LTX 2 Fast API for Text to Video

LTX‑2 Fast is the high‑speed text‑to‑video mode in the LTX‑2 family, built to turn prompts into 6–20 second cinematic clips with synchronized audio in roughly “real‑time” for brainstorming and rapid iteration.​

What LTX‑2 Fast is

• A distilled “Fast” flow of the LTX‑2 Diffusion‑Transformer video model, optimized for much lower latency while keeping motion quality and coherence high.​

• Text‑first: you give only a prompt, and it outputs a finished MP4 with video plus audio in one pass, suitable for drafts, storyboards, and social clips.​

Core capabilities

• Duration

  • Supports 6, 8, 10 seconds, and some APIs extend up to 20 seconds for Fast mode.​​

• Resolution & FPS

  • Common settings: 1080p, 1440p, or 2160p (4K) at 16:9, typically 25 or 50 fps; some deployments run 4K at ~24–48 fps.​

• Audio

  • Native synchronized audio (music/ambience/SFX) so you do not need a separate sound pass; generate_audio can usually be toggled on/off.​

Typical text‑to‑video workflow

• Provide a concise but structured prompt: subject, environment, camera move, motion type, and mood (for example “handheld vlog”, “cinematic tracking shot”, “anime fight”).​

• Choose:

  • Duration: 6/8/10 s (or up to 20 s where supported).​​

  • Resolution: usually 1080p for ideation; 1440p/2160p when you need sharper output.​

  • FPS: 25 for general content, 50 for extra‑smooth motion or slow‑downs.​​

• Generate, review, then iterate quickly by tweaking only one aspect at a time (motion, style, or framing) to converge on a usable shot.
  

Where it fits in workflows

• Excellent for rapid shot exploration before committing to Pro/Ultra re‑renders.​

• Suits high‑volume pipelines (many prompt variations, A/B tests, iterative storyboards) where low cost and fast turnaround matter more than absolute maximum fidelity.​

If you describe the kinds of shots you want (product ads, character moments, B‑roll, gameplay‑style clips), guidance can focus on LTX‑2 Fast prompt patterns and settings tailored to those.