CartVisualCreate account

Why is Claid async result_url not the finished image, and when should I use output_object.tmp_url?

Use Claid’s result_url, when returned, to check the async image-edit task. If that URL is absent, save the task ID and use GET /v1/image/edit/async/{task_id}. When status reaches DONE, inspect result.output_object for the processed file location, including tmp_url where supplied. For ERROR, reconcile the recorded request and errors before considering another submission.

Published · By CartVisual editorial team

Recognize the two URLs and their different purposes

The initial asynchronous response identifies the task before the image is ready. Claid’s example includes result_url, which points to the status and result endpoint. The schema does not mark that field as required.

Save the task ID and supplied status. Call result_url if present. Otherwise, use the documented GET /v1/image/edit/async/{task_id} lookup with that task ID.

Once processing succeeds, the result object contains input_object and output_object metadata. The output_object can include tmp_url, which is the temporary address of the processed image. The two URLs therefore serve different actions: result_url is for checking processing; tmp_url is for retrieving the image.

A task identifier, returned metadata and downloaded file should remain linked to one source revision. This prevents an accepted request from being mistaken for a completed catalog asset.

Gate image retrieval on the returned status

The submission example begins with ACCEPTED. Claid's result-page prose describes ACCEPTED, PROCESSING, ERROR and DONE, while the underlying status enumeration also contains WAITING, CANCELLED and PAUSED. The supplied documentation does not explain every transition involving those additional values.

Treat ACCEPTED, WAITING and PROCESSING as states requiring further status reconciliation, not as evidence that an output exists. For DONE, inspect result.output_object and the returned image location before preparing the asset for review.

For ERROR, preserve the errors list and original request. The documented errors structure may include a message, timestamp and information about the affected input. A repeated submission is a separate decision; the source does not promise that retrying will resolve an error.

For CANCELLED or PAUSED, keep the actual value and seek the relevant current integration behavior rather than inventing automatic resume, permanent deletion or refund rules.

Observed API stateField to inspectNext action
ACCEPTED, WAITING or PROCESSINGid and status. Use result_url if present.Check the existing task again; no completed image is established
DONEresult.output_object and tmp_url if returnedRetrieve the file and validate its delivered properties
ERRORerrors and saved requestDiagnose the reported error before considering another submission
CANCELLED or PAUSEDExact status and available task recordHold the workflow; do not infer undocumented recovery behavior

Keep processing success separate from asset retention

When an output destination is not defined, Claid's documented temporary storage retains images for one day. A temporary retrieval address is therefore not a permanent asset location. Store a downloaded copy under the seller's own retention policy when the asset is needed later.

The result metadata can report format, MIME type, width and height. Those values help establish what was delivered, but they do not establish that the product image is accurate or approved for a marketplace.

A configured webhook can report success or failure as an alternative notification route, but it does not turn result_url into an image URL. Keep task reconciliation, file retrieval and editorial approval as distinct steps. No CartVisual webhook or Claid integration is implied.

Hypothetical decision: accepted task versus completed product cutout

A merchant prepares an async edit for one approved product photograph and assigns it internal asset ID BOTTLE-27. The illustrative submission response contains task ID 842, status ACCEPTED and a result_url pointing to the task's API status endpoint.

The merchant does not mark the product image available or submit a duplicate request. Instead, the task record remains open. A later hypothetical GET response reports DONE and contains result.output_object with a temporary image URL and output dimensions.

That transition changes the next action: the merchant can now retrieve the processed file, compare its format and dimensions with the requested contract, and begin the separate product review.

If the response instead reports ERROR, record its errors and keep the image unavailable. Save the task ID in either case. Use the documented task-ID lookup when result_url is missing.

Source and evidence scope

Prepared by the CartVisual editorial team with AI assistance. Competitor behavior is documented rather than independently executed for this guide.

Claid: Async API Reference

Official source snapshot checked .

The supplied Claid Async API Reference distinguishes task lookup, result metadata and temporary image retrieval. Its status schema includes values not fully explained by the prose. The state-to-handoff procedure is editorial; no API request, webhook, file retrieval or CartVisual integration was executed.

Related decisions

Browse ecommerce creative workflows