I’m trying to create a prompt which will combine multiple movies into one image, such as “Star Wars. Titanic. Toy Story” - I want these all to be put into one image but DALLE3 is currently almost always making an image of a collage of the three different movies separate. Any way to get around this?
1 Like
I would use DALL-E 3 inpainting to try to merge them together.
You can describe what that mashup actually is in terms of individual element and how they should be composed.
Like this one where the AI actually re-wrote Star Wars as “characters from a famous space opera”, on a “grand ship”, thus distorting the meaning when broken down into elements, and giving opera-goers a presence on a clipper ship with rigging in the image. “Beloved animated movie known for its heartwarming tale of toys coming to life, with characters that might remind one of a cowboy doll” was instead inferred and reproduced in a striking fashion.
You can instead place lots of instructions telling the AI not to rewrite any prompts, but to pass them unaltered, and then do your own language that avoids trademarks that would make DALL-E return an error.
A prompt unaltered by ChatGPT with my own stand-in descriptions is closer:
