Auto Facial Rigging of Digital Humans Based on Normal Constrained Skinning Decomposition

This paper proposes an improved auto facial rigging method based on normal constrained skinning decomposition, aiming to enhance the presentation of digital human characters in virtual reality (VR). Compared to traditional skinning decomposition-based rigging methods, our approach introduces normal constraints. The normal constraint is used to avoid the "wave-like surface" issue that occurs in existing rigging methods. Our approach is especially effective in VR applications as it greatly enhances the authenticity of digital human expression and user immersion. Experimental results demonstrate that our proposed method outperforms existing approaches in replicating input character expressions, both qualitatively and quantitatively, showcasing significant practical value.