The good folk at Velocity have been undertaking some real-world testing by getting generative AI imaging tools to create variations on a free stock photo. This is a good test, because it’s something any of us might want to do. Royalty-free images can be great, but if we use them unedited they can crop up on other websites, which isn’t the best look. Or we might find one that’s almost right, but not close enough to use.
If we use an image as the basis for an AI-generated original alternative, we may be able to get exactly what we want, or at least something close enough.
In the test, the team took a ‘starter’ photo and asked Midjourney to describe it. Then they used the description as the basis for new AI-generated photos. It’s actually an experiment any of us could replicate. You can see how they got on here.