---
name: animate-multi-image
description: "Animate your AI character from separate images — character sheet, location and prop as Image 1, 2, 3 in MiniMax H3 Max reference-to-video with a two-sentence prompt. All three land; 5 s is too short for walk + stop + gesture; images under 256×256 are refused, pad them to 400×400. Use when the user has a character, a place and an object but no film frames, \"put my character in this location with this prop\". 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-multi-image · character + location + prop, one image each

> 🏆 **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.

The person's {character sheet} (Image 1), a {location} (Image 2), a {prop} (Image 3) → one shot. Tested with seed 20260929.
For an unusual prop that must keep its shape → `animate-props`.

## Choose

| Priority | Settings | Time | Result |
| --- | --- | --- | --- |
| ⚡ tested | `minimax/h3-max/reference-to-video` · 5 s · **480P** · 16:9 · 3 images · enhance **disabled** | 9 s submit to file | all three references land; action cut short |
| paced by the model | same, enhance **balanced** | 14 s | whole action fits 5 s — but it invented a cut at 3.0 s (see `animate-prompt-enhance`) |
| 🏆 768P / longer | same at 768P or a longer duration | not tested with this prompt | — |



## Steps

1. **Check every image is at least 256 × 256.** A 278 × 212 prop photo was refused at submit (`image_too_small`); centred on a **400 × 400 white canvas — no scaling** — it passed.
2. Body:
   ```json
   {"duration": 5, "resolution": "480P", "aspect_ratio": "16:9", "prompt_expansion_mode": "disabled", "seed": 20260929,
    "reference_image_urls": ["<{character sheet}>", "<{location}>", "<{prop}>"]}
   ```
3. Prompt — one sentence of roles, one of action; blind: references named by role, no looks in words:
   ```
   Image 1 is {character}. Image 2 is the location. Image 3 is {their prop}.
   {character} {moves} {through the place} of Image 2, {wearing / holding} the {prop} from Image 3 and their clothes from Image 1. {beat 2}, {beat 3}. {camera move}, {one detail of sound or motion}.
   ```
4. Generate through your supported external tool with the body and prompt above.

## What to expect

- ✓ the location with its light; the prop on the character; the outfit from the sheet. The prop may come out with a little more colour than its photo.
- ✓ the camera does what the words say (slow push-in to a close-up in the test); the location's own framing is not used.
- ✕ **5 s is too short for walk + stop + gesture**: at the last frame the gesture was only starting. Give fewer beats or more seconds (the length that fits was not measured). With enhance balanced / quality the rewrite timestamped each beat and the whole action fitted 5 s.

## Never

- An image under 256 × 256 — refused; pad, do not upscale.
- Describing an unusual prop in words to fix its shape — see `animate-props`.

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

The demo character Mara Vale's sheet (Image 1), a flooded tiled corridor (Image 2), round wire glasses (Image 3, the 278 × 212 photo padded to 400 × 400). Action: walks toward the camera down the corridor wearing the glasses, stops, takes them off, looks into the lens; slow push-in, water splashing around her boots. Result: the corridor with water and warm side light, round wire glasses on her face, lenses tinted teal; at the last frame she was only starting to lift the glasses. The generalized prompt above assigns roles instead of reusing that character-specific prompt.