AI children's books, body horror edition
Top Comments
I've found that by putting meaningful effort into AI storytelling, I can create bespoke stories that my kids love night after night.
My workflow is below: Caveat that it costs about $0.25-$0.50 to weave a book like this with Claude Sonnet and Gemini Nano Banana Pro. But to me the cost is worth it for the quality.
- Use Claude structured output and ask for page1, page2 ... pageN instead of an array of pages or wall of text.
- Pass a story arc as a set of values to the prompt. I.e. say each page has an emotional beat between 0.0 and 1.0. For a "man in hole" type of story: page1 starts at 0.6, page2 = 0.5, page5 = 0.25, page10 = 0.85. This ensures page 5 lands the "crisis" and page10 resolves higher than the start.
- For illustrations, have Claude generate the story text and an illustration prompt per page. i.e. page1: { "text": "...", "illustration": "..." }.
- For art consistency, add an "Art Direction" key to the structured output. Feed this into Gemini/OpenAI and ask for an art board visual guide & character reference sheet.
- Feed the page text, illustration prompt, and the art board to Gemini/ChatGPT images. I'm constantly surprised at the quality of the output.
Here's an example set of pages from a magic school bus style story about the immune system
[image] https://media.discordapp.net/attachments/839188039229112353/...
I'll start: John Rocco, How We Got to the Moon. (http://www.howwegottothemoon.com/)
I think a point missing is that this output all looks the same because the prompters are not specifying much more than the barest minimum to get what they want. If you just prompt "generate a cover for my book 100,000 whys which is childrens book that answers their questions about science" then you get images like from TFA using the models default style. However, the models are capable of reproducing any great artists style and any content you want.
If you have seen the prompts for images on communities of enthusiasts you may notice that they can be quite long and specify considerable detail about both the content and the style of the output.
Here is one of the four above the fold on the front page of CivitAI for me right now, it has both a positive and negative prompt. Not that long because this is a fairly simple image. However the image doesn't look like the slop in the 100,000 Why's book covers or the many commercial signs and advertisements I'm seeing when I leave the house.
Afaik, parents are super protective of their children and would never do something that could inhibit a childs learning
Visit the Original Link
Read the full content on lcamtuf.substack.com