OpenSubDiv: add support for an OpenGL evaluator

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

1 files changed, 72 insertions, 17 deletions

diff --git a/source/blender/draw/intern/mesh_extractors/extract_mesh_vbo_uv.cc b/source/blender/draw/intern/mesh_extractors/extract_mesh_vbo_uv.cc
index af279b08a59..6e9d8ef6926 100644
--- a/source/blender/draw/intern/mesh_extractors/extract_mesh_vbo_uv.cc
+++ b/source/blender/draw/intern/mesh_extractors/extract_mesh_vbo_uv.cc
@@ -23,6 +23,7 @@
 
 #include "BLI_string.h"
 
+#include "draw_subdivision.h"
 #include "extract_mesh.h"
 
 namespace blender::draw {
@@ -31,25 +32,27 @@ namespace blender::draw {
 /** \name Extract UV  layers
  * \{ */
 
-static void extract_uv_init(const MeshRenderData *mr,
-                            struct MeshBatchCache *cache,
-                            void *buf,
-                            void *UNUSED(tls_data))
+/* Initialize the vertex format to be used for UVs. Return true if any UV layer is
+ * found, false otherwise. */
+static bool mesh_extract_uv_format_init(GPUVertFormat *format,
+                                        struct MeshBatchCache *cache,
+                                        CustomData *cd_ldata,
+                                        eMRExtractType extract_type,
+                                        uint32_t &r_uv_layers)
 {
-  GPUVertBuf *vbo = static_cast<GPUVertBuf *>(buf);
-  GPUVertFormat format = {0};
-  GPU_vertformat_deinterleave(&format);
+  GPU_vertformat_deinterleave(format);
 
-  CustomData *cd_ldata = (mr->extract_type == MR_EXTRACT_BMESH) ? &mr->bm->ldata : &mr->me->ldata;
   uint32_t uv_layers = cache->cd_used.uv;
   /* HACK to fix T68857 */
-  if (mr->extract_type == MR_EXTRACT_BMESH && cache->cd_used.edit_uv == 1) {
+  if (extract_type == MR_EXTRACT_BMESH && cache->cd_used.edit_uv == 1) {
     int layer = CustomData_get_active_layer(cd_ldata, CD_MLOOPUV);
     if (layer != -1) {
       uv_layers |= (1 << layer);
     }
   }
 
+  r_uv_layers = uv_layers;
+
   for (int i = 0; i < MAX_MTFACE; i++) {
     if (uv_layers & (1 << i)) {
       char attr_name[32], attr_safe_name[GPU_MAX_SAFE_ATTR_NAME];
@@ -58,30 +61,47 @@ static void extract_uv_init(const MeshRenderData *mr,
       GPU_vertformat_safe_attr_name(layer_name, attr_safe_name, GPU_MAX_SAFE_ATTR_NAME);
       /* UV layer name. */
       BLI_snprintf(attr_name, sizeof(attr_name), "u%s", attr_safe_name);
-      GPU_vertformat_attr_add(&format, attr_name, GPU_COMP_F32, 2, GPU_FETCH_FLOAT);
+      GPU_vertformat_attr_add(format, attr_name, GPU_COMP_F32, 2, GPU_FETCH_FLOAT);
       /* Auto layer name. */
       BLI_snprintf(attr_name, sizeof(attr_name), "a%s", attr_safe_name);
-      GPU_vertformat_alias_add(&format, attr_name);
+      GPU_vertformat_alias_add(format, attr_name);
       /* Active render layer name. */
       if (i == CustomData_get_render_layer(cd_ldata, CD_MLOOPUV)) {
-        GPU_vertformat_alias_add(&format, "u");
+        GPU_vertformat_alias_add(format, "u");
       }
       /* Active display layer name. */
       if (i == CustomData_get_active_layer(cd_ldata, CD_MLOOPUV)) {
-        GPU_vertformat_alias_add(&format, "au");
+        GPU_vertformat_alias_add(format, "au");
         /* Alias to `pos` for edit uvs. */
-        GPU_vertformat_alias_add(&format, "pos");
+        GPU_vertformat_alias_add(format, "pos");
       }
       /* Stencil mask uv layer name. */
       if (i == CustomData_get_stencil_layer(cd_ldata, CD_MLOOPUV)) {
-        GPU_vertformat_alias_add(&format, "mu");
+        GPU_vertformat_alias_add(format, "mu");
       }
     }
   }
 
+  if (format->attr_len == 0) {
+    GPU_vertformat_attr_add(format, "dummy", GPU_COMP_F32, 1, GPU_FETCH_FLOAT);
+    return false;
+  }
+
+  return true;
+}
+
+static void extract_uv_init(const MeshRenderData *mr,
+                            struct MeshBatchCache *cache,
+                            void *buf,
+                            void *UNUSED(tls_data))
+{
+  GPUVertBuf *vbo = static_cast<GPUVertBuf *>(buf);
+  GPUVertFormat format = {0};
+
+  CustomData *cd_ldata = (mr->extract_type == MR_EXTRACT_BMESH) ? &mr->bm->ldata : &mr->me->ldata;
   int v_len = mr->loop_len;
-  if (format.attr_len == 0) {
-    GPU_vertformat_attr_add(&format, "dummy", GPU_COMP_F32, 1, GPU_FETCH_FLOAT);
+  uint32_t uv_layers = cache->cd_used.uv;
+  if (!mesh_extract_uv_format_init(&format, cache, cd_ldata, mr->extract_type, uv_layers)) {
     /* VBO will not be used, only allocate minimum of memory. */
     v_len = 1;
   }
@@ -116,10 +136,45 @@ static void extract_uv_init(const MeshRenderData *mr,
   }
 }
 
+static void extract_uv_init_subdiv(const DRWSubdivCache *subdiv_cache,
+                                   const MeshRenderData *UNUSED(mr),
+                                   struct MeshBatchCache *cache,
+                                   void *buffer,
+                                   void *UNUSED(data))
+{
+  Mesh *coarse_mesh = subdiv_cache->mesh;
+  GPUVertBuf *vbo = static_cast<GPUVertBuf *>(buffer);
+  GPUVertFormat format = {0};
+
+  uint v_len = subdiv_cache->num_subdiv_loops;
+  uint uv_layers;
+  if (!mesh_extract_uv_format_init(
+          &format, cache, &coarse_mesh->ldata, MR_EXTRACT_MESH, uv_layers)) {
+    // TODO(kevindietrich): handle this more gracefully.
+    v_len = 1;
+  }
+
+  GPU_vertbuf_init_build_on_device(vbo, &format, v_len);
+
+  if (uv_layers == 0) {
+    return;
+  }
+
+  /* Index of the UV layer in the compact buffer. Used UV layers are stored in a single buffer. */
+  int pack_layer_index = 0;
+  for (int i = 0; i < MAX_MTFACE; i++) {
+    if (uv_layers & (1 << i)) {
+      const int offset = (int)subdiv_cache->num_subdiv_loops * pack_layer_index++;
+      draw_subdiv_extract_uvs(subdiv_cache, vbo, i, offset);
+    }
+  }
+}
+
 constexpr MeshExtract create_extractor_uv()
 {
   MeshExtract extractor = {nullptr};
   extractor.init = extract_uv_init;
+  extractor.init_subdiv = extract_uv_init_subdiv;
   extractor.data_type = MR_DATA_NONE;
   extractor.data_size = 0;
   extractor.use_threading = false;