The research group that directly turns novels into comics released the comic-style AI image software StoryDALL-E

The AI ​​image software that has become popular recently is increasingly affecting the professionalism of the painting world and even the animation and game fields. According to foreign media reports, UNC Chapel Hill in the United States has announced a new AI image software architecture StoryDALL-E, which can directly generate AI "paintings" in a comic style. As long as you describe it well, novels can be turned into comics in minutes.

•The AI ​​image software architecture StoryDALL-E is called StoryDALL-E: Adapting Pretrained Text-to-Image Transformers for Story Continuation. Its core architecture is Text-to-Image, and its function is to build AI comics directly from text.

•It’s not that the original AI image software is incapable of generating similar comics, it’s just that the background generation of comics is simple, but it’s extremely difficult to build the authenticity of the characters. StoryDALL-E has made improvements and adjustments to this aspect, allowing operators to generate comic characters more easily. Currently, StoryDALL-E is still under further research, and the comics generated in the future will have more details and smoother results.