Back to MiniMax H3 prompts
Multi-video reference / Reference Generation
Official manual translationTranslated from a MiniMax H3 manual exampleHitchcock move and song transfer
Give camera, identity, and performance to separate videos instead of blending their roles.
Actual H3 output / 7s
Prompt -> result
What this output demonstrates
The output combines a dramatic reference camera move with a new singer and transferred vocal performance.
- Control pattern
- Give camera, identity, and performance to separate videos instead of blending their roles.
- Check in your run
- Unclear ownership can replace the performer or copy the wrong scene.
Shot structure
0s4s7s
INPUTR2V
ACTION01
CAMERA01
AUDIOON
Working principle
Give camera, identity, and performance to separate videos instead of blending their roles.
Watch for
Unclear ownership can replace the performer or copy the wrong scene.
Adapt this brief
Prompt draft
Use Video 1 only for the Hitchcock-style camera movement. Make the performer from Video 2 sing, while Video 3 defines the vocal performance and singing behavior. Keep the subject identity from Video 2 and align the facial performance to the referenced voice and camera timing.Replace the subject and reference roles first. Change camera, timing, and sound one variable at a time.