For some applications like detection tasks we may want to localize a small object precisely. Currently, we can only determine what face a ray intersects with. However, it's mathematically simple to determine the exact location of the intersection. This information is actually computed in PyTorch3D (another reason to work on #117) as the barycentric return of the rasterizer. We could also write our own code to take the ray and the triangle it intersects to compute the barycentric coordinates. An example that seems somewhat relevant is here.
For some applications like detection tasks we may want to localize a small object precisely. Currently, we can only determine what face a ray intersects with. However, it's mathematically simple to determine the exact location of the intersection. This information is actually computed in
PyTorch3D(another reason to work on #117) as thebarycentricreturn of the rasterizer. We could also write our own code to take the ray and the triangle it intersects to compute the barycentric coordinates. An example that seems somewhat relevant is here.