Generative Fill

What is Generative Fill?

Generative Fill lets you select part of an image in Photoshop, type what you want to appear there, and Adobe's AI fills it in convincingly: matching the surrounding colours, lighting, and style.

At a glance

Type of model
AI image synthesis and inpainting tool
Developed by
Adobe, powered by Adobe Firefly
Key capability
Seamlessly adds, removes, replaces, or extends image content based on text prompts within a selection
How it fits in AI workflow
Used in Photoshop to refine, composite, or expand AI-generated and live images, enabling non-destructive editing that blends generative content with existing photography or artwork

Ready to create?

Direct scenes, design characters, and ship full films

All-in-one AI creative platform with simple, transparent pricing, no speed throttles, and an infinite Canvas for max creativity.

Try Morphic
How it compares

How it compares

Generative Filltraditional content-aware fill

traditional content-aware fill in Photoshop samples pixels from the surrounding image to patch a selected area, making it effective for removing small objects against consistent backgrounds but prone to repetition artefacts on complex scenes. Generative Fill uses a large generative model to synthesise entirely new content, which allows it to add objects, change environments, and fill large or complex selections with far greater visual plausibility.

Pro tip

When using Generative Fill for object removal, make your selection slightly larger than the object you want to remove, extending into the surrounding background. This gives the model more contextual information about what the filled area should look like, and typically produces cleaner results with fewer visible boundaries or blending artefacts at the selection edge.

Types and variations

  • Generative Fill encompasses several related capabilities within the Adobe ecosystem.
  • Standard Generative Fill applies the AI synthesis within a user-defined selection area inside the image.
  • Generative Expand extends the canvas beyond the original image edges, hallucinating plausible content to fill the new space.
  • Remove Tool is a simplified variant that focuses specifically on eliminating unwanted objects from an image and filling the gap with background content.
  • Adobe also offers generative features in Adobe Express and Adobe Firefly's web interface, where similar fill and replace capabilities are available outside of Photoshop for users on different subscription tiers.

Ready to make your first scene in Morphic?

Try Morphic

Common use cases

  • Photographers use Generative Fill to remove unwanted people, vehicles, or objects from scenes without visible artefacts.
  • Advertising teams use it to extend backgrounds for different aspect ratios or to add branded elements to existing images.
  • Concept artists use it to rapidly explore compositional changes: swapping skies, adding architectural elements, or testing different costume options on figures.
  • AI content creators use it to refine and polish outputs from image generation models, fixing anatomical errors, removing watermarks, or seamlessly incorporating AI-generated subjects into photographic environments.

Ready to create?

Direct scenes, design characters, and ship full films

All-in-one AI creative platform with simple, transparent pricing, no speed throttles, and an infinite Canvas for max creativity.

Try Morphic

FAQs

What is Generative Fill?
Generative Fill is an AI-powered editing feature in Adobe Photoshop that allows users to add, remove, replace, or extend image content using text prompts. It is powered by Adobe Firefly and synthesises new pixels that blend seamlessly with the surrounding image based on the description provided.
How does Generative Fill work?
The user selects a region of the image and optionally types a prompt describing what should appear there. Adobe's Firefly model then generates multiple variations of plausible content for that region, matching the existing image's lighting, colour, and perspective. The results are placed on non-destructive generative layers so the original image data is preserved.
Is Generative Fill the same as content-aware fill?
They are related but distinct. Content-aware fill samples pixels from the surrounding image to fill a selection, making it effective for simple removals against consistent backgrounds. Generative Fill uses a large AI model to synthesise entirely new content, which allows it to add new objects, change environments, and handle complex or large selections more convincingly.
Can Generative Fill extend an image beyond its original borders?
Adobe's Generative Expand feature, closely related to Generative Fill, can extend the canvas beyond the original image boundaries. By selecting the empty area outside the image and either leaving the prompt blank or entering a description, the model generates contextually appropriate content to fill the expanded space.
What model powers Generative Fill?
Generative Fill is powered by Adobe Firefly, Adobe's proprietary family of generative AI models. Firefly was trained on Adobe Stock imagery and other licensed and public domain content, which Adobe states makes it commercially safer to use than models trained on unverified web-scraped data.
Does Generative Fill work on video?
As of mid-2025, Generative Fill is primarily designed for still images within Photoshop. Adobe has been extending generative capabilities into Premiere Pro for video editing, but frame-by-frame video generative fill operates differently from the still image version and is subject to separate feature availability and subscription tiers.
Can I use Generative Fill without writing a prompt?
Using Generative Fill with an empty prompt instructs the model to infer what should fill the selection based entirely on the surrounding image context. This works well for removing objects and filling with background content, but providing a prompt gives more control over what the model generates when adding or replacing elements.
Is Generative Fill output copyright-safe for commercial use?
Adobe states that content generated using Firefly, including through Generative Fill, is designed for commercial use and was trained on licensed data. However, copyright in AI-generated content is an evolving legal area, and creators working in commercial contexts should stay informed about current guidance in their jurisdiction.
Can't find what you are looking for?
Contact us and let us know.
bg