How to Edit AI Video in Runway Gen-4 Without Regenerating
Runway Gen-4 includes camera control, clip extension, upscaling, and Act-Two. Here is when to edit in place instead of burning credits on a fresh generation.
Runway Act-Two Paid
Runway's performance-capture tool that drives any character with your face and body
Runway Act-Two is a performance-capture model that transfers an actor’s facial expressions, hand gestures and body motion onto any character — a photo, an illustration, or a 3D-rigged avatar. It is included with Runway plans starting around $15 per month.
What is Runway Act-Two?
Announced alongside Runway’s Gen-4 and Aleph work, Act-Two focuses on a narrower problem than full video generation: getting believable human performance onto a target character. You record yourself with a webcam or phone, and the model transfers that motion. Because it is transfer rather than generation, results stay closer to the source performance than text-to-video does, which directors find more controllable.
What does Act-Two actually do?
- Facial expression transfer, including eyebrow and lip detail
- Body and hand motion capture from a single reference video
- Character swapping onto stills, artwork or pre-rigged 3D avatars
- Real-time and recorded reference performance modes
- Works with the Runway Gen-4 pipeline for finishing shots
- Available in the Runway web app and the API
Why directors use it instead of full video generation
Text-to-video is unpredictable. Act-Two keeps the actor’s timing and intent, which is what you need for dialogue and dance. The tradeoff is that it does not invent new shots — it transfers the performance you give it, so the creative burden stays on the performer. That is a feature for anyone who has tried to get a generated clip to hit a specific beat.
- Controllable performance: the actor, not the prompt, drives the result.
- Cheaper than mocap: a phone camera replaces a stage with markers.
- Fits existing pipelines: output drops into Gen-4 for background and cleanup.
Where it earns its keep
- Animated shorts where a real actor’s nuance is the point.
- Pre-visualising a scene before committing to a full shoot.
- Driving a mascot character in a marketing spot.
- Game prototyping with fast placeholder dialogue animation.
- Music videos mixing live performance with illustrated characters.
Is Act-Two free, and its limits
No free tier — it lives inside Runway’s Standard and higher plans, with credits consumed per generation. It also needs decent lighting and a stable camera, since poor reference footage transfers its problems along with the performance. For long-form or photoreal human output, a traditional pipeline still beats it; Act-Two shines on stylised characters and short, expressive clips.
Quick Summary
- Runway Act-Two transfers facial and body performance onto any target character.
- It is included with Runway plans from about $15/month; credits are consumed per use.
- It favours controllable, actor-driven animation over text-to-video generation.
- Needs well-lit reference footage to look right.
- Best for stylised characters and short expressive shots, not photoreal long-form.
Related Tools
Tavus CVI Paid
Real-time conversational video avatars you can embed in your own app
HeyGen Avatar IV Freemium
Photoreal AI avatars that lip-sync from a single photo and a script
Elai Freemium
Elai AI Avatar Video Generator for Training & Marketing
Colossyan Freemium
Colossyan AI Video Creator for Training & Corporate Content
Hour One Paid
Hour One: Enterprise AI Video Platform with Realistic Virtual Humans
D-ID Freemium
D-ID: AI Avatar Video Platform for Talking-Head Content at Scale
Related Tutorials