Why might the AI change a label or collar?

The AI infers parts of the garment that are hidden in the source photo. It may add a neck label, leave one out or redraw the inside collar. Before publishing, check each generated image against the actual garment and your source photos.

Consider a jacket photographed on a mannequin. You may be able to see the outside edge of its collar while the mannequin covers the inside neckline. Removing the mannequin leaves an area for which the front photo contains no detail. What appears in that area of the generated image is an inference about the missing part.

Compare both the label and its position with the garment. A label can go missing, or lettering can appear where the real lining is plain. Follow the collar edge from one shoulder to the other and look at where the lining meets the outer fabric. Check each garment separately when reviewing a batch.

If you still have the jacket, take a second photo looking directly into the neck opening. Add it as the inside-collar reference in Single item processing to give the AI more information about the lining and reduce guesswork. The reference costs no extra credits. Compare the new result with both photos, especially around any lettering, before publishing the listing.