Avatar video image URLs: product_image, scene photo and backgrounds
Sume avatar video takes three kinds of image URL: product_image, scene.image_url and scene-background images. All must be public HTTPS. Checks to run first.
A Sume avatar video request can carry three kinds of image URL: product_image, scene.image_url for a photo scene, and image URLs in scene backgrounds inside video_inputs. The docs say that each must be a fetchable public HTTPS URL (read 2026-10-09). Check all three before you submit, because a URL that works in your browser behind a login will not work for a server fetching it.
What each field does
All three are optional. Leave out product_image for a productless avatar video. Use scene with a prompt to direct a scene in words or with a photo to supply a reference image.
| Field | Purpose | Required? | URL rule |
|---|---|---|---|
| product_image | Product shown with the avatar | No | Public HTTPS, fetchable |
| scene.image_url (type photo) | Photo scene reference | No | Public HTTPS, fetchable |
| video_inputs[].background image | Per-scene background | No | Public HTTPS, fetchable |
What Sume rejects before submit
For avatar creation, the docs say Sume rejects localhost, private-network URLs, non-HTTPS URLs and non-image responses before submitting the generation. The avatar-video docs say media fields must be fetchable public HTTPS URLs and point to the same media-input rules. In practice, that means these will fail: a link on a corporate VPN, a signed URL that has expired, an http:// link, a link that returns an HTML login page.
The error documentation lists image_not_fetchable and input_media_unreachable for media Sume cannot fetch or mirror safely, with the advice to make sure the input is a public HTTPS image URL and retry, or contact support with the request id. Do not include signed URLs or private ids in a support message.
A short pre-flight
Open each URL in a private browser window with no cookies. If you see the image, a server probably can. Check that the link returns an image content type, not a page. Make sure the file will exist for the whole job; a link that expires in ten minutes can break a queued job that starts later, since jobs may wait as queued when your workspace is busy.
If the picture matters a lot, run an avatar video preview first. It uses the same inputs and returns first-frame stills, so a bad product image shows up before the full render. The preview keeps your quality choice open: you can change the final tier at generate-video without a new preview.
- No localhost, private networks or http:// links.
- Return an image content type, not a page.
- Keep the link alive until the job finishes.
Where the images should live
If your product photos sit in a private bucket, create a public copy for the render, or publish through your own CDN with a stable HTTPS URL. Sume's media inputs doc covers asset-library handling. Avoid URLs with short-lived signatures, since a queued job may fetch the image minutes after you submit it.
One practical tip: use the same product image URL in the preview and in the final render. A preview that approved one image and a render that fetched another is a surprise you can avoid.
Sources
Related posts
More in Developers
- File-size cap and max length: the average bitrate ceiling
Divide a platform's size cap by its longest allowed video to get an average bitrate ceiling: LinkedIn about 2.2 Mbps, TikTok 6.7, Pinterest 17.8. Python check.
- Before a 20-clip MCP burst: dry_run, admission preview, max_spend_usd
A single Sume MCP create needs none of these. A 20-job burst should use dry_run or generation_admission_preview, set max_spend_usd, and wait in one jobs_wait.
- Duration dropdown from supported_durations: 27, 29 and 8 choices
Seedance 2.5 offers 27 lengths, Wan 3.0 29, Omni Flash 1.1 8. Map supported_durations to options in TypeScript, with the price of the longest option.
- Timeline body from a cut list in TypeScript: starts and coverage
Turn a cut list of source ranges into Timeline slots: starts are running sums rounded to 3 decimals, and 3 ranges of 3.55, 4.68 and 3.36 s make 11.59 s.
Written by Sume