Original Article (Russian language)

Translated Article (Engilsh)

Live demo

This project explores a single recursive fractal generation system — evolving it not through predefined fitness functions, but through direct user interaction. You pick the fractals you like, and the algorithm adapts.

Instead of natural selection, it’s artificial selection by hand. Each fractal is a product of user-guided evolution, based on angle sequences (the genotype) applied recursively to subdivide line segments.

The system creates a surprisingly deep space of visual forms.

This project was created in 2017, two years before the Cellular Automata Evolution project — and stands as its conceptual predecessor.

• The genotype is an array of angles (in degrees).
• Starting from two points (A → B), a third point C is calculated using the current angle.
• The recursion splits the segment in two, creating more points based on those angles.
• The result: self-similar, often pseudo-chaotic shapes — fractals.

• Genetic Algorithm with User Selection: Evolves fractal angles (genotype) using crossover and mutation (e.g., 20% mutation rate), with users manually selecting preferred fractals to guide evolution across populations (e.g., 60 fractals, 3 populations, 20 per population).
• Fractal Generation: Produces fractals via a recursive geometric method starting with two points A and B, computing a third point C using an angle CAB, and recursively splitting into segments AC and CB with new angles.
• Interactive Visualization: Renders fractals in a web browser using HTML5 canvas, with controls for evolving populations and inspecting patterns.
• Manual Fractal Design: Supports user-defined fractal creation (2d.html, 2d.js) by adding or editing angles, with mouse-based dynamic visualization of the evolving fractal.
• Local Storage Persistence: Stores populations and fractal parameters in browser local storage, enabling users to resume evolution sessions after closing the browser.
• 3D Fractal Experiments: Includes advanced 3D fractal generation in geom3d/, extending the recursive method to three-dimensional space with mouse-driven rotation; while the PHP backend is legacy, the 3D visualization remains a significant feature.
• Dope as Hell!: I mean, c'mon! Check out the live demo. All those buttons are win98 style, and all those fractals have parent trees. o_O

fractal-monsters-evolution/
├── css/                 # Styling for the interface
├── geom3d/              # Code for 3D fractal generation (with legacy PHP backend)
├── 2d.html              # Interface for building fractals with mouse input
├── 2d.js                # Logic for mouse-based fractal construction
├── 2points.html         # Same as 2d.html, but uses quadrilaterals instead of triangles
├── draw2points.js       # Drawing functions for quadrilaterals fractals
├── draw.js              # Fractal drawing and recursive generation functions
├── index.html           # Main interface for fractal visualization and evolution
├── init.js              # Genetic algorithm and population management logic
├── screenshots/         # Sample fractal images or GIFs
│   └── demo.gif
└── README.md            # Project documentation

• Genetic Algorithm: Implements crossover and mutation to evolve angles defining fractal geometry, with evolution driven by user selection rather than a predefined fitness function.
• Fractal Algorithm: Uses a recursive geometric method starting with points A and B, computing point C with an angle CAB (genotype), and splitting into segments AC and CB, repeating with new angles up to a specified depth.
• Visualization: Renders fractals on an HTML5 canvas, with a simple Win98-inspired interface for user interaction.
• Modularity: Separates fractal generation (draw.js), evolution logic (init.js), and manual design (2d.js) from interfaces (index.html, 2d.html).
• Local Storage: Persists population data and fractal parameters in browser local storage for continuity across sessions.
• 3D Fractal Generation: Extends the recursive method to 3D in geom3d/, using angle pairs to define rotations around axes; includes mouse-driven rotation for dynamic viewing, though the PHP backend is legacy.
• Output: Supports exporting fractal images for analysis or documentation.

This project, created in 2017, explores the use of genetic algorithms for evolving geometric fractals through user-driven selection, predating the 2019 Cellular Automata Evolution project’s automated fitness-based approach. By focusing on interactive evolution, it provides a platform for studying user-influenced emergent patterns in computational geometry. Shared on GitHub, it offers researchers and enthusiasts a tool to investigate fractal systems and evolutionary dynamics.

MIT License. See LICENSE for details.

Serhii Herasymov
sergeygerasimofff@gmail.com
https://github.com/xcontcom