Actual generation · October 12, 2026
One photo. A new lead.
A completed studio test using the exact photo below and the original performance reference. The person in the input image is fictional.
- Video request
- 14 seconds · native 768P
- Actual output file
- 14.375 seconds · 1344 × 768
- Model / attempts
- MiniMax H3 · one generation
One-time price per generation. Prepare a cast for free, then pay before creating your video. View current pricing.
The original
Reference performance · 14 seconds · 1024 × 576
View original sourceA new main character
Actual H3 output · 14.375 seconds · native 1344 × 768
Choose a music style when creating your own video.
Download this exampleThe generated preview starts with the model’s own audio. “Original music” plays a separately prepared demo with the reference audio; new orders let you choose original or AI-generated music. Both demo versions keep the same native video frames.
This test used a party image with other people in the background. For your own lead, a clear single-person portrait is a better starting point.
What this result shows
- The denim-clad female lead replaces the foreground performer.
- The dark dining room, chandelier, visible guests and silver cup remain recognizable.
- The result follows the broad camera progression and gesture sequence.
What to check for yourself
- Finger poses and lip movements differ from the reference.
- This is one solo-lead test; it does not verify added friend identities.
- Likeness, clothing and movement can vary in another generation.
How we made this test · usage and estimated cost
- Supplied the 14-second performance reference and the exact lead image shown above.
- Requested one 14-second native 768P generation, replacing the foreground lead while guiding the scene, camera and action with the reference.
- Downloaded the full native output. Its 14.375-second runtime includes provider padding; the video was not resized or re-encoded.
- Prepared the original-audio comparison separately, without another generation.
The provider reported 14 input video seconds, 14 output seconds and one image. At the published 768P rates of CNY 0.50 per input/output second, the estimated API cost is CNY 14. The image is within the free image allowance. MiniMax pricing, checked October 12, 2026.
This is a provider-cost estimate, not a verified account invoice or a customer checkout receipt. Your retail purchase is $5.99 for one generation; this case was a direct studio test.
Follow the same photo-to-video workflow
- Add your lead photo
- Choose music and review the price
- Pay, then create
- Watch, download and share
