Fixes an issue that appeared when trying to compose view transforms from
multiple poses. (See this sample for a demonstration of the issue.) This was
primarily caused by re-using the same viewerPose object for each pose.
The entire system was built on fragile special cases, though, and it was badly
in need of a more general approach.
Input spaces are still a special case but don't really need to be
anymore. I just don't want to make this PR any bigger.
Fixes an issue that appeared when trying to compose view transforms from multiple poses. (See this sample for a demonstration of the issue.) This was primarily caused by re-using the same viewerPose object for each pose. The entire system was built on fragile special cases, though, and it was badly in need of a more general approach.
Input spaces are still a special case but don't really need to be anymore. I just don't want to make this PR any bigger.