I’m probably late and everyone already talked about GPT Images 2.0, but I wanted to test it for one specific thing - YouTube thumbnails. This is basically a short version of the test I did for my YouTube video, so I’ll try to keep it focused on the practical stuff. Not just making a nice AI image, but actually trying to use it in a normal creator workflow with faces, text, products, references, edits, style matching and cleanup. Some of this may be obvious for people who already work with ...