AI Background Changer

An AI background changer replaces the scene behind a subject while leaving the subject alone. The person, their pose, their clothes and the way they hold their hands stay as they were. The street, the studio wall or the cluttered living room behind them becomes whatever you describe: a beach at golden hour, a neon alley, a clean backdrop for a product.

Mage does it as an edit. You attach your photo to an edit model such as Mango 3, Nano Banana 2.1 or GPT Image 2.5 Flare and write what the new background should be. The model repaints the scene around your subject in one pass, so the edges, the shadow on the ground and the reflection in a window are generated together instead of pasted on afterwards. When you need tighter control, Inpaint lets you paint over exactly the area to replace.

Mage does not cut subjects out into transparent PNG files. Every result is a finished image with a full background, which is what you want for social posts, listings, thumbnails and profile photos. If a job needs a transparent cutout for a layout tool, use a dedicated background remover for that one step, and Mage for the scene itself.

The most common failure is a model that changes too much. It decides the new beach needs different clothes, or that the person should turn to face the sea. Prevent it by naming what stays before naming what changes. Start with the subject and the details you care about, the face, the hair, the outfit, the pose and the framing, and say they stay exactly as they are. Then describe the background.

Describe the new scene the way a photographer would: the place, the time of day, the light, and how far away things are. 'A sunlit café terrace in Lisbon, late afternoon, warm side light from the left, blurred tables behind her' gives the model a camera position and a light direction to match. 'Put her in Lisbon' leaves it to guess.

If the same subject has to survive many backgrounds, for a set of listing photos or a character who travels through a story, save them once as a character and mention them by @handle, so the face stays the same from one scene to the next. The AI character generator guide shows how the Characters system works.

• Name what stays first, then what changes.
• Give the light a direction and a time of day.
• Say how blurred the background should be: 'shallow depth of field' or 'everything in focus'.
• Keep the camera height: a waist-level photo wants a waist-level scene.
• For a plain backdrop, name the color and the material: 'seamless warm grey paper'.

A prompt edit is the fast path. The model sees the whole image, works out which pixels belong to the person, and rebuilds the rest. It handles hair, glass and sheer fabric well because it regenerates them together with the new scene. The trade-off is that it can touch the subject slightly, so check hands and faces against your original before you keep a result.

Inpaint is the precise path. You paint the mask over the background yourself and describe what should be there, and everything you did not paint stays untouched. It takes more care around fine hair, but nothing about the person can drift. The same tool removes a single distracting object, which the AI photo editor guide covers in more detail.

Many good results use both: a prompt edit for the new scene, then Inpaint for the one corner that came out wrong.

A swapped background gives itself away through light. If the new scene is a sunset and your subject is lit by flat office light from above, the eye notices before it can say why. Relight fixes that after the fact: choose the direction the new light comes from, soft or hard light, the brightness and a color, and the subject picks up light that agrees with the scene.

Shadows matter just as much. Ask for 'a soft contact shadow under his feet' when the subject stands on a floor, and for reflections when they stand near water or glass. Finish with Image Enhancer, which upscales the result to 2K, 4K or 6K so it holds up as a print or a large banner.