LTX-Video is an official repository for a diffusion transformer (DiT) model that generates videos from text prompts or input images. It provides the model weights, inference code, and is integrated with ComfyUI for workflow automation. The repository also links to companion training tools (LTX-Video-Trainer) for fine-tuning and customization. The model is available on HuggingFace and has a hosted demo on LTX Studio.