Skip to content
IFAI

Generative media

Runway drops Act-Two for full-body mocap

Published Reach 52

Runway just introduced Act-Two, its next-gen motion capture model that tracks head, face, body, and hands using just a driving performance video and reference character. It brings major quality bumps, though it's currently limited to enterprise clients and creative partners.

Runway drops Act-Two for full-body mocap
Image: x.com

Runway just launched Act-Two, its next-generation motion capture AI model designed to animate digital characters without the expensive suits, tracking dots, or specialized studio setups.

Traditional motion capture has long been a massive budget sink across Hollywood, AAA gaming, and commercial production. Runway wants to compress that entire pipeline down to simple video inputs.

Full-body tracking with just a phone clip

Act-Two introduces substantial upgrades to generation quality and broadens its tracking scope to cover head, face, body, and hand movements simultaneously. Capturing clean hand gestures and subtle facial micro-expressions has historically been a major pain point for generative video, making full-body tracking a critical milestone.

The input workflow is remarkably lean. Creators only need to provide two assets: a single driving performance video of an actor, and a reference character image.

According to Runway, Act-Two can transfer that recorded performance across a wide variety of character designs, diverse visual styles, and complex digital environments, all while preserving the fidelity and nuances of the original acting.

There is a catch for solo hobbyists: Act-Two is currently rolling out exclusively to Runway Enterprise customers and official Creative Partners.

Why it matters

For commercial directors, indie game developers, and VFX boutiques, high-end motion capture has always been gatekept by six-figure hardware and grueling cleanup work. If Act-Two holds up under production scrutiny, turning an actor's webcam test into fully realized character animation could slash pre-visualization and production timelines from weeks to hours.

However, everyday creators will have to wait on the sidelines for now. Restricting access to enterprise tiers suggests that computing demands remain heavy, and edge cases around consistency might still require high-touch enterprise support.

Across the AI video landscape, the battle lines are shifting. Raw text-to-video novelty is wearing off, and control is the new frontier. With models like Moonvalley's Marey focusing on commercial safety, and video editing suites like Premiere Pro and DaVinci Resolve embedding native AI utilities, Runway is fighting to secure the high-end creative pipeline.

We think targeting markerless full-body and hand capture is a smart play. The film industry does not need more random hallucinations; it needs reliable digital puppetry. The lingering question is production tolerance: whether a single reference video can handle extreme angles and complex physical interactions cleanly enough to replace seasoned animators, or if it will primarily serve as a rapid prototyping engine.

Sources (2)

  1. [1]
    @runwayml(opens in a new tab)Originalx.com
  2. [2]

Similar events

Past happenings like this one, for context.

Best of AI, daily