Open
Description
We use UVAtlas (https://github.com/microsoft/UVAtlas), which produces good results but is not parallelized and can take minutes to compute.
We have implemented a naive parallelization by first partitioning the mesh and then running UVAtlas on each part.
This reduces the quality as it produces meshes with more UV islands.
- Analyze the UVAtlas library and identify performance optimizations
- investigate GPU porting. [Algorithms, SYCL]
Benefits all 3D mesh creation and editing tasks.
Metadata
Metadata
Assignees
Type
Projects
Status
Backlog