Generative Fill Online

What you can do with generative fill

Generative fill lets you change one part of a photo without touching the rest. Brush over an area, describe what belongs there, and the AI paints it in. Use it to add a vase of flowers to an empty table, put a hat on someone, replace a blank wall with a window, or swap a plain mug for a branded one in a product mockup.

Behind the scenes, the tool uses FLUX.1 Fill [dev], an inpainting model that generates new pixels only inside the mask and blends them with the surrounding light and colors.

Writing a good prompt

• Describe only what goes in the masked area. "A small wooden bench" works better than describing the whole scene.
• Add useful details. Color, material and size help: "a tall green plant in a white ceramic pot".
• Match the mask to the object. Draw a mask roughly the size and shape of the thing you want. A tiny mask can't fit a large object.
• Try a few variations. Each fill is cheap, so run the same mask with slightly different wording and keep the best one.

Resolution and limits

The image is processed at about 1024 × 1024 pixels, so very large photos are scaled down. If you need a high-resolution result, send it to the Image Upscaler afterwards. Generated details can sometimes look off, especially hands, text and faces. Generating again with a clearer prompt or a better-fitting mask usually helps.

Results have no watermark, even when you use free credits, so you can use them in mockups, posts and presentations right away.

Generative fill vs. other tools

If you want to remove something without adding anything new, use the AI Object Remover instead. To add space around the edges of a photo, use the AI Image Extender. And if you want to change the whole image, such as its style, lighting or clothing, the AI Photo Editor works on the full picture from a single text instruction.