New outfit, same person

Outfit Swap takes two images: a photo of a person and an image of an outfit. It dresses the person in that outfit, with the shoes, bags, hats, and jewelry the outfit image shows, and keeps everything else from the original photo: the face, the hair, the body, the pose, the light, and the background. The result keeps the shape of the photo.

Because the outfit comes from a picture, you never have to describe it. The model reads the cut, color, fabric, and pattern from the reference and fits them to the pose of the person in the base photo. That makes it better than a written prompt for one specific piece: a dress from a product page, a jacket from a street photo, or a costume you designed earlier.

It works like a virtual try-on: see how clothes look on a person before you buy them, shoot them, or draw them. To change clothes from a description instead of a picture, try the AI clothes changer.

The base photo decides the pose, the framing, and the light. Pick one where the clothes you want to replace are fully visible: a full-body shot for a full outfit, a waist-up shot for a top or a jacket. Arms crossed over the body, a bag held in front, or another person overlapping the subject give the model less to work with.

Simple poses swap most cleanly: standing, walking, or sitting upright. Clear, even light helps too, since the new clothes have to take on the light of the photo to look like they belong in it.

• Full body for a full outfit, waist up for a top.
• Hands and arms away from the clothes you want to replace.
• One person in the frame.

The outfit image is the reference for what the person will wear. Show the whole look in one image: top, bottom, shoes, and accessories. A product photo on a plain background, a flat lay, or a photo of someone else wearing the outfit all work. The clearer the details, the more faithfully the pattern, the logo, and the texture come through.

To change a single item, use an image that shows only that item. An image of a full outfit tends to change the full look.

Saved characters work in both slots. Put a character in Base Image to dress them in a new look while their face stays the same, or use one as the outfit reference to borrow their costume. To make a new character first, use the Character Builder: Attributes.

For a lookbook, keep one base photo and run a swap for each outfit. Every result lands in History, so you can compare the looks side by side and keep the ones that work. Running the same swap on another model from the Model menu gives you a second take on the fit.

To recast a person in a video, costume included, Video Character Swap replaces the whole character in every frame. To replace the whole person in a photo rather than their clothes, use Character Swap. To design a new outfit from a description, try the AI outfit generator.

Use photos of people who agreed to it, and outfits you have the rights to show. Your uploads and results stay private to your account unless you post them.

Clothing shops can show one garment on several models without a new shoot, or a new collection on the models they already work with, with their permission. Use product photos on a plain background as the outfit images, so the cut and the fabric come through clearly.

Stylists can show a client several looks on the client's own photo and decide together. Creators can keep one look across a series of posts, or try a new style on themselves before they buy anything.

For a character, keep a wardrobe: run swaps with an outfit for each part of a story, such as a uniform for work and a coat for winter, and keep the results you will use again.