diff options
author | Hans Goudey <h.goudey@me.com> | 2022-01-13 23:37:58 +0300 |
---|---|---|
committer | Hans Goudey <h.goudey@me.com> | 2022-01-13 23:38:25 +0300 |
commit | cfa53e0fbeed7178c7876413e2010fd3347d7f72 (patch) | |
tree | f5cdc39134a3477614ae98699f90da3722161ed0 /source/blender/blenkernel/intern/subdiv_mesh.c | |
parent | 800fc17367480b82e07570d5234726aa61c38563 (diff) |
Refactor: Move normals out of MVert, lazy calculation
As described in T91186, this commit moves mesh vertex normals into a
contiguous array of float vectors in a custom data layer, how face
normals are currently stored.
The main interface is documented in `BKE_mesh.h`. Vertex and face
normals are now calculated on-demand and cached, retrieved with an
"ensure" function. Since the logical state of a mesh is now "has
normals when necessary", they can be retrieved from a `const` mesh.
The goal is to use on-demand calculation for all derived data, but
leave room for eager calculation for performance purposes (modifier
evaluation is threaded, but viewport data generation is not).
**Benefits**
This moves us closer to a SoA approach rather than the current AoS
paradigm. Accessing a contiguous `float3` is much more efficient than
retrieving data from a larger struct. The memory requirements for
accessing only normals or vertex locations are smaller, and at the
cost of more memory usage for just normals, they now don't have to
be converted between float and short, which also simplifies code
In the future, the remaining items can be removed from `MVert`,
leaving only `float3`, which has similar benefits (see T93602).
Removing the combination of derived and original data makes it
conceptually simpler to only calculate normals when necessary.
This is especially important now that we have more opportunities
for temporary meshes in geometry nodes.
**Performance**
In addition to the theoretical future performance improvements by
making `MVert == float3`, I've done some basic performance testing
on this patch directly. The data is fairly rough, but it gives an idea
about where things stand generally.
- Mesh line primitive 4m Verts: 1.16x faster (36 -> 31 ms),
showing that accessing just `MVert` is now more efficient.
- Spring Splash Screen: 1.03-1.06 -> 1.06-1.11 FPS, a very slight
change that at least shows there is no regression.
- Sprite Fright Snail Smoosh: 3.30-3.40 -> 3.42-3.50 FPS, a small
but observable speedup.
- Set Position Node with Scaled Normal: 1.36x faster (53 -> 39 ms),
shows that using normals in geometry nodes is faster.
- Normal Calculation 1.6m Vert Cube: 1.19x faster (25 -> 21 ms),
shows that calculating normals is slightly faster now.
- File Size of 1.6m Vert Cube: 1.03x smaller (214.7 -> 208.4 MB),
Normals are not saved in files, which can help with large meshes.
As for memory usage, it may be slightly more in some cases, but
I didn't observe any difference in the production files I tested.
**Tests**
Some modifiers and cycles test results need to be updated with this
commit, for two reasons:
- The subdivision surface modifier is not responsible for calculating
normals anymore. In master, the modifier creates different normals
than the result of the `Mesh` normal calculation, so this is a bug
fix.
- There are small differences in the results of some modifiers that
use normals because they are not converted to and from `short`
anymore.
**Future improvements**
- Remove `ModifierTypeInfo::dependsOnNormals`. Code in each modifier
already retrieves normals if they are needed anyway.
- Copy normals as part of a better CoW system for attributes.
- Make more areas use lazy instead of eager normal calculation.
- Remove `BKE_mesh_normals_tag_dirty` in more places since that is
now the default state of a new mesh.
- Possibly apply a similar change to derived face corner normals.
Differential Revision: https://developer.blender.org/D12770
Diffstat (limited to 'source/blender/blenkernel/intern/subdiv_mesh.c')
-rw-r--r-- | source/blender/blenkernel/intern/subdiv_mesh.c | 102 |
1 files changed, 16 insertions, 86 deletions
diff --git a/source/blender/blenkernel/intern/subdiv_mesh.c b/source/blender/blenkernel/intern/subdiv_mesh.c index 5bf519e5f70..c334d9a2c33 100644 --- a/source/blender/blenkernel/intern/subdiv_mesh.c +++ b/source/blender/blenkernel/intern/subdiv_mesh.c @@ -21,6 +21,7 @@ * \ingroup bke */ +#include "BKE_mesh.h" #include "BKE_subdiv_mesh.h" #include "atomic_ops.h" @@ -58,23 +59,8 @@ typedef struct SubdivMeshContext { /* UV layers interpolation. */ int num_uv_layers; MLoopUV *uv_layers[MAX_MTFACE]; - /* Accumulated values. - * - * Averaging is happening for vertices along the coarse edges and corners. - * This is needed for both displacement and normals. - * - * Displacement is being accumulated to a vertices coordinates, since those - * are not needed during traversal of edge/corner vertices. - * - * For normals we are using dedicated array, since we can not use same - * vertices (normals are `short`, which will cause a lot of precision - * issues). */ - float (*accumulated_normals)[3]; /* Per-subdivided vertex counter of averaged values. */ int *accumulated_counters; - /* Denotes whether normals can be evaluated from a limit surface. One case - * when it's not possible is when displacement is used. */ - bool can_evaluate_normals; bool have_displacement; } SubdivMeshContext; @@ -102,20 +88,12 @@ static void subdiv_mesh_ctx_cache_custom_data_layers(SubdivMeshContext *ctx) static void subdiv_mesh_prepare_accumulator(SubdivMeshContext *ctx, int num_vertices) { - if (!ctx->can_evaluate_normals && !ctx->have_displacement) { - return; - } - /* TODO(sergey): Technically, this is overallocating, we don't need memory - * for an inner subdivision vertices. */ - ctx->accumulated_normals = MEM_calloc_arrayN( - num_vertices, sizeof(*ctx->accumulated_normals), "subdiv accumulated normals"); ctx->accumulated_counters = MEM_calloc_arrayN( num_vertices, sizeof(*ctx->accumulated_counters), "subdiv accumulated counters"); } static void subdiv_mesh_context_free(SubdivMeshContext *ctx) { - MEM_SAFE_FREE(ctx->accumulated_normals); MEM_SAFE_FREE(ctx->accumulated_counters); } @@ -450,48 +428,23 @@ static void subdiv_mesh_tls_free(void *tls_v) /** \} */ -/* -------------------------------------------------------------------- */ -/** \name Evaluation helper functions - * \{ */ - -static void eval_final_point_and_vertex_normal(Subdiv *subdiv, - const int ptex_face_index, - const float u, - const float v, - float r_P[3], - short r_N[3]) -{ - if (subdiv->displacement_evaluator == NULL) { - BKE_subdiv_eval_limit_point_and_short_normal(subdiv, ptex_face_index, u, v, r_P, r_N); - } - else { - BKE_subdiv_eval_final_point(subdiv, ptex_face_index, u, v, r_P); - } -} - /** \} */ /* -------------------------------------------------------------------- */ /** \name Accumulation helpers * \{ */ -static void subdiv_accumulate_vertex_normal_and_displacement(SubdivMeshContext *ctx, - const int ptex_face_index, - const float u, - const float v, - MVert *subdiv_vert) +static void subdiv_accumulate_vertex_displacement(SubdivMeshContext *ctx, + const int ptex_face_index, + const float u, + const float v, + MVert *subdiv_vert) { Subdiv *subdiv = ctx->subdiv; const int subdiv_vertex_index = subdiv_vert - ctx->subdiv_mesh->mvert; float dummy_P[3], dPdu[3], dPdv[3], D[3]; BKE_subdiv_eval_limit_point_and_derivatives(subdiv, ptex_face_index, u, v, dummy_P, dPdu, dPdv); - /* Accumulate normal. */ - if (ctx->can_evaluate_normals) { - float N[3]; - cross_v3_v3v3(N, dPdu, dPdv); - normalize_v3(N); - add_v3_v3(ctx->accumulated_normals[subdiv_vertex_index], N); - } + /* Accumulate displacement if needed. */ if (ctx->have_displacement) { /* NOTE: The subdivided mesh is allocated in this module, and its vertices are kept at zero @@ -589,13 +542,6 @@ static void evaluate_vertex_and_apply_displacement_copy(const SubdivMeshContext BKE_subdiv_eval_limit_point(ctx->subdiv, ptex_face_index, u, v, subdiv_vert->co); /* Apply displacement. */ add_v3_v3(subdiv_vert->co, D); - /* Copy normal from accumulated storage. */ - if (ctx->can_evaluate_normals) { - float N[3]; - copy_v3_v3(N, ctx->accumulated_normals[subdiv_vertex_index]); - normalize_v3(N); - normal_float_to_short_v3(subdiv_vert->no, N); - } /* Remove facedot flag. This can happen if there is more than one subsurf modifier. */ subdiv_vert->flag &= ~ME_VERT_FACEDOT; } @@ -622,15 +568,6 @@ static void evaluate_vertex_and_apply_displacement_interpolate( BKE_subdiv_eval_limit_point(ctx->subdiv, ptex_face_index, u, v, subdiv_vert->co); /* Apply displacement. */ add_v3_v3(subdiv_vert->co, D); - /* Copy normal from accumulated storage. */ - if (ctx->can_evaluate_normals) { - const float inv_num_accumulated = 1.0f / ctx->accumulated_counters[subdiv_vertex_index]; - float N[3]; - copy_v3_v3(N, ctx->accumulated_normals[subdiv_vertex_index]); - mul_v3_fl(N, inv_num_accumulated); - normalize_v3(N); - normal_float_to_short_v3(subdiv_vert->no, N); - } } static void subdiv_mesh_vertex_every_corner_or_edge(const SubdivForeachContext *foreach_context, @@ -644,7 +581,7 @@ static void subdiv_mesh_vertex_every_corner_or_edge(const SubdivForeachContext * Mesh *subdiv_mesh = ctx->subdiv_mesh; MVert *subdiv_mvert = subdiv_mesh->mvert; MVert *subdiv_vert = &subdiv_mvert[subdiv_vertex_index]; - subdiv_accumulate_vertex_normal_and_displacement(ctx, ptex_face_index, u, v, subdiv_vert); + subdiv_accumulate_vertex_displacement(ctx, ptex_face_index, u, v, subdiv_vert); } static void subdiv_mesh_vertex_every_corner(const SubdivForeachContext *foreach_context, @@ -793,8 +730,7 @@ static void subdiv_mesh_vertex_inner(const SubdivForeachContext *foreach_context MVert *subdiv_vert = &subdiv_mvert[subdiv_vertex_index]; subdiv_mesh_ensure_vertex_interpolation(ctx, tls, coarse_poly, coarse_corner); subdiv_vertex_data_interpolate(ctx, subdiv_vert, &tls->vertex_interpolation, u, v); - eval_final_point_and_vertex_normal( - subdiv, ptex_face_index, u, v, subdiv_vert->co, subdiv_vert->no); + BKE_subdiv_eval_final_point(subdiv, ptex_face_index, u, v, subdiv_vert->co); subdiv_mesh_tag_center_vertex(coarse_poly, subdiv_vert, u, v); } @@ -1141,12 +1077,6 @@ static void subdiv_mesh_vertex_of_loose_edge(const struct SubdivForeachContext * /* TODO(sergey): This matches old behavior, but we can as well interpolate * it. Maybe even using vertex varying attributes. */ subdiv_vertex->bweight = 0.0f; - /* Reset normal, initialize it in a similar way as edit mode does for a - * vertices adjacent to a loose edges. - * See `mesh_evaluate#mesh_calc_normals_vert_fallback` */ - float no[3]; - normalize_v3_v3(no, subdiv_vertex->co); - normal_float_to_short_v3(subdiv_vertex->no, no); } /** \} */ @@ -1161,8 +1091,8 @@ static void setup_foreach_callbacks(const SubdivMeshContext *subdiv_context, memset(foreach_context, 0, sizeof(*foreach_context)); /* General information. */ foreach_context->topology_info = subdiv_mesh_topology_info; - /* Every boundary geometry. Used for displacement and normals averaging. */ - if (subdiv_context->can_evaluate_normals || subdiv_context->have_displacement) { + /* Every boundary geometry. Used for displacement averaging. */ + if (subdiv_context->have_displacement) { foreach_context->vertex_every_corner = subdiv_mesh_vertex_every_corner; foreach_context->vertex_every_edge = subdiv_mesh_vertex_every_edge; } @@ -1212,8 +1142,6 @@ Mesh *BKE_subdiv_to_mesh(Subdiv *subdiv, subdiv_context.coarse_mesh = coarse_mesh; subdiv_context.subdiv = subdiv; subdiv_context.have_displacement = (subdiv->displacement_evaluator != NULL); - subdiv_context.can_evaluate_normals = !subdiv_context.have_displacement && - subdiv_context.subdiv->settings.is_adaptive; /* Multi-threaded traversal/evaluation. */ BKE_subdiv_stats_begin(&subdiv->stats, SUBDIV_STATS_SUBDIV_TO_MESH_GEOMETRY); SubdivForeachContext foreach_context; @@ -1227,9 +1155,11 @@ Mesh *BKE_subdiv_to_mesh(Subdiv *subdiv, Mesh *result = subdiv_context.subdiv_mesh; // BKE_mesh_validate(result, true, true); BKE_subdiv_stats_end(&subdiv->stats, SUBDIV_STATS_SUBDIV_TO_MESH); - if (!subdiv_context.can_evaluate_normals) { - BKE_mesh_normals_tag_dirty(result); - } + /* Using normals from the limit surface gives different results than Blender's vertex normal + * calculation. Since vertex normals are supposed to be a consistent cache, don't bother + * calculating them here. The work may have been pointless anyway if the mesh is deformed or + * changed afterwards. */ + BKE_mesh_normals_tag_dirty(result); /* Free used memory. */ subdiv_mesh_context_free(&subdiv_context); return result; |