Can anyone explain this to me? I have been stuck on this recentering questions for long time.
I understand from the matrix multiplication that the recentered poses map all coordinates from different cameras to one average camera space, but the relative difference of these poses are the same, so how can this recentering benefit?
The author said 'recenter_poses very simply applies the inverse of this average pose to the dataset (a rigid rotation/translation) so that the identity extrinsic matrix is looking at the scene, which is nice because normalizes the orientation of the scene for later rendering from the learned NeRF' https://github.com/bmild/nerf/issues/34#issuecomment-616175083 . My question is, why 'the identity extrinsic matrix is looking at the scene' is important? And why it 'normalizes the orientation of the scene' ?
Can anyone explain this to me? I have been stuck on this recentering questions for long time.
I understand from the matrix multiplication that the recentered poses map all coordinates from different cameras to one average camera space, but the relative difference of these poses are the same, so how can this recentering benefit?
The author said 'recenter_poses very simply applies the inverse of this average pose to the dataset (a rigid rotation/translation) so that the identity extrinsic matrix is looking at the scene, which is nice because normalizes the orientation of the scene for later rendering from the learned NeRF' https://github.com/bmild/nerf/issues/34#issuecomment-616175083 . My question is, why 'the identity extrinsic matrix is looking at the scene' is important? And why it 'normalizes the orientation of the scene' ?