LTX-2 Overview
Meet LTX-2
LTX-2 is a DiT-based audio-video foundation model designed to generate synchronized video and audio within a single model. It brings together the core building blocks of modern video generation, with open weights and a focus on practical, local execution.The model supports:
High-Fidelity Generation: Generation of up to ~20 seconds of synchronized audio and video, with support for high resolutions and high frame rates depending on configuration and hardware. LTX-2 is designed to scale from fast iteration to high-quality outputs.
Native Audio-Video Sync: Dialogue, lip movement, and ambient audio are generated together in a single pass, enabling coherent speech, expressive performances, and consistent timing without external alignment steps.
Motion Realism: Dynamic scenes with stable motion, consistent identity, and strong frame-to-frame coherence.
Fine-Grained Control: LoRA-based customization, camera-aware motion logic, and multimodal inputs including text, image, video, audio, and depth for precise creative direction.
Efficient by Design: A compact latent space and step-distilled architecture allow LTX-2 to run efficiently on high-end consumer GPUs, making high-fidelity multimodal generation accessible without specialized infrastructure.
To learn more about LTX-2 architecture, and the research behind it, read our full research paper.
