Presigned URL expiry for image edit references on Sume

Reference images must be reachable by Sume's servers. The docs say the URL must be public HTTPS; localhost, private networks and non-HTTPS are rejected before submission. A private bucket with a presigned URL is the usual workaround, and it adds one risk: the link can expire before the job fetches it.

When the download fails, the failed job reports input_media_unreachable with the message "Could not download an input media URL (image_url). Verify the URL is publicly reachable, then retry." It is classed as validation, retryable: false and next_action: fix_input. A retry with the same dead URL will not help.

Choosing the expiry

POST /v1/images waits up to 30 seconds and falls back to a job for slow requests. With mode: "async" or webhook mode, the job can sit in a queue. The docs do not publish when the reference is fetched, so give the link more time than your longest expected wait.

Generating the link

With boto3, generate_presigned_url takes an ExpiresIn value in seconds:

import boto3

s3 = boto3.client("s3")
url = s3.generate_presigned_url(
    "get_object",
    Params={"Bucket": "my-private-bucket", "Key": "refs/product.jpg"},
    ExpiresIn=3600,
)
print(url)

Verify first

A presigned URL is a long https link, which is allowed. Fetch it once with a plain HTTP client from outside your network before you send it. If the link only works from your account or VPN, Sume will not reach it. Rotate to a fresh link per attempt rather than reusing one.

How this was checked

Vendor facts come from the pages listed in the sources, read on 2026-10-05. Sume facts come from the Image API docs and the catalog code on main on the same date. Catalogs and limits change, so read the descriptors from GET /v1/images/models before you pin a number in production code.