Make an AI Clothes Try-On With ChatGPT (2026)

Want to see how an outfit looks on you before you buy it, borrow it, or commit to it? Instead of hunting for the “best” try-on app, you can build your own in a few minutes — one that fits exactly how you shop and dress.

Why build your own instead of downloading one

There are plenty of ways to try on clothes virtually with GPT-Image-1, and you’ll find generic try-on apps online. Most are one-size-fits-all: they lock you into their layout, their prompts, and their idea of what “trying on” means.

Building your own in Auto means it works your way. Want a catalogue-style background? A carousel of past looks? A quick edit button to tweak a fit? You decide. Auto is camera-first: you describe what you want, snap a photo, and it builds you a personal mini-app — a Frame — around it. People have already built ones like Virtual Try-On this way.

Build it in Auto: the quick walkthrough

1. Open Auto and describe your goal in plain words: “Create a try-on app. The first photo is me. Every other photo is an outfit to try on it.”
2. Add the core behavior: display the outfit with a Try On button that swaps the clothing onto your photo — keeping your face and pose the same.
3. Snap a photo of yourself, then a photo of any outfit. Tap Try On and watch it generate the result.
4. Refine as you go. Adjust the layout, colors, and prompt wording until it feels like yours.

The trick to good results is the prompt behind the swap — describing the garment in detail, not just its color.

How someone actually built theirs

A real Auto user built their try-on Frame prompt by prompt. Here’s the honest arc:

They started simple: set the first photo as “me,” treat every later photo as clothing, and add a button to apply the swap while keeping the original face and pose intact.

Early results were off — the generated image didn’t show the right garment. They diagnosed it out loud, asking whether both the pose photo and the clothing photo were being sent to the AI. That led to the key fix: make the prompt analyze the outfit first. Not “green jacket” but “olive green jacket, wide peak lapels.” Not “red dress” but “red halter-neck, ankle-length, fitted bodice gathered at the waist.” Those extracted details — color, fit, named style elements, embellishments — get sent alongside the photo so the output actually matches. This is the heart of good ChatGPT fashion prompts for try-on.

From there it was all personalization: a plain catalogue-style background, a scrollable carousel of past try-ons, a “Set as me” button, and an edit sheet to type tweaks and re-run the generation. Then the polish — orange accent color, staggered entrance animations, a two-column photo grid, and a processing bar that sweeps a sheen across the thumbnail while it works. They even asked Auto for an expert iOS design pass to make it feel native.

The point: they didn’t settle for a stock app. They shaped it, screen by screen, until it fit.

FAQ

Do I need to know how to code? No. You describe what you want in plain language and Auto builds it.

How does it use ChatGPT to put clothing on a model? Your Frame sends your photo plus a detailed description of the garment to an AI image model, which renders the outfit onto you while preserving your face and pose.

Can I edit a result I don’t like? Yes — add an edit button that lets you type a change and regenerate from your source photo, the created image, and your new prompt.

How long does it take to build? A basic version takes minutes. Refining the look and prompt is optional and ongoing.