Generative Media

LTX-2

Open video that comes with its own soundtrack

Open code / restricted weightsModel

What it is

A DiT-based audio-video model — the first open-weights system generating synchronized sound and picture together, up to 4K at 50fps. The 22B LTX-2.3 (with a fast distilled variant) covers text-to-video, image-to-video, interpolation, camera-control LoRAs, and ships an official LoRA trainer.

Why it's interesting

Speed and audio are the edge: distilled variants run near real time on an H100 and 2-3x faster than rivals on consumer cards, and native synchronized audio puts an open-weights model in territory otherwise exclusive to closed systems.

Use cases

  • Fast stylized video with synchronized audio
  • Real-time and interactive video experiments
  • LoRA-customized brand motion

Who it's for

Motion designers, app developers, indie filmmakers

Setup

Moderate. Official Python packages, ComfyUI, or Diffusers; distilled variants from ~8-16GB VRAM, the 22B quality tier 24GB+

Limitations & cautions

Community-licensed weights with scale thresholds; photoreal humans trail Wan 2.2; frame counts must be a multiple of 8 plus 1.

Editorial takeaway

Sound changes everything about generated video — silence was the uncanny part. First open model to figure that out.

Related & alternatives