InstructHumans: Editing Animated 3D Human Textures with Instructions
arXiv:2404.04037v1 Announce Type: cross
Abstract: We present InstructHumans, a novel framework for instruction-driven 3D human texture editing. Existing text-based editing methods use Score Distillation Sampling (SDS) to distill guidance from generative models. This work shows that naively using such scores is harmful to editing as they destroy consistency with the source avatar. Instead, we propose an alternate SDS for Editing (SDS-E) that selectively incorporates subterms of SDS across diffusion timesteps. We further enhance SDS-E with spatial smoothness regularization and gradient-based viewpoint sampling to achieve high-quality edits with sharp and high-fidelity detailing. InstructHumans significantly outperforms existing 3D editing methods, consistent with the initial avatar while faithful to the textual instructions. Project page: https://jyzhu.top/instruct-humans .