Question about USkeletalMesh internals (FRigidVertex and FSoftVertex)


I am looking for some advice from people who have already worked with the USkeletalMesh class in C++.

For a project I am working on I have to generate skeletal meshes from source meshes at runtime. Animations will be done with a custom AnimationInstance implementation.

Basically I load a simple skeletal mesh, which serves as a template, and create a new skeletal mesh in C++ with N (N=around 200) bones.
The template mesh is repeated N times. This way I need only one single SkeletalMesh to render hundreds of animated simple meshes.

I managed to get the generated mesh to render( with some inspiration from the SkeletalMeshMerge class) but I am facing one little problem, which has to do with the reference pose of the mesh.

Each USkeletalMesh contains FSkelMeshChunk and FSkelMeshSection entries.

A FSkelMeshChunk entry contains two arrays:

TArray<FRigidVertex> RigidVertices; // influenced by a single bone
TArray<FSoftVertex> SoftVertices; // influenced by multiple bones ( blend weights)

My question: Are the vertices stored in FSkelMeshChunk transformed to reference pos space ?

Looking at the SkeletalMeshMerge.cpp, there they extract the vertex position from the vertex buffer and not the FSkelMeshChunk.

const TGPUSkinVertexBase<bExtraBoneInfluencesT>* SrcBaseVert = SrcLODModel.VertexBufferGPUSkin.GetVertexPtr<bExtraBoneInfluencesT>(VertIdx);
				DestVert.Position = SrcLODModel.VertexBufferGPUSkin.GetVertexPositionFast<bExtraBoneInfluencesT>(SrcBaseVert);

Do I have to transform the vertices back to refpose(source mesh)->bone space->refpose(new mesh)?

Thank you for your help in advance