Only leaving CL 26157603 reverted since accessing GenerateSource in FDistanceFieldVolumeData::CacheDerivedData is not safe due to async mesh compilation.
#prelight 64948190a26e02126bc0a64f
#rb Krzysztof.Narkowicz
[CL 26206875 by tiago costa in ue5-main branch]
[FYI] tiago.costa
Original CL Desc
-----------------------------------------------------------------
Use FSignedDistanceFieldBuildSectionData when generating distance fields for Nanite meshes.
Context:
- Nanite meshes distance fields are generated using the original mesh data instead of the coarse representation generated by Nanite, however that code path was not checking section data such as blend mode, two sided, etc.
Fix:
- Modified FStaticMeshBuilder::BuildMeshVertexPositions(...) to also output a FStaticMeshSectionArray.
- Updated MeshRepresentation::SetupEmbreeScene(...) to check section flags when using SourceMeshData.IsValid() code path.
#prelight
#rb Krzysztof.Narkowicz
[CL 26174853 by bob tellez in ue5-main branch]
Context:
- Nanite meshes distance fields are generated using the original mesh data instead of the coarse representation generated by Nanite, however that code path was not checking section data such as blend mode, two sided, etc.
Fix:
- Modified FStaticMeshBuilder::BuildMeshVertexPositions(...) to also output a FStaticMeshSectionArray.
- Updated MeshRepresentation::SetupEmbreeScene(...) to check section flags when using SourceMeshData.IsValid() code path.
#prelight
#rb Krzysztof.Narkowicz
[CL 26160411 by tiago costa in ue5-main branch]
Highlights:
* Only allocate and use a wedge map for non-Nanite LODs, or if the mesh is sub 300kindices and it will be cache optimized
* Only allocate and use a remap if there are no overlapping corners
* When the Nanite landscape mesh is created, we now ensure the mesh description is fully compacted, computed, and optimized prior to committing it, along with making sure there are no overlapping corners.
* Since Nanite landscape is checked to ensure no overlapping corners before the build, we can also now skip computing the overlapping corners, which was an intense number of tiny allocations.
* When Nanite landscape is being built as a static mesh, we now avoid super expensive temporary copies of the mesh description, knowing that no actual mutation will occur.
* The original code (after building) would release the memory from the mesh description copy, but it would not release the original immutable cached copy from memory (now that is released too)
* Fixed NaniteLODIndex incorrectly creating a mesh description at a LOD other than the expected 0
* Removed a full loop over all vertex positions by calculating the bounding box while we're building the vertex buffer (and already processing every vertex)
* Added a NeedsCompact() helper to MeshDescription to avoid expensive and unnecessary Compact() calls that do lots of reallocs/copies even if the mesh description is already contiguous\packed in memory.
* Skip building distance field and mesh card representations for the Nanite landscape mesh, which also avoids more cached mesh description deserialization and copies into memory.
Note: In the near future the APIs around this wil be cleaned up further.
#rb richard.talbotwatkin, yuriy.odonnell
[FYI] zousar.shaker, brian.karis, rune.stubbe, juan.canada, danny.couture
#preflight 640b7390482188d710464d39
[CL 24601442 by graham wihlidal in ue5-main branch]
-Nanite-specific settings for tangent and degenerates are no longer also applied to the non-nanite mesh LODs on the same mesh.
-100% fallback meshes now have identical tangents to input mesh. Don't skip the tangent generation in StaticMeshBuilder for these meshes and don't recalculate tangents in Nanite builder.
#preflight 632b04c0b40000c8f0c29adb
#rb brian.karis
[FYI] jamie.hayes, graham.wihlidal
#lockdown marc.audy
[CL 22201439 by rune stubbe in ue5-main branch]
Tangent generation now also respects the Nanite project setting, not just the per-mesh setting.
#lockdown marc.audy
[CL 21862271 by rune stubbe in ue5-main branch]
Headers are updated to contain any missing #includes needed to compile and #includes are sorted. Nothing is removed.
#ushell-cherrypick of 21064294 by bryan.sefcik
#jira
#preflight 62d5c2111062f2e63015e598
#ROBOMERGE-OWNER: bryan.sefcik
#ROBOMERGE-AUTHOR: bryan.sefcik
#ROBOMERGE-SOURCE: CL 21155249 via CL 21158121 via CL 21161259
#ROBOMERGE-BOT: UE5 (Release-Engine-Staging -> Main) (v972-20964824)
[CL 21182053 by bryan sefcik in ue5-main branch]
- Cook side extra low lod coarse mesh will be created at 1/8th of the original coarse mesh if mesh streaming or editor data is available for the platform
- Nanite static meshes with streamable LODs are registered to a Nanite::CoarseMeshStreamingManager which will handle the stream in/out requests for these assets in fixed memory budget (memory budget is part of the total mesh streaming budget)
- Nanite proxy needs to handle different raytracing proxy LOD indices now for dynamic and cached indices
- What to stream is driven by what's used in the TLAS build - each deferred render pass all used streamable meshes are collected and forwarded to the CoarseMeshStreamingManager
- CoarseMeshStreamingManager is updated after all views are rendered and will make stream in/out requests depending on the TLAS usage
- When LODs of UStaticMeshes renderdata is loaded or unloaded then a rebuild of the cached render data is requested to make sure the correct LOD & RT BLAS is used (to do this the UStaticMeshComponent are registered to the UStaticMesh so a request for rebuild can be made for all the proxies on the scene)
#preflight 612659e872e9eb00010a27d9
#ROBOMERGE-SOURCE: CL 17314263
#ROBOMERGE-BOT: (v861-17282326)
[CL 17315687 by kenzo terelst in ue5-main branch]
This is later used to build static mesh derived data in cases where Nanite coarse mesh representation replaces original static mesh data and we can't depend on it. In extreme cases it lowers main thread derived data build time from ~7.5s to ~0.1s.
#rb Richard.TalbotWatkin
[CL 14578414 by Krzysztof Narkowicz in ue5-main branch]
