# FantasyID: Face Knowledge Enhanced ID-Preserving Video Generation

arXiv: https://arxiv.org/abs/2502.13995 · HTML: https://mlsanity.com/paper/2502.13995/

## Key facts

- Cited by 18 papers in the map (data through 14 Oct 2026).
- Placed in Image generation, text-to-image and editing, one of 14 directions in Diffusion models.
- First posted on arXiv 19 Feb 2025.

## Placed in

- Diffusion models / [Image generation, text-to-image and editing](https://mlsanity.com/diffusion/image-generation-text-to-image-and-editing/)

## Built on by (most cited first, 18 of 18)

- [OpenS2V-Nexus: A Detailed Benchmark and Million-Scale Dataset for Subject-to-Video Generation](https://mlsanity.com/paper/2505.20292/) (2025)
- [DreamRelation: Relation-Centric Video Customization](https://mlsanity.com/paper/2503.07602/) (2025)
- [MAGREF: Masked Guidance for Any-Reference Video Generation with Subject Disentanglement](https://mlsanity.com/paper/2505.23742/) (2025)
- [FantasyPortrait: Enhancing Multi-Character Portrait Animation with Expression-Augmented Diffusion Transformers](https://mlsanity.com/paper/2507.12956/) (2025)
- [SynMotion: Semantic-Visual Adaptation for Motion Customized Video Generation](https://mlsanity.com/paper/2506.23690/) (2025)
- [Identity-Preserving Image-to-Video Generation via Reward-Guided Optimization](https://mlsanity.com/paper/2510.14255/) (2025)
- [Evolution of Video Generative Foundations](https://mlsanity.com/paper/2604.06339/) (2026)
- [ARGUS: Stacked Multi-View Identity Mosaic Injection for Subject-Preserving Video Generation](https://mlsanity.com/paper/2606.11670/) (2026)
- [Proteus-ID: ID-Consistent and Motion-Coherent Video Customization](https://mlsanity.com/paper/2506.23729/) (2025)
- [LumosX: Relate Any Identities with Their Attributes for Personalized Video Generation](https://mlsanity.com/paper/2603.20192/) (2026)
- [MoSA: Motion-Coherent Human Video Generation via Structure-Appearance Decoupling](https://mlsanity.com/paper/2508.17404/) (2025)
- [FlashI2V: Fourier-Guided Latent Shifting Prevents Conditional Image Leakage in Image-to-Video Generation](https://mlsanity.com/paper/2509.25187/) (2025)
- [DreamVideo-Omni: Omni-Motion Controlled Multi-Subject Video Customization with Latent Identity Reinforcement Learning](https://mlsanity.com/paper/2603.12257/) (2026)
- [FantasyTalking2: Timestep-Layer Adaptive Preference Optimization for Audio-Driven Portrait Animation](https://mlsanity.com/paper/2508.11255/) (2025)
- [Soul: Breathe Life into Digital Human for High-fidelity Long-term Multimodal Animation](https://mlsanity.com/paper/2512.13495/) (2025)
- [SwapAnyone: Consistent and Realistic Video Synthesis for Swapping Any Person into Any Video](https://mlsanity.com/paper/2503.09154/) (2025)
- [Collaborative Face Experts Fusion in Video Generation: Boosting Identity Consistency Across Large Face Poses](https://mlsanity.com/paper/2508.09476/) (2025)
- [FaithfulFaces: Pose-Faithful Facial Identity Preservation for Text-to-Video Generation](https://mlsanity.com/paper/2605.04702/) (2026)

## Abstract (arXiv)

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.