README

Using OpenAI and Stability.AI to generate a strip of comic.

The scenario must mention the characters with a physical description.

How it works

Forked from original project aschen/comic-generator. Huge thanks !

First, a LLM (OpenAI API) is used to split the scenario into 6 panels with their description and associated text.

Then for each panel:

• an image is generated with Stable Diffusion (Stability API).
• the panel text is added to the image The 6 generated images with their texts are then merged into a final strip !

Lessons learnt:

Creating a consistent aesthetics through prompting throughout the comic is definitely an "art". e.g. The tie that Thaliavar wears in each panel is different :) Ensure your prompts are consistently supplying as much physical details about the character and the background. In this case, the prompts for Stability is AI generated from OpenAPI - so it probably needs better prompt template to generate consistent outcomes

Usage

Export OPENAI_API_KEY and STABILITY_KEY.

Install dependencies: pip install langchain openai stability-sdk pillow

Then edit the SCENARIO variable in kartoon.py.
You can also edit the STYLE variable to change the style.

Run the script: python kartoon.py

Examples

Style american comic

Characters: Superstar is a guy who looks like young Rajnikanth has a moustache and wears a black suit.
New York City is in danger of an alien attack from a Spaceship. 
Superstar the young Rajnikanth lookalike flies towards the large Alien spaceship.
Superstar the young Rajnikanth lookalike saves the day by destroying the large alien spaceship.