Quality mode reviews every result against a fidelity bar — colour, print, construction, drape — and re-generates anything that misses, once, free. You spend a credit and see only the better attempt.
Generative image tools do not fail constantly. They fail occasionally, and that is worse, because it turns every result into something you have to check. A seller who has been burned once by a colour that drifted does not trust the next ninety-nine images either, and the time saved by generating them disappears into inspecting them.
The obvious fix — let the user retry — sends the bill for our miss to the customer. Charging a credit to correct a below-par render is charging twice for one image, and it is the moment a seller decides the pricing is not honest.
Quality mode moves the check to before you see the image, and absorbs the cost of the second attempt.
You pick the model, the garment and the scene exactly as in standard mode. Nothing about the setup changes and the credit cost is identical.
The result is scored against your own garment photo on the things that matter commercially — colour, print, construction and drape — not on whether the picture is attractive.
If the result misses the bar, a second generation runs steered at the specific defect that was measured, and the better of the two attempts is what you get. The retry costs you nothing.
A second provider call costs us real money. Absorbing it is the point: a below-par render is our miss, and billing you for the fix would be charging twice for one image.
The retry is not a blind re-roll. The reviewer says which check failed, and that finding is fed into the second attempt, so it aims at the measured defect rather than hoping for a better dice roll.
If you mark a finished result with a thumbs-down and the reviewer agrees with the complaint, the fix is generated free within the same daily allowance. You are not paying to report a problem that turned out to be real.
Quality mode costs the same credits as standard. It is a mode, not a premium SKU, so there is no arithmetic to do before choosing it.
Three reviewed retries a day on Pro and five on Business and above. Publishing the number is deliberate — an unbounded promise would either be untrue or priced into everyone's plan.
Standard generates once, and it is the right default for volume work where you are reviewing the batch anyway. Quality is for the shots that have to be right first time.
| Standard | Quality | |
|---|---|---|
| Plans | Every plan | Pro and above |
| Generations per request | One | Up to two, second one free |
| Credit cost | 1 at HD, 2 at 2K, 4 at 4K | Identical |
| Automated review | Result is still scored, and flagged in a bulk run | Scored and re-rolled on a clear miss |
| Judge-confirmed fix after the fact | Costs a credit | Free within the daily allowance |
| Daily allowance | — | 3 a day on Pro, 5 on Business and above |
| Best for | Volume runs you are reviewing as a batch | Hero shots and anything going straight to a listing |
Every result is scored on every plan — see batch review for how that surfaces in a bulk run.
Pro and above. On Starter and the free plan the setting is simply treated as standard rather than shown as a dead end, and every result is still scored — what the lower plans do not get is the automatic free re-roll.
No. It is the same credit price as standard: one at HD, two at 2K, four at 4K. The second generation triggered by a failed review is absorbed rather than billed.
Three reviewed retries a day on Pro, and five a day on Business and above. The allowance covers both the automatic re-rolls and the judge-confirmed fixes described below, and it resets daily.
The generated image against your own garment photo, on colour, print, construction and drape. It is a fidelity check rather than an aesthetic one: the question is whether the buyer would receive what the image shows.
The complaint is checked against the original garment. If the reviewer agrees with you, the fix is generated free within the same daily allowance. If it disagrees, you can still regenerate normally at the usual credit cost.
No. The reviewer reports which check failed, and that finding steers the second attempt at the specific defect it measured. A blind re-roll would be a lottery; this is aimed.
No — you see the better one. The purpose is to spend your attention on choosing garments and scenes, not on adjudicating between two renders of the same shot.
Usually not. Standard is the right default for volume, because a bulk run is already scored and the misses are flagged for you to retry as a set. Quality mode earns its place on hero shots and on anything going straight into a listing without review.
Five free credits on signup; quality mode unlocks on Pro.