Open LTX-2 generates 4K video with synchronized audio—the first fully open-source model for creating video and sound sim

Open LTX-2 generates 4K video with synchronized audio—the first fully open-source model for creating video and sound simultaneously. Lightricks has released the weights for LTX-2, a multimodal model for synchronous video and audio generation. Here's what the new model can do: 🕤 Generates native 4K video at 50 FPS with dialogues, music, and sound effects, up to 20 seconds long; 🕤 Maintains character identity throughout the clip without loss of quality; 🕤 Accurately synchronizes lip movements with speech for realistic dialogues; 🕤 Supports control via keyframes and full 3D camera logic; 🕤 Accepts text, images, video, audio, and depth maps as input; 🕤 Model weights, inference pipelines, training code, and full documentation are available; According to blind tests on AI Arena, LTX-2 is the leader among open solutions for video generation. 🔗 Try it out: https://huggingface.co/Lightricks/LTX-2