by eskibars on 4/14/24, 6:37 AM with 0 comments
The text comes from gpt-4-turbo. Images are courtesy of dall-e-3.
The biggest challenge so far has been getting the images to feel cohesive across the story: since they're auto-generated as the image goes, it needs to describe each of the characters very specifically so that they don't look like completely different people on each frame. At some point, I might use a different approach.