← Back to the presentation

The actual inputs behind the demo

Bruce’s two identity photos and the product reference were used for Seedance A and MiniMax H3 A. Seedance B uses A’s actual final frame. These are the submitted generation prompts, distinct from the beginner three-stage exercise.

Bruce’s photos illustrate this run; use your own authorized identity or a virtual character. Official Apple source.

Seedance A

{
  "model": "seedance-2-0-fast",
  "videoTaskType": "reference",
  "duration": 15,
  "resolution": "480p",
  "ratio": "16:9",
  "generateAudio": true
}
Download .txt

Seedance B

{
  "model": "seedance-2-0-fast",
  "videoTaskType": "frames",
  "duration": 15,
  "resolution": "480p",
  "ratio": "adaptive",
  "generateAudio": true
}
Download .txt

MiniMax H3 A

{
  "model": "MiniMax-H3",
  "duration": 15,
  "resolution": "768P",
  "ratio": "16:9"
}
Download .txt

H3 B was not generated: its actual bridge failed face consistency verification. No second-stage H3 prompt is presented as an executed result.