---
name: animate-video-ref
description: "Borrow motion and camera from a video clip for your AI character with MiniMax H3 Max reference-to-video — character sheet as Image 1, a short clip with the sound removed as Video 1; the dance and the low still camera transfer, the costume and face of the clip do not. Use when the user says make my character move like this clip, dance like, copy the motion, motion reference, video reference. Part of the animate meta-skill."
---

These creative references describe source experiments with external tools. Use your own supported account and tools; this download installs no automation. Settings and timings are experimental observations, not current FlashBoards availability.

# animate-video-ref · motion and camera from a clip

> 🏆 **Source quality recommendation:** Seedance 2.5 reference-to-video (`bytedance/seedance-2.5/reference-to-video`, references as @Image1… / @Audio1…, 4–30 s, native sound) is the supplier's preferred quality route. That recommendation is supported by the product-commercial comparison, not a measurement of every character task. The H3 methods below retain their own measured inputs, settings and timings. Use supported external tools; this is not a FlashBoards integration claim.

{character sheet} as Image 1 + a short clip (sound removed) as Video 1 → {character} doing the clip's motion with the clip's camera.

## Choose

| Priority | Settings | Time | Result |
| --- | --- | --- | --- |
|  tested | `minimax/h3-max/reference-to-video` · 5 s · **480P** · `aspect_ratio: "21:9"` · sheet + 5 s clip | not measured | steps, arm swings, kicks, timing and the low still camera transfer |
| 🏆 768P | same at 768P | not run | — |
| ⚡ | not measured | — | — |



## Steps

1. Cut the clip to the moves you need (2–15 s; video references 15 s together at most). **Remove its sound** (`ffmpeg -i in.mp4 -an -c:v copy clip.mp4`) — we did; a clip with sound was not tested.
2. Upload your own images and clip through supported external tools. Body:
   ```json
   {"duration": 5, "resolution": "480P", "aspect_ratio": "21:9", "prompt_expansion_mode": "disabled",
    "reference_image_urls": ["<{character sheet}>"], "reference_video_urls": ["<clip>"]}
   ```
3. Prompt — one job per reference, blind (no looks in words), the action in words:
   ```
   Image 1 is only the look of {character}: exactly their face, hair and outfit from Image 1, nothing added. Image 1 never appears on screen.
   Video 1 is the motion and camera reference only: copy its {motion} - every {move}, {move}, {move}, the timing - and its camera, {the clip's camera: framing, angle, movement}.
   Live-action film scene, one continuous shot: {character}, exactly as in Image 1, {action} in {place}, exactly like the person in Video 1. Exactly one person is in the frame: {character}. Sound: {sound}.
   ```
4. Generate through your supported external tool; compare side by side with the clip.

## What transfers

- ✓ every step, the arm swings, the kicks, the timing; the camera — low, wide, looking up, holding still (in the tested clip).
- ✕ the performer's face, hair, make-up and costume — the character keeps their own outfit from Image 1.

## Never

- Expect the clip to dress or cast the character — the look comes only from Image 1.
- A long clip “just in case” — choose only the motion you need and keep within the reference duration limits.

## Example run (reference only — never reuse its files, prompts, faces, voices or frames for a person)

5 s of the stairs dance from *Joker* (2019), sound removed, + the demo character Mara Vale's sheet → Mara dancing down concrete steps between brick walls at dusk; she kept her bomber, turtleneck and ponytail from the sheet. The historical prompt still described Mara's looks in words; the blind wording above replaces that.