AI Clothes Changer

Upload a photo of a person and a photo of a garment. The AI swaps the outfit in the first photo for the garment in the second and keeps the face, pose and background. It is free, needs no sign-up, and gives each network two tries per day.

How the AI clothes changer works

Your browser shrinks both photos to at most 1024 pixels on the long side before anything is uploaded. That keeps uploads small and fast on mobile data.
One request goes to the image model. We send the two photos with a fixed instruction: keep the person, pose, lighting and background from photo 1, and dress them in the garment from photo 2. You cannot add your own text prompt, which keeps results predictable and blocks misuse.
The result comes straight back to you as a 1K image you can download. Nothing is saved on our side.

The model is Google's Nano Banana 2.1, released on October 6, 2026. Google says it improves visual quality, instruction following and character consistency over Nano Banana 2, which matters here: the garment's neckline, sleeves and print have to follow photo 2 while the face stays the person from photo 1.

Tips for better results

The person photo

Use a front-facing photo with even light. Strong shadows and backlight make fabric look wrong.
Match the framing to the garment. A dress or a coat needs a full-body photo; a T-shirt works from the waist up.
Keep arms away from the body if you can. Crossed arms, bags and hands in pockets hide the area the garment has to cover.
Only one person in the frame. With several people, the model may dress the wrong one.

The garment photo

A product photo or flat lay on a plain background works best.
Show the whole garment from the front. A cropped sleeve or hem leaves the model guessing.
Use a sharp image. Blurry prints and small text on the garment come out blurrier.

What works well and what doesn't

T-shirts, shirts, hoodies, sweaters, jackets, coats and most dresses give the most reliable results. Results are weaker for fine printed text, very busy patterns, sheer or reflective fabric, and items the person photo does not show, such as shoes in a waist-up photo. Expect small differences in colour and drape: the output is a preview of the look, not a guarantee of how the real garment fits.

Common problems and how to fix them

The face looks slightly different: Small or distant faces give the model less to work with. Crop the person photo so the face takes up more of the frame, or use a sharper photo. Avoid sunglasses and heavy filters.
The colour of the garment is off: Product photos under warm or coloured light carry that tint into the result. Use a garment photo taken in daylight or on a white background, and check that the person photo is not strongly tinted either.
The print or logo is simplified: Fine details survive best when the garment fills most of the garment photo. If a print matters, crop tightly around the garment before uploading.
The old outfit shows through: A long garment over short clothes usually works; a short garment over a long coat often leaves traces. Use a person photo in fitted, simple clothes for the cleanest swap.
The pose or hands look odd: Complex poses, crossed arms and hands near the torso are the hardest cases. A relaxed, straight standing pose gives the most natural result.

What people use it for

Checking an online purchase. Save the product photo from the shop and see how the style and colour sit on you before you order.
Choosing between two options. Run the same person photo with two jackets and compare them side by side.
Planning an outfit for an event, a trip or a photo shoot.
Previewing your own designs. Small shops and print-on-demand sellers can put a new T-shirt design on a model photo they own to check placement and colour. Label these images as AI previews when you share them.

Clothes changer or virtual fitting room?

Virtual fitting rooms in some online shops use your measurements and the brand's size data to estimate fit. This tool works from photos only. It is good at showing style, colour and how a cut changes your silhouette, and it cannot predict whether a size will be tight or loose. If fit matters, use the shop's size chart alongside the preview.

Why there is a daily limit

Each try sends a real request to a paid image model. At Google's list price, one 1K image costs about $0.034, plus a small amount for the input photos. To keep the tool free, each network gets two tries per day, and the total number of free images per day is capped. Limits reset at 00:00 UTC.

When daily capacity runs out, the form tells you and the button stays off until the next day. The service also pauses itself if the image provider returns errors that need a person to check, so you may occasionally see a paused message.

Privacy in short

There is no account and no tracking cookie. Your photos go from your browser to our server and on to Google's Gemini API, then they are discarded. To enforce the daily limit we keep a hashed identifier of your network that changes every day; the IP address itself is never stored. Full details are on the privacy page.

Rules

Do not upload photos of children, photos of anyone who has not agreed, or anything meant to sexualise, humiliate or impersonate someone. The tool is for previewing clothes, and it refuses revealing results. The full list is on the terms page.

Frequently asked questions

Is the AI clothes changer really free?

Yes. Every network gets two free tries per day, and there is no paid plan yet. Each image costs us real money to generate, so total free capacity per day is also limited. When it runs out, the tool says so instead of failing silently.

Do I need to sign up or log in?

No. There is no account. A short Cloudflare Turnstile check runs before each try to keep bots from using up the free capacity.

Are my photos stored?

No. Your two photos are resized in your browser, sent once to the image model, and discarded after the result comes back. We never save the photos or the result on our servers. See the privacy page for what Google logs on its side.

Why did a failed or blocked try still count?

The image service can charge for a request even when it refuses or returns nothing, so every attempt that reaches it counts toward the daily limit. Requests rejected before they reach the model, such as a missing photo, don't count.

Why was my result blocked?

The model has a safety filter, and our instructions ask it to refuse anything revealing. Swimwear, lingerie, very sheer fabric and photos that look like they show a child are common reasons. Try a different garment photo or a clearer, fully clothed person photo.

Can it tell me which size to buy?

No. It shows how the garment could look, not how it will fit. The model does not know your measurements or the garment size chart.

Which AI model does this use?

Google's Nano Banana 2.1 image model (gemini-nano-banana-2.1) through the Gemini API, at 1K resolution. The site is independent and not affiliated with Google.

Can I use the images for my shop or social media?

Only if you have the rights to both photos and the person shown agreed. Results carry Google's invisible SynthID watermark that identifies them as AI-generated. Do not present them as real product photos of an item someone actually wore.