Why did a try-on result look wrong?

Photos are the usual culprit

A try-on result reflects two inputs: the shopper's photo and your product imagery. Weak input on either side shows up in the output. On the shopper side, poor lighting, unusual angles, or partial framing gives the model less to work with. On the product side, clear, product-focused photos generate best — and for apparel, full views of the garment beat cropped or heavily styled shots. If one product misses repeatedly while others render well, inspect its photography first.

The model is inferring, not scanning

Looksy generates a photo try-on in about 20 seconds from one uploaded photo. Any model working from a single image has to estimate things the photo never shows directly — body position under clothing, how a fabric drapes, how a frame sits on a face. Sometimes an estimate lands wrong: a hem at an odd length, a pattern that warps, a fit that reads too loose or too tight. That is a limit of single-image generation in general, not a sign the tool is broken.

What the shopper can do right away

A bad result is not a dead end. Shoppers can retry a result straight from the product page, and a second attempt with a clearer photo — evenly lit, showing the relevant part of the body — gives the model more to work with. They can also leave feedback when a result misses. Because the whole flow runs in the browser with no app download and no account, a retry costs the shopper about 20 seconds, not a re-signup.

How merchants can cut the miss rate

Treat wrong results as data. Looksy's analytics show what shoppers try on, add to cart, and buy — a product with plenty of try-ons but few adds to cart is worth testing yourself. Shoppers can also leave feedback when a result misses; check in the app how that feedback is surfaced to you rather than assuming a particular report exists. Reshoot repeat offenders with clean, full, product-focused images before concluding the tool is at fault. If a product still misses with good photos, contact Looksy support with concrete examples — a before-and-after is faster to diagnose than a description.

Sources: the public Shopify App Store listing for Looksy. Last checked 27 July 2026.

Related questions: How is AI try-on different from social media filters? · Can shoppers download and share try-on results? · What photo should shoppers upload for try-on?