Difference in result between monocular vs multi-view VHAP tracking

[pratch]

Dear author, I tried driving GaussianAvatars using VHAP tracking in monocular (left) vs multiview (right) mode and got quite different results. May I ask if this is expected?

vhapdiff.mp4