PuLID
Pure and Lightning ID customization for diffusion models without fine-tuning.
About
PuLID, short for Pure and Lightning ID customization, is a ByteDance research project accepted to NeurIPS 2024 that adds high-fidelity identity control to diffusion models without any per-subject fine-tuning. Given a face photo, it generates new images that preserve the person's identity across poses, expressions, styles, and scenes. Its distinguishing technique is contrastive alignment, which minimizes disruption to the base model's behavior so prompts stay editable and outputs remain natural. Implementations cover both SDXL and FLUX, with PuLID-v1.1 improving compatibility, editability, and facial similarity over the first release, and PuLID-FLUX running on consumer GPUs with 16 GB of VRAM. The Apache 2.0 licensed repository provides local Gradio demos alongside hosted Hugging Face demos, and the community has built ComfyUI nodes, WebUI extensions, and Replicate deployments around it. Common uses include consistent-character generation and personalized portrait work.
Reviews (0)
Leave a Review
No reviews yet. Be the first to review!
Details
- Category
- Image Generation
- Price
- Free
- Platform
- Local/Desktop
- Difficulty
- Advanced (4/5)
- License
- Apache-2.0
- Minimum VRAM
- 8 GB
- Added
- Apr 3, 2026
Related Tools
State-of-the-art open image generation model by Black Forest Labs with rectified flow transformers.
Next-generation image generation model by Black Forest Labs.
Bilingual text-to-image diffusion transformer by Tencent with Chinese and English support.
Zero-shot identity-preserving image generation from a single face photo.
Image prompt adapter for pre-trained text-to-image diffusion models.
Neural network architecture for adding spatial control to diffusion models.