School open-day teaser: three 10 s MiniMax H3 Max takes cost $3.00

Order three 10-second takes of MiniMax H3 Max at 768p: $1.00 each, $3.00 for the set, on Sume. Use your own photos of the entrance, the library and the playground as reference images so the teaser shows your campus, not a generic one. 768p is H3's native size, so you are not paying for an upscale.

Why three takes

An open-day teaser is usually a gate shot, a classroom shot and a crowd shot. Separate takes let you pick the best of each and cut them to a music bed, instead of hoping one 30-second generation gets all three right. H3 and H3 Max on Sume run 5 to 15 seconds per job.

MiniMax's H3 post, read 2026-10-08, describes multimodal references (video, image and audio inputs) and multi-shot generation, with native stereo sound that does not separate voice, effects and music. On Sume the model takes up to 9 reference images, 3 reference videos and 3 reference audio files, 12 in total, and audio cannot be the only reference.

Costs by tier

Sume bills the provider list price times 1.25, rounded up to the cent per job.

A workable order of operations

Spend the cheap tier on decisions, the expensive tier on the take you keep.

• Draft each shot on minimax-h3 at 480p, $0.63 per 10 s, to check the framing.
• Re-run the keeper on H3 Max at 768p ($1.00) or 1080p ($2.00). H3 Max's 1080p is a refinement from 768p, per the catalog notes.
• Send up to 9 photos of the real building as image references. Past 5 images, plain H3 reference-to-video adds a per-image charge on Sume, so keep the set small.
• Overlay the school name, date and address as text afterwards rather than prompting for them.

Submit it and read the result

Sume reserves the billable amount from your workspace USD balance when the job is submitted, then returns a polling_url. Submit the three takes as three separate jobs, so one weak take can be re-run alone. Poll the job until status is completed or failed; a completed job lists its files in unsigned_urls.

import os
import time
import requests

H = {"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"}
body = {
    "model": "minimax-h3-max",
    "prompt": "A slow push-in on the scene, soft natural light",
    "duration": 10,
    "resolution": "768p",
    "aspect_ratio": "16:9",
}
r = requests.post("https://api.sume.com/v1/videos", headers=H, json=body, timeout=60)
r.raise_for_status()
url = r.json()["polling_url"]
while True:
    job = requests.get(url, headers=H, timeout=30).json()
    if job["status"] in ("completed", "failed"):
        break
    time.sleep(10)
print(job["status"], job.get("unsigned_urls"))

Care with children on screen

Do not upload photos of identifiable pupils as references without the written permission your school requires. Prompt for crowds seen from behind or at a distance, or use staff who have consented. I did not test how H3 Max renders large groups of people, so draft before you commit the budget.