Abstracta2.0

A Python-based tool that recreates images using simple geometric shapes (circles or triangles). This project builds upon geometric primitive image generation techniques to create abstract representations of target images.

Features

• Support for multiple geometric primitives (circles and triangles)
• Command-line interface with customizable parameters
• Progress tracking with time estimates
• Intermediate result saving
• Configurable output resolution
• Performance optimizations using NumPy

Installation

1. Clone the repository:

git clone https://github.com/Muneer320/Abstracta2.0
cd Abstracta2.0

2. Install required dependencies:

pip install numpy Pillow

Usage

The program can be run from the command line with various options:

python main.py -i <input_image> [options]

Required Arguments:

• -i, --image: Path to the input image

Optional Arguments:

• -s, --shape: Shape to use ('circle' or 'triangle', default: circle)
• -n, --number: Number of shapes to generate (default: 1200)
• -r, --resolution: Output resolution (default: 512)
• -w, --working-res: Working resolution (default: 256)
• -k, --keep-progress: Save progress images (default: True)
• --no-progress: Disable saving progress images

Example:

python main.py -i input/photo.jpg -s circle -n 1000 -r 1024

Algorithm Overview

1. Image Initialization

   • Load and resize target image to working resolution
   • Initialize canvas with average color of target image

2. Shape Generation

   • Supports two primitive types:
     • Circles: Defined by center point (x, y), radius, and RGB color
     • Triangles: Defined by three vertices and RGB color

3. Optimization Process

   • Uses hill climbing algorithm to find optimal shapes
   • Performs multiple attempts with random starting positions
   • Mutates shapes to minimize root mean square error
   • Implements efficient numpy-based calculations

4. Progress Tracking

   • Real-time progress updates with score
   • Time estimation for completion
   • Optional intermediate result saving

Technical Implementation

The project uses several key components:

ShapeImageGenerator Class

class ShapeImageGenerator:
    def __init__(self, target_image, shape='circle', count=300,
                 base_resolution=256, output_directory="output",
                 keep_progress=True)

Main class responsible for:

• Image processing and shape generation
• Optimization algorithm implementation
• Progress tracking and saving
• Final image rendering

Key Methods:

• generate_random_shape(): Creates random circle or triangle
• mutate_shape(): Modifies existing shape parameters
• hill_climbing(): Optimizes shape placement
• generate_abstract_image(): Main generation process

Example Results

1. Original image
2. Final generated image (1500 Iterations)
   • Circle:
   • Triangles:

Performance

The generator includes several performance optimizations:

• NumPy-based calculations for faster processing
• Efficient random number generation
• Optimized image manipulation using PIL
• Progress tracking with time estimates

Acknowledgments

This project is an extension of Abstracta by datavorous, with additional features including:

• Multiple shape support
• Command-line interface
• Progress tracking
• Performance optimizations