The represents the absolute best generative image editing available to the public right now. It beats Midjourney simply because of seam integration —you don't port images in and out; the AI lives inside your mask panel.
Previously, Photoshop's AI tools required an existing base image. The new feature changes that by bringing text-to-image creation directly into the app.
Adobe Photoshop Beta updates (version 27.x/2026 release) have
This feature automatically conforms text layers to circular, arched, or bowed paths with one click. adobe photoshop beta ai upd
The is a killer for ghost mannequins. If you shoot a t-shirt on a white background, you can now select the empty neck hole and use Generative Fill with the prompt "continue fabric." The new beta respects the knit weave direction. It saves 15 minutes of Clone Stamp work per image.
These refined selections are directly accessible via the updated Contextual Taskbar, speeding up the workflow for complex masking projects. 5. Other Notable AI and Workflow Updates
: The expanded prompt window and centralized Harmonize tool improve ergonomics. Consistency The represents the absolute best generative image editing
Are you running into any with the Creative Cloud desktop app? Share public link
Once installed, launch the Beta version directly from that menu.
: Users can upload multiple images to guide a single generation, giving the AI a better sense of style, color, and composition. Larger Prompt Window The new feature changes that by bringing text-to-image
Adobe has a roadmap. The current (Firefly 3) is Phase 1. Phase 2 (expected later this quarter) includes:
Adobe continues to lead the industry in ethical AI deployment. Because the Firefly model is trained exclusively on licensed content, such as Adobe Stock, and public domain images, all content generated in the Photoshop Beta is commercially safe.
After installation, go to Preferences > Image Processing and ensure "Use Firefly 3" is toggled ON. Adobe sometimes leaves legacy models active for stability.
To help tailor future guides, let me know you primarily focus on, or which specific AI feature you want a deep-dive tutorial for next. Share public link