Crowdfunding page video: Wan 3.0 hero plus Kling 3 loops, priced

For a crowdfunding page, use Wan 3.0 for the hero clip and Kling 3 for the short feature loops. The hero needs your prototype photos as references, and only Wan 3.0 takes them; Kling 3 on Sume accepts a first frame and an end frame but no reference URLs. A 15-second 720p Wan hero is $1.88, three 6-second 1080p silent Kling loops are $2.52, and the set costs $4.40.

Why the split

Backers judge a campaign by whether the product looks real. A hero built from three or four photos of the actual prototype keeps the product shape stable, and that is reference-to-video. Wan 3.0 lists reference-to-video on Sume with up to 10 reference images, 5 reference videos and 5 reference audio files, and a 2 to 30 second duration range. Alibaba's launch post, read 2026-10-08, describes fine-grained reference consistency for characters, props, spaces and style as a headline feature.

Feature loops are different: you start from one still of the finished render, animate it, and want cheap 1080p. Kling 3 charges the same for 720p and 1080p, so you take 1080p.

The numbers on Sume

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

A short build order

Draft before you pay for the final.

• Render the hero at 480p first: 15 s is $0.94. Check that the product silhouette holds, then re-run at 720p.
• Pin the resolution. Wan 3.0 defaults to 1080p when the request omits it, which is the highest of its three tiers.
• Keep the feature loops silent, then add one music bed under the whole page video on a timeline instead of paying per clip for sound.
• Keep a separate 9:16 job for social; do not crop the 16:9 hero.

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. Draft the hero at 480p, then change only the resolution for the final. 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": "wan-3.0",
    "prompt": "A slow push-in on the scene, soft natural light",
    "duration": 15,
    "resolution": "480p",
    "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"))

Limits to know

Wan 3.0 reference video must be 15 seconds or less in total and at least 16 fps, per the catalog constraints. Alibaba's own post says audio texture and on-screen text accuracy are "improving but not yet where we want them", so put pledge tiers and prices in the page layout, not in the generated video. Kling 3 lists 4 to 15 seconds, so a 3-second loop is not possible; use 4.