Identity-Consistent Video Generation under Large Facial-Angle Variations

TL;DR

This paper introduces a multi-view conditioned framework, $ ext{Mv}^2 ext{ID}$, for identity-consistent video generation under large facial-angle variations, effectively balancing identity preservation and motion naturalness without costly cross-paired data.

Contribution

The paper proposes a novel multi-view conditioned approach with region-masking and reference decoupled-RoPE mechanisms, enabling identity-consistent video synthesis with in-paired supervision.

Findings

Significantly improves identity consistency in generated videos.

Maintains natural facial motion despite large angle variations.

Outperforms existing methods trained with cross-paired data.

Abstract

Single-view reference-to-video methods often struggle to preserve identity consistency under large facial-angle variations. This limitation naturally motivates the incorporation of multi-view facial references. However, simply introducing additional reference images exacerbates the \textit{copy-paste} problem, particularly the \textbf{\textit{view-dependent copy-paste}} artifact, which reduces facial motion naturalness. Although cross-paired data can alleviate this issue, collecting such data is costly. To balance the consistency and naturalness, we propose $\mathrm{Mv}^2\mathrm{ID}$, a multi-view conditioned framework under in-paired supervision. We introduce a region-masking training strategy to prevent shortcut learning and extract essential identity features by encouraging the model to aggregate complementary identity cues across views. In addition, we design a reference decoupled-RoPE mechanism that assigns distinct positional encoding to video and conditioning tokens for better modeling of their heterogeneous properties. Furthermore, we construct a large-scale dataset with diverse facial-angle variations and propose dedicated evaluation metrics for identity consistency and motion naturalness. Extensive experiments demonstrate that our method significantly improves identity consistency while maintaining motion naturalness, outperforming existing approaches trained with cross-paired data.