alexgreensh/anidoodle

What it solves

Anidoodle solves the problem of inconsistent and expensive AI-generated art by replacing token-based image generation with a deterministic, code-based drawing system. It allows users to create high-quality illustrations, animations, and music that are identical across every machine and size, eliminating the randomness and compute costs associated with traditional picture models.

How it works

The project uses a "art core" where every mark, note, and movement is defined as a pure function of code. By using seeded randomness and deterministic functions, the same source code always produces the exact same render. It features 31 distinct artistic styles (such as risograph, sumi-e, and pixel art) and a synthesized music engine that creates scores without samples or recordings. The system can be driven by AI agents (via plugins for Claude Code or Codex) to route plain-language requests to specific style recipes and workflows.

Who it’s for

It is designed for designers, brand managers, and developers who need consistent visual assets for websites (hero illustrations, interactive mascots), explainers, reports, social media (GIFs, stickers), and storyboards/films with original scores.

Highlights

• 31 Artistic Styles: Includes diverse mediums from charcoal and oil paint to toy bricks and woodcut prints.
• Deterministic Rendering: Guarantees byte-for-byte identical results on any machine.
• Code-Based Music: Synthesizes instruments like piano and marimba directly from data without using recordings.
• Interactive Web Art: Provides a lightweight ES module and custom HTML element for illustrations that react to cursor movement and user input.
• Image-to-Style: Can study an existing image to borrow its style or recreate a photo as a code-drawn illustration.
• Educational Tools: Automatically generates drawing lessons with step-by-step sheets and captioned timelapses.