TOOL SCHOOL
Photoshop’s AI-powered feature brings revolutionary new tools to image editing
With the emergence of Midjourney, Dall-E, Stable Diffusion and similar W AI-powered image generators over the past year, it was inevitable we would eventually see a generative AI feature in Photoshop. Powered by Adobe Firefly, the new Generative Fill and Expand features in Photoshop could be the most seismic update in the application’s 33-year history. The tool works like other promptbased image generators. Select an area, type something and watch as it appears in your image. It’s equal parts dumbfounding, scarily intelligent, and also somehow reassuringly disappointing.
There are some things it does brilliantly – the way it matches the backlighting in our original portrait is mind-boggling. The Expand feature works well and there’s much enjoyment to be had by cropping outside your original frame. But ask it to generate a new object and the results are hit-and-miss. Of course, the technology is in its infancy but for now, it is reassuring to know that the low-res, smudgy fills aren’t a match for what you can create with a camera.
1 Make a selection
Define an area to be filled by making a selection with the Lasso tool. Keep in mind that the entire selection area will be altered and the shape of the selection affects the fill. Here, we made a loose selection to create steampunk goggles, but the results didn’t match the face. By using Quick Mask to paint a rough shape of goggles, the resulting fill was more successful.
2 Type a prompt
Type a text prompt in the Properties Panel or Contextual Taskbar. This is where you can let your imagination run riot, but there are limits. Prompts deemed offensive, violent or fraudulent will trigger a warning. Adobe states that ‘Using generative AI features to create content that violates third-party copyright
