Why AI Isn’t Always the Answer for Photo Edits

Getting natural-looking results when expanding your images can be frustrating, especially when relying solely on new AI tools. Knowing alternative methods can often yield far better outcomes.

Coming to you from Glyn Dewis, this practical video highlights why generative AI isn't always the ideal solution for expanding backgrounds in Photoshop. Dewis demonstrates with a composite portrait of his friend, combined from separate background images to create a realistic environment. When attempting to widen the background using Adobe’s Generative Expand, the results initially seem impressive, but a closer look reveals disappointing limitations. The AI-generated details lack clarity and become noticeably blurred, creating an obvious mismatch with the original photo. Dewis tries enhancing the AI output, yet even after enhancement, the results remain unacceptably soft.

The limitations Dewis experiences with generative AI aren't unique to this specific example. He tests another popular method—Photoshop’s Content-Aware Fill—and initially lets the software handle everything automatically. However, automatic Content-Aware Fill completely fails to deliver usable results, resulting in strange repetitions and textures. Dewis then attempts manual control over the Content-Aware process, selectively telling Photoshop what areas to use and avoid when filling. Unfortunately, even careful manual adjustments fail to produce convincing backgrounds in this case, leaving him dissatisfied again.

Where Dewis finds real success is with an often-overlooked Photoshop tool: Content-Aware Scale. This feature has been available for years but tends to go unnoticed in the flood of new AI features. Dewis clearly demonstrates how Content-Aware Scale intuitively preserves subjects while stretching or duplicating backgrounds smoothly. Using Content-Aware Scale, the expansion is seamless and clean, effortlessly producing natural-looking results that outshine both Generative Expand and Content-Aware Fill in these scenarios.

However, Content-Aware Scale isn’t perfect either—especially if your subject dominates the frame. Dewis acknowledges this issue and carefully shows how you can overcome it. He reveals a lesser-known but powerful tip: creating a protective Alpha channel mask around your subject. By isolating and slightly expanding the selection, you provide Photoshop with clear boundaries. The software then knows precisely which elements must remain untouched, allowing you to significantly stretch backgrounds without distorting the main subject. This method proves remarkably effective, letting you dramatically resize images without ruining the realism.

Beyond this practical advice, Dewis offers a grounded perspective on AI. He emphasizes that although AI technologies provide rapid results, traditional tools hidden deep within Photoshop’s menus can still be superior. Relying entirely on generative AI can lead to frustrating dead-ends, especially when quality matters. Check out the video above for the full rundown from Dewis.

Alex Cooke's picture

Alex Cooke is a Cleveland-based portrait, events, and landscape photographer. He holds an M.S. in Applied Mathematics and a doctorate in Music Composition. He is also an avid equestrian.

Log in or register to post comments