LTX 2 Retake Video for Video Editing

LTX‑2 Retake Video is a video‑to‑video editing mode that lets you “reshoot” only part of an existing clip with a prompt, while keeping the rest of the shot’s look, motion, and timing intact.​

What Retake actually does

• Segment‑level editing: You pick a time range (for example seconds 4–8), and only that segment is regenerated; the rest of the clip stays untouched.​

• Continuity preservation: Retake matches camera movement, lighting, character design, and motion direction so the edited segment blends seamlessly into the surrounding frames.​

• Prompt‑based redirection: You use text instructions to change actions, framing, emotion, dialogue, or even remove/replace objects in that segment.​

Core controls and modes

• Time range

  • Choose start_time and duration (typically 2–20 seconds depending on the platform).​

• Replacement mode

  • Video‑only: change visuals, keep original audio.

  • Audio‑only: regenerate voice/dialogue/ambience, keep visuals.

  • Audio+video: fully re‑interpret that moment.​

• Use cases

  • Fix a gesture, reaction, or timing without redoing the whole shot.

  • Remove an object or person from a few seconds of footage.

  • Change emotional tone or camera framing for an ending.​

Typical Retake workflow

• Import or select your video in an LTX‑2–enabled tool (LTX Studio, RunDiffusion, WaveSpeed, etc.).​

• Mark the exact section to edit with in/out points or timecodes.​​

• Write a focused prompt describing how that moment should change (for example “make the character look shocked and step back quickly” or “remove the red car and show only the street”).​

• Choose replacement mode (video, audio, or both) and run Retake; only those seconds are re‑rendered and blended back into the original clip.​​

Where Retake fits in a workflow

• After initial generation with LTX‑2 Fast/Pro/Ultra, you use Retake to surgically fix problem moments instead of re‑rolling full clips.​

• Acts like a director’s tool: closer to doing multiple “takes” on set than traditional NLE trimming or masking, because performance, pacing, and framing can all be re‑interpreted with text.​

If you describe what kind of edits you want (fixing endings, removing objects, changing emotion, adjusting camera), the next step can be concrete Retake prompt patterns and guardrails for those scenarios.