Use larger, soft-edged selections to provide context for better blending

Generative Fill always starts with a selection of the area that you want to change. Create a broader, soft-edged selection that includes more image detail around the part you want to modify. This will provide more context and result in a better blend of the generated detail with the original image.

Use clear, descriptive prompts

Once you have selected the area you want to change, enter a prompt that clearly describes how you want to change an object, or even the entire image.

Add to a previous generation with a new prompt

To modify a previously generated layer, create a new selection of the area you want to change and enter a new prompt.

Replace an entire background and maintain scene coherence

To replace a background, or change the location of a scene, select the entire image, and enter a prompt with a detailed description of the new background or scene, including details such as time of day and the lighting. The original subject will be preserved and blended seamlessly with the new background scene.

Combine Generative Fill results with traditional Photoshop tools

The best thing about using Generative Fill partner models in Photoshop is that you can take advantage of tools like Remove, Find Distractions, and more, for additional editing flexibility.

Transform poses and camera angles while preserving subject identity

To change the pose of a person, or show the scene from a different camera angle, select the entire image, and provide a descriptive prompt. When changing the camera angle, tell the AI to keep the same subject pose.

Use annotations on an image to tell the AI what you want

You can write directly on an image to add annotations to guide Nano Banana generative edits for additional options or visual workflows. Use an easily described color for the annotations, such as red, and tell Nano Banana to remove the red text when it is done.