Change Details

When transforming a Mesh into Edit Mode, some updates should be ignored in favor of optimization. Among the most time-consuming operations are called operations due to `DEG_id_tag_update(tc->obedit->data, ID_RECALC_GEOMETRY)`: - deg_evaluate_copy_on_write - The Mesh, although not changed, is completely copied and the batch cache is completely destroyed. - BKE_object_data_batch_cache_dirty_tag - Tags the batch cache to be redone This patch proposes a solution to skip mesh CoW during a transformation and not tag dirty All. **Depsgraph Changes:** Currently, the graph can be compared to this simplified one: {F10187860} Where `DEG_id_tag_update(id, ID_RECALC_GEOMETRY)` would trigger `geom_eval_mesh` and `geom_eval_obj_init`. And `DEG_id_tag_update(id, ID_RECALC_SELECT)` would trigger `tag_dirty_select_mesh` and `tag_dirty_select_obj`. This patch proposes to separate `"tag_dirty"` from `"geom_eval"` and create a separate node for it. In addition to adding a option to `tag_dirty_deform_only` The graph becomes something like this: {F10187861} The new `ID_RECALC_GEOMETRY_DEFORM` also doesn't trigger CoW. **Disadvantages:** - Some might say that the graph is too complex and difficult to maintain. - `ID_RECALC_GEOMETRY_DEFORM` could further crash if misused since the loose element cache would not be freed. **Alternative solution:** Instead of using `DEG_id_tag_update` we could update everything we need in the operator itself and somehow trigger the Geometry dependent objects to be updated.

During a mesh transformation in edit mode (Move, Rotate...), not the entire draw cache needs to be recreated. But this occurs because `DEG_id_tag_update(tc->obedit->data, ID_RECALC_GEOMETRY)`, chain a call to `BKE_object_data_batch_cache_dirty_tag` that tags all the batch cache to be redone This patch proposes not tag dirty All, but only what participates in the deformation of geometry. **Depsgraph Changes:** Currently, the graph can be compared to this simplified one: {F10187860} Where `DEG_id_tag_update(id, ID_RECALC_GEOMETRY)` triggers `geom_eval_mesh` or `geom_eval_obj_init`. And `DEG_id_tag_update(id, ID_RECALC_SELECT)` triggers `update_select` or `update_select_obj`. This patch proposes to separate `"tag_dirty"` from `"geom_eval"` and create a separate node for it (`update_all` in the image bellow). In addition to adding an node to `update_deform` The graph becomes something like this: {F10194537} **Benchmarking:** ||master:|patch: |---|---|---| |large_mesh_editing:|Average: 16.727632 FPS|Average: 26.424897 FPS ||rdata 9ms iter 26ms (frame 60ms)|rdata 0ms iter 19ms (frame 38ms) |large_mesh_editing_ledge:|Average: 17.761902 FPS|Average: 28.070558 FPS ||rdata 9ms iter 24ms (frame 56ms)|rdata 0ms iter 18ms (frame 36ms) |looptris_test:|Average: 5.537827 FPS|Average: 5.456050 FPS ||rdata 11ms iter 26ms (frame 169ms)|rdata 11ms iter 28ms (frame 172ms) |subdiv_mesh_cage_and_final:|Average: 2.095824 FPS|Average: 2.140402 FPS ||rdata 7ms iter 21ms (frame 242ms)|rdata 0ms iter 20ms (frame 237ms) ||rdata 7ms iter 22ms (frame 233ms)|rdata 0ms iter 21ms (frame 227ms) |subdiv_mesh_final_only:|Average: 6.626541 FPS|Average: 7.974115 FPS ||rdata 3ms iter 13ms (frame 145ms)|rdata 0ms iter 10ms (frame 122ms) |subdiv_mesh_final_only_ledge:|Average: 6.590914 FPS|Average: 7.978224 FPS ||rdata 3ms iter 13ms (frame 143ms)|rdata 0ms iter 10ms (frame 121ms)

WhenDuring a mesh transforming a Mesh into Edit Modeation in edit mode (Move, Rotate...), some updates should be ignored in favor of optimizationnot the entire draw cache needs to be recreated. Among the most time-consuming operations are called operations due to `DEG_id_tag_update(tc->obedit->data, ID_RECALC_GEOMETRY)`: - deg_evaluate_copy_on_write - The MeshBut this occurs because `DEG_id_tag_update(tc->obedit->data, although not changedID_RECALC_GEOMETRY)`, is completely copied and the batch cache is completely destroyed. - chain a call to `BKE_object_data_batch_cache_dirty_tag - Tags` that tags all the batch cache to be redone This patch proposes a solution to skip mesh CoW during a transnot tag dirty All, but only what participates in the deformation and not tag dirty Alln of geometry. **Depsgraph Changes:** Currently, the graph can be compared to this simplified one: {F10187860} Where `DEG_id_tag_update(id, ID_RECALC_GEOMETRY)` would triggers `geom_eval_mesh` andor `geom_eval_obj_init`. And `DEG_id_tag_update(id, ID_RECALC_SELECT)` would trigger `tag_dirtys `update_select_mesh` and `tag_dirty` or `update_select_obj`. This patch proposes to separate `"tag_dirty"` from `"geom_eval"` and create a separate node for it (`update_all` in the image bellow). In addition to adding a optionan node to `tag_dirtyupdate_deform_only`` The graph becomes something like this: {F10187861}94537} The new `ID_RECALC_GEOMETRY_DEFORM` also doesn't trigger CoW. **Disadvantages:****Benchmarking:** - Some might say that the graph is too complex and difficult to maintain.||master:|patch: - `ID_RECALC_GEOMETRY_DEFORM` could further crash if misused since the loose element cache would not be freed. **Alternative solution:**|---|---|---| |large_mesh_editing:|Average: 16.727632 FPS|Average: 26.424897 FPS ||rdata 9ms iter 26ms (frame 60ms)|rdata 0ms iter 19ms (frame 38ms) |large_mesh_editing_ledge:|Average: 17.761902 FPS|Average: 28.070558 FPS ||rdata 9ms iter 24ms (frame 56ms)|rdata 0ms iter 18ms (frame 36ms) |looptris_test:|Average: 5.537827 FPS|Average: 5.456050 FPS ||rdata 11ms iter 26ms (frame 169ms)|rdata 11ms iter 28ms (frame 172ms) |subdiv_mesh_cage_and_final:|Average: 2.095824 FPS|Average: 2.140402 FPS ||rdata 7ms iter 21ms (frame 242ms)|rdata 0ms iter 20ms (frame 237ms) ||rdata 7ms iter 22ms (frame 233ms)|rdata 0ms iter 21ms (frame 227ms) |subdiv_mesh_final_only:|Average: 6.626541 FPS|Average: 7.974115 FPS ||rdata 3ms iter 13ms (frame 145ms)|rdata 0ms iter 10ms (frame 122ms) |subdiv_mesh_final_only_ledge:|Average: 6.590914 FPS|Average: 7.978224 FPS Instead of using `DEG_id_tag_update` we could update everything we need in the operator itself and somehow trigger the Geometry dependent objects to be updated.||rdata 3ms iter 13ms (frame 143ms)|rdata 0ms iter 10ms (frame 121ms)