A Mesh Agent System Based on Color Similarity for Image Style Transfer
Tech Stack: Rhino/Grasshopper Python Scripting, Agent-Based Modeling-ABxM.Core library
Teammate: Joyce Wu
[Chroma]Morph explores how agent-based modeling can be used to transfer image styles through localized, color-driven interactions within a mesh environment. Inspired by wet-on-wet painting techniques, the system simulates how colors dynamically evolve across a mesh as agents interact with their surroundings.
This project implements a similarity-based agent system where:
- Each mesh face serves as a potential agent in either a passive or active state.
- Active agents seek out other similarly colored agents and influence the mesh faces they traverse.
- A decay rate defines how much color influence is transferred.
- The simulation halts once all mesh faces reach an "active threshold" based on cumulative agent interactions.
- 🎨 Agent-based image style transfer based on color similarity.
- 🧩 Localized transformations of mesh faces within a simulated environment.
- 🔄 Real-time visualization of the propagation and blending process.
/GH Script → Core Grasshopper and Python scripts
/assets → Sample images and mesh inputs
/README.md → Project overview
This project is an early-stage exploration that combines visual computing and agent-based simulation. It aims to simulate painterly style transfer effects by mimicking physical dynamics within generative art.