WildActor: Unconstrained Identity-Preserving Video Generation

TL;DR

WildActor introduces a new framework for generating human videos that maintain consistent full-body identities across diverse viewpoints and motions, overcoming limitations of face-centric and pose-locked methods.

Contribution

The paper presents Actor-18M dataset and WildActor framework, enabling unconstrained, identity-preserving human video generation with novel attention and sampling strategies.

Findings

WildActor outperforms existing methods in identity preservation across viewpoints.

The Actor-18M dataset provides extensive data for training and evaluation.

WildActor maintains consistent full-body identities in complex motion and viewpoint changes.

Abstract

Production-ready human video generation requires digital actors to maintain strictly consistent full-body identities across dynamic shots, viewpoints and motions, a setting that remains challenging for existing methods. Prior methods often suffer from face-centric behavior that neglects body-level consistency, or produce copy-paste artifacts where subjects appear rigid due to pose locking. We present Actor-18M, a large-scale human video dataset designed to capture identity consistency under unconstrained viewpoints and environments. Actor-18M comprises 1.6M videos with 18M corresponding human images, covering both arbitrary views and canonical three-view representations. Leveraging Actor-18M, we propose WildActor, a framework for any-view conditioned human video generation. We introduce an Asymmetric Identity-Preserving Attention mechanism coupled with a Viewpoint-Adaptive Monte Carlo Sampling strategy that iteratively re-weights reference conditions by marginal utility for balanced manifold coverage. Evaluated on the proposed Actor-Bench, WildActor consistently preserves body identity under diverse shot compositions, large viewpoint transitions, and substantial motions, surpassing existing methods in these challenging settings.