = BKE_mesh_vert_positions(mesh);
const blender::Span<MPoly> polys = mesh->polys();
const blender::Span<int> corner_verts = mesh->corner_verts();
/* Mark vertices which needs new coordinates. */
/* TODO(sergey): This is annoying to calculate this on every update,
* maybe it's better to cache this mapping. Or make it possible to have
* OpenSubdiv's vertices match mesh ones? */
BLI_bitmap *vertex_used_map = BLI_BITMAP_NEW(mesh->totvert, "vert used map");
for (int poly_index = 0; poly_index < mesh->totpoly; poly_index++) {
const MPoly &poly = polys[poly_index];
for (int i = 0; i < poly.totloop; i++) {
BLI_BITMAP_ENABLE(vertex_used_map, corner_verts[poly.loopstart + i]);
}
}
/* Use a temporary buffer so we do not upload vertices one at a time to the GPU. */
float(*buffer)[3] = static_cast<float(*)[3]>(
MEM_mallocN(sizeof(float[3]) * mesh->totvert, __func__));
int manifold_vertex_count = 0;
for (int vertex_index = 0, manifold_vertex_index = 0; vertex_index < mesh->totvert;
vertex_index++) {
if (!BLI_BITMAP_TEST_BOOL(vertex_used_map, vertex_index)) {
continue;
}
const float *vertex_co;
if (coarse_vertex_cos != nullptr) {
vertex_co = coarse_vertex_cos[vertex_index];
}
else {
vertex_co = positions[vertex_index];
}
copy_v3_v3(&buffer[manifold_vertex_index][0], vertex_co);
manifold_vertex_index++;
manifold_vertex_count++;
}
subdiv->evaluator->setCoarsePositions(
subdiv->evaluator, &buffer[0][0], 0, manifold_vertex_count);
MEM_freeN(vertex_used_map);
MEM_freeN(buffer);
}
/* Context which is used to fill face varying data in parallel. */
struct FaceVaryingDataFromUVContext {
OpenSubdiv_TopologyRefiner *topology_refiner;
const Mesh *mesh;
blender::Span<MPoly> polys;
const float (*mloopuv)[2];
float (*buffer)[2];
int layer_index;
};
static void set_face_varying_data_from_uv_task(void *__restrict userdata,
const int face_index,
const TaskParallelTLS *__restrict /*tls*/)
{
FaceVaryingDataFromUVContext *ctx = static_cast<FaceVaryingDataFromUVContext *>(userdata);
OpenSubdiv_TopologyRefiner *topology_refiner = ctx->topology_refiner;
const int layer_index = ctx->layer_index;
const MPoly &poly = ctx->polys[face_index];
const float(*mluv)[2] = &ctx->mloopuv[poly.loopstart];
/* TODO(sergey): OpenSubdiv's C-API converter can change winding of
* loops of a face, need to watch for that, to prevent wrong UVs assigned.
*/
const int num_face_vertices = topology_refiner->getNumFaceVertices(topology_refiner, face_index);
const int *uv_indices = topology_refiner->getFaceFVarValueIndices(
topology_refiner, face_index, layer_index);
for (int vertex_index = 0; vertex_index < num_face_vertices; vertex_index++, mluv++) {
copy_v2_v2(ctx->buffer[uv_indices[vertex_index]], *mluv);
}
}
static void set_face_varying_data_from_uv(Subdiv *subdiv,
const Mesh *mesh,
const float (*mloopuv)[2],
const int layer_index)
{
OpenSubdiv_TopologyRefiner *topology_refiner = subdiv->topology_refiner;
OpenSubdiv_Evaluator *evaluator = subdiv->evaluator;
const int num_faces = topology_refiner->getNumFaces(topology_refiner);
const float(*mluv)[2] = mloopuv;
const int num_fvar_values = topology_refiner->getNumFVarValues(topology_refiner, layer_index);
/* Use a temporary buffer so we do not upload UVs one at a time to the GPU. */
float(*buffer)[2] = static_cast<float(*)[2]>(
MEM_mallocN(sizeof(float[2]) * num_fvar_values, __func__));
FaceVaryingDataFromUVContext ctx;
ctx.topology_refiner = topology_refiner;
ctx.layer_index = layer_index;
ctx.mloopuv = mluv;
ctx.mesh = mesh;
ctx.polys = mesh->polys();
ctx.buffer = buffer;
TaskParallelSettings parallel_range_settings;
BLI_parallel_range_settings_defaults(&parallel_range_settings);
parallel_range_settings.min_iter_per_thread = 1;
BLI_task_parallel_range(
0, num_faces, &ctx, set_face_varying_data_from_uv_task, &parallel_range_settings);
evaluator->setFaceVaryingData(evaluator, layer_index, &buffer[0][0], 0, num_fvar_values);
MEM_freeN(buffer);
}
static void set_vertex_data_from_orco(Subdiv *subdiv, const Mesh *mesh)
{
const float(*orco)[3] = static_cast<const float(*)[3]>(
CustomData_get_layer(&mesh->vdata, CD_ORCO));
const float(*cloth_orco)[3] = static_cast<const float(*)[3]>(
CustomData_get_layer(&mesh->vdata, CD_CLOTH_ORCO));
if (orco || cloth_orco) {
OpenSubdiv_TopologyRefiner *topology_refiner = subdiv->topology_refiner;
OpenSubdiv_Evaluator *evaluator = subdiv->evaluator;
const int num_verts = topology_refiner->getNumVertices(topology_refiner);
if (orco && cloth_orco) {
/* Set one by one if have both. */
for (int i = 0; i < num_verts; i++) {
float data[6];
copy_v3_v3(data, orco[i]);
copy_v3_v3(data + 3, cloth_orco[i]);
evaluator->setVertexData(evaluator, data, i, 1);
}
}
else {
/* Faster single call if we have either. */
if (orco) {
evaluator->setVertexData(evaluator, orco[0], 0, num_verts);
}
else if (cloth_orco) {
evaluator->setVertexData(evaluator, cloth_orco[0], 0, num_verts);
}
}
}
}
static void get_mesh_evaluator_settings(OpenSubdiv_EvaluatorSettings *settings, const Mesh *mesh)
{
settings->num_vertex_data = (CustomData_has_layer(&mesh->vdata, CD_ORCO) ? 3 : 0) +
(CustomData_has_layer(&mesh->vdata, CD_CLOTH_ORCO) ? 3 : 0);
}
bool BKE_subdiv_eval_begin_from_mesh(Subdiv *subdiv,
const Mesh *mesh,
const float (*coarse_vertex_cos)[3],
eSubdivEvaluatorType evaluator_type,
OpenSubdiv_EvaluatorCache *evaluator_cache)
{
OpenSubdiv_EvaluatorSettings settings = {0};
get_mesh_evaluator_settings(&settings, mesh);
if (!BKE_subdiv_eval_begin(subdiv, evaluator_type, evaluator_cache, &settings)) {
return false;
}
return BKE_subdiv_eval_refine_from_mesh(subdiv, mesh, coarse_vertex_cos);
}
bool BKE_subdiv_eval_refine_from_mesh(Subdiv *subdiv,
const Mesh *mesh,
const float (*coarse_vertex_cos)[3])
{
if (subdiv->evaluator == nullptr) {
/* NOTE: This situation is supposed to be handled by begin(). */
BLI_assert_msg(0, "Is not supposed to happen");
return false;
}
/* Set coordinates of base mesh vertices. */
set_coarse_positions(subdiv, mesh, coarse_vertex_cos);
/* Set face-varying data to UV maps. */
const int num_uv_layers = CustomData_number_of_layers(&mesh->ldata, CD_PROP_FLOAT2);
for (int layer_index = 0; layer_index < num_uv_layers; layer_index++) {
const float(*mloopuv)[2] = static_cast<const float(*)[2]>(
CustomData_get_layer_n(&mesh->ldata, CD_PROP_FLOAT2, layer_index));
set_face_varying_data_from_uv(subdiv, mesh, mloopuv, layer_index);
}
/* Set vertex data to orco. */
set_vertex_data_from_orco(subdiv, mesh);
/* Update evaluator to the new coarse geometry. */
BKE_subdiv_stats_begin(&subdiv->stats, SUBDIV_STATS_EVALUATOR_REFINE);
subdiv->evaluator->refine(subdiv->evaluator);
BKE_subdiv_stats_end(&subdiv->stats, SUBDIV_STATS_EVALUATOR_REFINE);
return true;
}
void BKE_subdiv_eval_init_displacement(Subdiv *subdiv)
{
if (subdiv->displacement_evaluator == nullptr) {
return;
}
if (subdiv->displacement_evaluator->initialize == nullptr) {
return;
}
subdiv->displacement_evaluator->initialize(subdiv->displacement_evaluator);
}
/* --------------------------------------------------------------------
* Single point queries.
*/
void BKE_subdiv_eval_limit_point(
Subdiv *subdiv, const int ptex_face_index, const float u, const float v, float r_P[3])
{
BKE_subdiv_eval_limit_point_and_derivatives(
subdiv, ptex_face_index, u, v, r_P, nullptr, nullptr);
}
void BKE_subdiv_eval_limit_point_and_derivatives(Subdiv *subdiv,
const int ptex_face_index,
const float u,
const float v,
float r_P[3],
float r_dPdu[3],
float r_dPdv[3])
{
subdiv->evaluator->evaluateLimit(subdiv->evaluator, ptex_face_index, u, v, r_P, r_dPdu, r_dPdv);
/* NOTE: In a very rare occasions derivatives are evaluated to zeros or are exactly equal.
* This happens, for example, in single vertex on Suzannne's nose (where two quads have 2 common
* edges).
*
* This makes tangent space displacement (such as multi-resolution) impossible to be used in
* those vertices, so those needs to be addressed in one way or another.
*
* Simplest thing to do: step inside of the face a little bit, where there is known patch at
* which there must be proper derivatives. This might break continuity of normals, but is better
* that giving totally unusable derivatives. */
if (r_dPdu != nullptr && r_dPdv != nullptr) {
if ((is_zero_v3(r_dPdu) || is_zero_v3(r_dPdv)) || equals_v3v3(r_dPdu, r_dPdv)) {
subdiv->evaluator->evaluateLimit(subdiv->evaluator,
ptex_face_index,
u * 0.999f + 0.0005f,
v * 0.999f + 0.0005f,
r_P,
r_dPdu,
r_dPdv);
}
}
}
void BKE_subdiv_eval_limit_point_and_normal(Subdiv *subdiv,
const int ptex_face_index,
const float u,
const float v,
float r_P[3],
float r_N[3])
{
float dPdu[3], dPdv[3];
BKE_subdiv_eval_limit_point_and_derivatives(subdiv, ptex_face_index, u, v, r_P, dPdu, dPdv);
cross_v3_v3v3(r_N, dPdu, dPdv);
normalize_v3(r_N);
}
void BKE_subdiv_eval_vertex_data(
Subdiv *subdiv, const int ptex_face_index, const float u, const float v, float r_vertex_data[])
{
subdiv->evaluator->evaluateVertexData(subdiv->evaluator, ptex_face_index, u, v, r_vertex_data);
}
void BKE_subdiv_eval_face_varying(Subdiv *subdiv,
const int face_varying_channel,
const int ptex_face_index,
const float u,
const float v,
float r_face_varying[2])
{
subdiv->evaluator->evaluateFaceVarying(
subdiv->evaluator, face_varying_channel, ptex_face_index, u, v, r_face_varying);
}
void BKE_subdiv_eval_displacement(Subdiv *subdiv,
const int ptex_face_index,
const float u,
const float v,
const float dPdu[3],
const float dPdv[3],
float r_D[3])
{
if (subdiv->displacement_evaluator == nullptr) {
zero_v3(r_D);
return;
}
subdiv->displacement_evaluator->eval_displacement(
subdiv->displacement_evaluator, ptex_face_index, u, v, dPdu, dPdv, r_D);
}
void BKE_subdiv_eval_final_point(
Subdiv *subdiv, const int ptex_face_index, const float u, const float v, float r_P[3])
{
if (subdiv->displacement_evaluator) {
float dPdu[3], dPdv[3], D[3];
BKE_subdiv_eval_limit_point_and_derivatives(subdiv, ptex_face_index, u, v, r_P, dPdu, dPdv);
BKE_subdiv_eval_displacement(subdiv, ptex_face_index, u, v, dPdu, dPdv, D);
add_v3_v3(r_P, D);
}
else {
BKE_subdiv_eval_limit_point(subdiv, ptex_face_index, u, v, r_P);
}
}