1
I know it's fun to ask ChatGPT for a joke about X, but I wanted to take a step further by asking it for:
1. a comic in three parts 2. to include lines and a title 3. prompts to draw it
I'm asking ChatGPT to output JSON with the comic and then I parse it and move it to the next step of the pipeline (drawing). Tools on their own are cool, but chaining them is the true multiplier of creativity.
Check the source on Github: https://github.com/infomiho/dailyaicomics
The hardest thing so far was figuring out how to get a bit more consistent style since each panel is draw in isolation. I've tried asking "draw in the style of X famous comic writer" but that had two drawbacks: - their style varies (color, black and white) - since it's a comic author, it oftentimes draws a whole comic (multiple panels)
Now I'm being very explicit around the style (ink, pastel colors etc.) and it's okay.
I'm wondering what's the best approach to get consistency?
I whipped it up to help me learn Italian using a method I used way back in high school to faster retain historical facts. It has a pool of 100ish phrases and after you go through them all at least once, it will start prioritising phrases that you had trouble with. This way you are focusing on the toughest phrases more. This is kinda like roulette wheel selection used often with genetic algorithms.
I wanted to blend in when I was in Italy and not speak English all the time. I only had a week, and of course, I built an app to help with that. It was awesome saying "Mi scusi" all the time and at some point, they helped us to find our way when our phones died!
It's built using a full-stack framework I'm also developing called Wasp (http://wasp-lang.dev) and I rebuilt it from Supabase to Wasp when I joined the company.