The biggest issue I am having lately with ChatGPT image generation is that when I request text to be over layed its often slightly off. Off enough to show that sloppiness and the text is embeded into the image so the only way to change it is to reprompt.
One work around I had was to have chatgpt create the image then I move it into canva and apply the text. I just saw canva can create images with 'magic layers' I am wondering if anyone has come across a method or a tool that can use a model to create an image on 1 layer and then create the text on another layer allowing you to make adjustments?