zsxkib / infinite-you

Transform your portrait photos into any style or setting while preserving your facial identity

  • Public
  • 1.1K runs
  • GitHub
  • Weights
  • Paper
  • License
Run with an API
Playground API Examples README Versions

Examples

View more

Run time and cost

This model costs approximately $0.062 to run on Replicate, or 16 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia L40S GPU hardware. Predictions typically complete within 65 seconds. The predict time for this model varies significantly based on the inputs.

Readme

InfiniteYou-FLUX

InfiniteYou is an AI model that preserves your facial identity while transforming your portrait into new contexts, styles, and settings.

About This Model

InfiniteYou addresses the challenges of identity-preserved image generation using Diffusion Transformer (DiT) technology. Key features include:

  • Identity Preservation: Maintains your facial identity while generating entirely new images
  • High Text-Image Alignment: Accurately reflects the details in your text prompts
  • Natural Integration: Avoids awkward face-pasting issues common in other models
  • Two Model Variants:
  • aes_stage2: Better text-image alignment and aesthetic quality
  • sim_stage1: Higher identity similarity for stronger likeness preservation

The model works by injecting identity features into the FLUX.1-dev base model via residual connections, allowing it to maintain your identity while applying new styles, backgrounds, clothing, and contexts.

Model Parameters

  • Model Version: Choose between identity similarity (sim_stage1) or better aesthetics (aes_stage2)
  • Enhancement Options: Optional realism and anti-blur LoRAs for additional quality improvements
  • Advanced Controls: Adjust identity strength and guidance parameters for fine-tuned results

Technical Architecture

InfiniteYou uses InfuseNet, a component that injects identity features into the DiT base model through residual connections. The model underwent a multi-stage training process including pretraining and supervised fine-tuning with synthetic single-person-multiple-sample data.

The plug-and-play design ensures compatibility with various approaches including ControlNets, LoRAs, and other enhancement techniques.

License & Disclaimer

Important: This model is for non-commercial use only and is released under the Creative Commons Attribution-NonCommercial 4.0 International Public License for academic research purposes only.

The model makes use of several components including: - Face models from InsightFace - FLUX.1-dev base model - Optional LoRAs (Realism and Anti-blur)

Use of these components must follow their original licenses.

This research aims to positively impact the field of Generative AI. While users can create images with this tool, they must comply with local laws and use it responsibly. The developers do not assume responsibility for potential misuse.

Citation

If you use this model in your research, please cite:

@article{jiang2025infiniteyou,
  title={{InfiniteYou}: Flexible Photo Recrafting While Preserving Your Identity},
  author={Jiang, Liming and Yan, Qing and Jia, Yumin and Liu, Zichuan and Kang, Hao and Lu, Xin},
  journal={arXiv preprint},
  volume={arXiv:2503.16418},
  year={2025}
}