Generative Fill in Adobe Photoshop

← Back to Blog

What is Generative Fill?

Generative Fill is an AI-powered feature in Adobe Photoshop, introduced in 2023 as part of the Adobe Firefly integration. It allows users to add, remove, or replace image content using a text prompt, with the AI generating new pixels that match the surrounding image in terms of lighting, perspective, style, and colour. The results are placed on a non-destructive generative layer, preserving the original image data.

How It Works

Generative Fill is powered by Adobe Firefly, Adobe's family of generative AI models trained exclusively on licensed, royalty-free content. This makes it safe for commercial use without the copyright concerns associated with some other AI image tools.

To use Generative Fill:

1. Make a selection in Photoshop using any selection tool (lasso, rectangular marquee, subject selection, etc.)
2. Click Generative Fill in the contextual task bar that appears
3. Enter a text prompt describing what you want to appear in the selection (or leave it blank to simply remove content)
4. Click Generate — Photoshop produces three variations to choose from
5. Select the best variation; you can regenerate as many times as needed

Key Use Cases

Removing Objects

Select any object in a photo and leave the prompt empty. Photoshop fills the area with contextually appropriate background content, effectively removing the object. This is significantly more powerful than Content-Aware Fill for complex backgrounds.

Extending Images

Generative Fill can extend the canvas beyond the original image boundaries. Select the empty canvas area and generate content to seamlessly extend the scene — useful for changing aspect ratios or creating wider compositions.

Adding Objects

Describe an object in a prompt and Photoshop places it in the selected area, matching the lighting and perspective of the existing image. A prompt of "a red umbrella" will insert a plausible umbrella that fits naturally into the scene.

Replacing Backgrounds

Use Select Subject to isolate a foreground subject, invert the selection, and generate a new background. The AI respects the lighting on the subject when creating the background scene.

Non-Destructive Workflow

All Generative Fill content is placed on a dedicated generative layer. The original image is never modified. You can hide, delete, or modify the generative layer at any time, and regenerate different variations independently. This makes it suitable for use in professional production workflows where change history matters.

Limitations

• Results can be inconsistent for very specific or detailed prompts
• Generating requires an internet connection (processing happens in Adobe's cloud)
• Complex structural elements (hands, text, specific faces) can produce imperfect results
• Very large selections may produce lower-quality results than smaller, focused generations

Generative Expand

A related feature, Generative Expand, works on the Crop tool. Extend the canvas in any direction and Generative Expand fills the new area with AI-generated content that matches the existing image. This makes it easy to reformat images for different aspect ratios without cropping.

Photoshop Automation and Plugins

For developers, Generative Fill is accessible through Photoshop's UXP plugin API, allowing automated generation workflows within custom plugins. Mapsoft builds custom Photoshop plugins and automation tools — see our Photoshop development page for more information.