Generation involves several AI systems working in sequence, and sometimes a step fails. Imagefall is built so a failure costs you nothing and never blocks the rest of your refresh. Here’s what happens and what to do next.
Partial delivery, automatic refunds
A Listing Studio refresh is a series of independent steps: cutout, packshot, scenes, model shot, video, alt text, optionally 3D. If one step fails:
- Every other step still completes and is delivered to your review gallery.
- The failed step’s credits are refunded automatically you’ll see the refund noted on the step and reflected in your balance. You only ever pay for delivered work.
Retrying
Failed steps show a Retry button. Retrying re-runs just that step (charging its normal cost again, which the refund has already covered). Transient failures (provider timeouts, queue hiccups) usually succeed on the second attempt. If the same step fails twice, the cause is more likely the input; see below.
Common causes
- Low-resolution or blurry source photo. The studio needs enough pixels to work with. Swap in a sharper photo, even a good phone shot, and re-run the step.
- Product barely visible in the photo. If the product is tiny in the frame, cropped, or heavily obscured, cutout and reconstruction steps can fail. Use a photo where the product dominates the frame.
- 3D on unsuitable products. Clothing and soft goods often fail reconstruction; the beta is aimed at rigid products. See the 3D beta guide.
- Provider outages. Rarely, an upstream AI provider has an incident. These resolve on their own; retry after a few minutes.
Stuck rather than failed?
Steps can legitimately take a few minutes; video and 3D are the slowest. A refresh that’s been “running” for much longer than ten minutes is worth reporting. Applied videos and 3D models are also processed by Shopify after upload, so a short delay before they appear on the storefront is normal.
When to contact us
If a step fails repeatedly on a photo that looks fine, or a refund doesn’t appear in your balance, email support@imagefall.com with the product name and roughly when the run happened; we can see the run’s step log and will sort it out.