Exploring Fungal Morphology Simulation and Dynamic Light Containment from a Graphics Generation Perspective

TL;DR

This study equates fungal morphology simulation to a two-dimensional graphic time-series generation problem and proposes a zero-coding, neural network-driven cellular automaton, which successfully spreads into pre-designed complex shapes in reality.

Abstract

Fungal simulation and control are considered crucial techniques in Bio-Art creation. However, coding algorithms for reliable fungal simulations have posed significant challenges for artists. This study equates fungal morphology simulation to a two-dimensional graphic time-series generation problem. We propose a zero-coding, neural network-driven cellular automaton. Fungal spread patterns are learned through an image segmentation model and a time-series prediction model, which then supervise the training of neural network cells, enabling them to replicate real-world spreading behaviors. We further implemented dynamic containment of fungal boundaries with lasers. Synchronized with the automaton, the fungus successfully spreads into pre-designed complex shapes in reality.

Figures (7)

• Figure 1: Framework of our method. The E-ViT model first captures temporal edge segmentation images by observing real fungal spread. Subsequently, the TCN model learns spread patterns from these temporal images. Third, the TCN model is used for supervised training of the NN that drive the cellular automaton, aligning it with real-world fungal growth patterns. Finally, the laser device used to limit the spread is connected to the NN-cellular automaton, facilitating the execution of simulations while achieving the spread of pre-designed complex shapes.
• Figure 2: AI model chain learn fungal spread pattern and supervise the training of NN-cell for simulation. Upper: E-ViT captures the edge of sequential fungal images and transfers them to the TCN model to learn its spreading rules. Lower: The TCN model supervises the training of NN-cells, enabling them to execute fungal spread patterns in the simulation environment that are consistent with reality.
• Figure 3: Random spreading simulations of Phanerochaete velutina, Physarum polycephalum, and Aspergillus niger.
• Figure 4: Experiment on light containment of Rhizopus Oligosporus. Upper left: Experimental setup. Upper right: Normalized containment rates of Rhizopus Oligosporus with lasers of different powers and wavelengths. Lower left: Detail of containment using a 405 nm 20 mW laser. Lower right: Simulated light containment using NN-cellular automata.
• Figure 5: Dynamic light containment testing. Upper: The NN-cellular automaton, with the addition of light entities, achieves dynamic light containment, turning off in stable fungal areas and on in active areas. Lower: The laser device, connected to the simulation environment via an API, successfully forms fungal spread patterns consistent with the simulation after executing a dynamic light containment strategy that mirrors the simulation.