Everyone can build a storybook business, but not everyone can create a compelling story and style. This is where my edge is. Creative writing is one of the things humans are still better at. So I am even considering hiring an actual flesh-and-blood writer to write my story.
A good story and great style further my moat. The fact that I’m tempted to shortcut with AI is evidence that a non-shortcut is differentiating.
First I tried using the Gemini web interface to create all storybook spreads in one go. To avoid the images slightly changing over the course of the 10 spreads I thought this might help with consistency. Spoiler alert: it didn’t. Gemini just stopped after a few minutes and did not do anything.
So I tried separate prompts for each spread instead. That didn’t work either, the first and last image looked like they belonged to different books.
Then I realized that I didn’t need one continuous scene across all the spreads. I was fighting to make the consistency work, which I didn’t actually need. I just needed some elements to be consistent throughout the story.
It made more sense to have close ups of elements or change the scene entirely over the course of the book.
So I did another run trying to change the prompts more into what I wanted. It improved a lot this way. But still it required a lot of re-tries and fine tuning the prompt.
Key Insight:
Anyone can prompt Gemini to make images. Almost no one will sit with it long enough to get something that actually feels like a book worth reading your kids.