LTX-2.3 22B IC-LoRA LipDub

This is an IC-LoRA trained on top of LTX-2.3-22b, enabling lip dubbing on top of the LTX model.

It is based on the LTX-2 foundation model.

• Project Page: JustDubIt project page
• Paper: JustDubIt paper
• Code: GitHub Repository
• Inference Pipeline: packages/ltx-pipelines/src/ltx_pipelines/lipdub.py

ltx-2.3-22b-ic-lora-lipdub-0.9.safetensors

See the LTX-2-community-license for full terms.

• Base Model: LTX-2.3
• Training Type: IC-LoRA
• Control Type: Video & Audio
• Reference Downscale Factor: 1 (reference resolution is the same as output resolution)

1. Copy the LoRA weights into models/loras.
2. Use the official lipdub workflow from the LTX-2 ComfyUI repository.

The model was trained using a lip-dubbing dataset.

@article{chen2026just,
  title={JUST-DUB-IT: Video Dubbing via Joint Audio-Visual Diffusion},
  author={Chen, Anthony and Korem, Naomi Ken and Zeevi, Gal and Halperin, Tavi and Yosef, Matan Ben and Jelercic, Urska and Bibi, Ofir and Patashnik, Or and Cohen-Or, Daniel},
  journal={arXiv preprint arXiv:2601.22143},
  year={2026}
}

• Base model by Lightricks
• Training infrastructure: LTX-2 Community Trainer