New LoRA model preserves identity in reference-to-video generation

A new LoRA model, Alissonerdx/LTX-Best-Face-ID, has been released for LTX-2.3, enabling identity-preserving reference-to-video generation. This model allows users to input a reference photo of a person and a text prompt to create a video where the individual's identity is maintained. It utilizes techniques such as overlap reference conditioning, TASS-RoPE, and an ArcFace identity loss during training to achieve accurate identity transfer. AI

IMPACT Enables more precise identity control in video generation, potentially improving creative tools and synthetic media applications.

RANK_REASON This is a release of a specific model (LoRA) for a particular AI system (LTX-2.3), not a frontier model release from a major lab.

Read on Hugging Face Trending Models →

• Alissonerdx/LTX-Best-Face-ID
• ArcFace
• arXiv:2606.02441
• BFS Nodes
• ComfyUI
• Gemma-vision
• IPT2V
• Lora
• LTX-2.3
• LTX-Best-Face-ID
• St-DRC

• model release
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →