ltx-video

LTX-Video is an open-source video generation model based on DiT (Diffusion Transformer) architecture, developed by Lightricks. It unifies multiple video generation capabilities into a single model. Core capabilities: - Text-to-video and image-to-video generation - Video extension in both forward and backward directions - Multi-keyframe conditioning and keyframe-based animation - Video-to-video transformations - Synchronized audio generation (available in LTX-2) - Native 4K resolution output up to 50 FPS Model variants: - 13B and 2B parameter versions - Distilled models for faster inference (HD video in seconds) - FP8 quantized versions for reduced VRAM usage - 2B distilled LoRA variant requiring only 1GB VRAM Integrations and ecosystem: - ComfyUI and Diffusers support - Python library API - Control models for depth, pose, and Canny edge conditioning - LoRA fine-tuning for style customization - Community tools: ComfyUI-LTXTricks (RF-Inversion, RF-Edit, FlowEdit), LTX-VideoQ8 (8-bit optimized inference), TeaCache (inference acceleration) - OpenRAIL-M license allowing commercial use Successor LTX-2 adds synchronized audio-video generation, longer clip support, and enhanced creative controls.