The models suck at removing the background of an image, and they fail at cropping the image to the edge of the graphics to reduce whitespace for things like logos. It would be a lot easier if I could just manually remove the background and crop with a button or tool instead of asking the model to do it over and over again with unsuccessful results. Took about four tries to get the final correct result.
Hi and welcome to the community @emc2,
I just wanted to recommend to check this topic out (if you already haven’t): Having trouble getting transparent backgrounds in ChatGPT images