FantasyID: Face Knowledge Enhanced ID-Preserving Video Generation

TL;DR

FantasyID is a tuning-free video generation framework that enhances face knowledge in diffusion transformers, ensuring identity preservation and realistic facial dynamics through 3D priors and adaptive feature injection.

Contribution

It introduces a novel face knowledge enhancement method with 3D priors and adaptive feature guidance for identity-preserving text-to-video synthesis.

Findings

Outperforms existing tuning-free IPT2V methods.

Effectively maintains facial identity during dynamic video synthesis.

Improves facial expression and pose diversity in generated videos.

Abstract

Tuning-free approaches adapting large-scale pre-trained video diffusion models for identity-preserving text-to-video generation (IPT2V) have gained popularity recently due to their efficacy and scalability. However, significant challenges remain to achieve satisfied facial dynamics while keeping the identity unchanged. In this work, we present a novel tuning-free IPT2V framework by enhancing face knowledge of the pre-trained video model built on diffusion transformers (DiT), dubbed FantasyID. Essentially, 3D facial geometry prior is incorporated to ensure plausible facial structures during video synthesis. To prevent the model from learning copy-paste shortcuts that simply replicate reference face across frames, a multi-view face augmentation strategy is devised to capture diverse 2D facial appearance features, hence increasing the dynamics over the facial expressions and head poses. Additionally, after blending the 2D and 3D features as guidance, instead of naively employing cross-attention to inject guidance cues into DiT layers, a learnable layer-aware adaptive mechanism is employed to selectively inject the fused features into each individual DiT layers, facilitating balanced modeling of identity preservation and motion dynamics. Experimental results validate our model's superiority over the current tuning-free IPT2V methods.