An AI cover design workflow

I’m going to document a new process I like for book cover designs using Ideogram. One issue with generating covers with AI is the inability to separate out different elements, like the title. But if you generate a background image alone and add text later, the design might not be fully integrated. This process starts with a complete cover design, then separates out the different pieces.

1. Generate a full cover image. Here I was using another image I generated as a reference style, so I can actually create multiple books in a series with a similar style.
2. Use Ideogram’s Canvas tool (Magic Fill) to remove the text.
3. Use ChatGPT to create a vignette effect I want to apply to the image.
4. Use an image editor to select the original text and copy/paste it pack in as a layer. In fact, here you can see I used a different title from another image I generated.

This allows me to move the text around, add effects, and reuse the text in other places.