Gone are the days of blurry upscaling. The new tool, developed in partnership with Topaz Labs , brings the power of Gigapixel AI directly into Photoshop. This feature uses machine learning to enlarge images while intelligently adding detail, producing sharper and cleaner results than traditional bicubic interpolation. Whether you are restoring a low-res historical photo or need a thumbnail sized for a billboard, Generative Upscale maintains edge integrity and texture resolution at a level previously impossible inside Photoshop.
This integration is built for hyper-realism. It excels at preserving the original background and texture when processing minor human alterations, such as removing clothing imperfections or touching up subtle facial elements.
At the core of the 27.0 Beta is the integration of the Adobe Firefly Image 4 Foundation Model. This underlying AI engine delivers drastic improvements in photorealism, text rendering, and complex prompt comprehension. Ultra-Photorealistic Textures and Lighting photoshop 27 beta
: The March update (v27.6) includes improved logic for removing wires, people, and other complex distractions from images.
Now fully optimized in the v27 architecture, acts as an automated composition blender. When a user drops a foreign object into a new background layer, Harmonize automatically samples the ambient environment. It recalibrates the foreground object’s color temperature, introduces matching highlights, and projects realistic environmental shadows to unify the composite in seconds. 🎨 Non-Destructive Design Workflows Photoshop 2026: 12 New Features, One Big Catch! Gone are the days of blurry upscaling
Selecting intricate details like human hair, translucent fabrics, and complex motion blurs has been completely solved in this beta release.
The beta builds introduce several significant enhancements aimed at automation and non-destructive editing: Dynamic Text System : The experimental Dynamic Text Whether you are restoring a low-res historical photo
When using Generative Fill on a person, you can now input a reference image, allowing the AI to keep the same person, clothing style, and lighting across multiple generated poses or backgrounds.
The Spot Healing tool has evolved into a background-aware . By simply circling an unwanted element, the AI removes it and reconstructs the hidden background elements using multi-angle patch prediction. The engine analyzes the broader scene structure to continue perspective lines and repeating patterns seamlessly. 4. Interface Modernization and Fluid Workspace
To generate content in the (v27.0), you can use the expanded Generative Fill and Generate Image features , which now integrate high-end third-party AI models for more realistic and precise results. Key Methods to Generate Content