Information-Regularized Constrained Inversion for Stable Avatar Editing from Sparse Supervision

ArXi:2604.02883v1 Announce Type: new Editing animatable human avatars typically relies on sparse supervision, often a few edited keyframes, yet naively fitting a reconstructed avatar to these edits frequently causes identity leakage and pose-dependent temporal flicker. We argue that these failures are best understood as an ill-conditioned inversion: the available edited constraints do not sufficiently determine the latent directions responsible for the intended edit.