Runway Act-Two uses a driving performance and a character image or video. For a rap project, that makes the recorded performance a central part of the creative process: expressions, speech, and movement are prepared before generation rather than described entirely through text.
Workflow
- Record or select the driving performance for a short, complete excerpt. Keep the speaker and movement easy to see. Use a stable version of the soundtrack so subsequent edits do not change the timing you are asking the character to follow.
- Choose the character input according to the shot. Act-Two accepts a character image or video, but gesture control depends on the input type. Read the specific guidance for the route you select rather than assuming the two options provide identical controls.
- Generate the single-character performance and inspect the whole clip. Check facial stability, speech timing, and whether the movement fits the character. Keep notes on the input combination so a successful shot can be reproduced or revised deliberately.
- For multiple characters, prepare separate performances and assemble the scene in an editor following Runway's multi-character workflow. Align the clips to one timing reference and decide when to use close-ups, reactions, or a combined shot.
When this workflow fits
Choose Act-Two when you want to direct a performance through a recording and can prepare suitable source material. The single-character input is an important production constraint for duos; do not budget a two-person scene as though it were necessarily one generation.
Alternatives for your project
Viggle is relevant to motion-reference character animation. Hedra is relevant when the source is a portrait and audio. RapVideo.ai offers a more preset-oriented duo route when the available scene is already close to your intended result.
Inputs, output and limitations
Runway Act-Two
Input: A recorded performance and a character image or video.
Output: A character performance with transferred expressions, speech, and movement.
Main limitation: Act-Two supports one character per input. Multi-character scenes require separate generations and editing. Gesture control depends on using a character image.
Multiple people: Separate generation & editing
Related guides
Updated October 26, 2026. Editorial policy