[Andrew Clark]

A satellite image, 120 Crayola crayons, and the mathematics between them. Color-a-Pixel maps Landsat imagery to a numbered grid that can be filled in collaboratively: a NASA-featured outreach activity rebuilt as an interactive investigation of color quantization. A 3D color space visualizer with linked brushing lets you explore the geometry of the mapping: watch a dozen ocean blues converge on a single crayon, see where the Crayola palette crowds the greens and abandons the blue-green boundary your eye distinguishes most finely. Toggle between CIELAB and RGB matching and the color space reorganizes in real time. "Nearest" means something different when you ask the eye instead of the sensor.

A 200-band hyperspectral pixel is not a color. It is a point in 200-dimensional space, and you cannot look at 200 dimensions. This tool projects high-dimensional datasets into 3D and lets you change which projection you're looking at, morphing smoothly between PCA, t-SNE, and manual axis views. Parallel coordinates show the full-dimensional signatures the projection hides. A spatial minimap connects abstract clusters back to physical locations on the ground. Five bundled datasets (Synthetic hyperspectral, AVIRIS hyperspectral imagery, world development indicators, a knowledge graph of 120 scientists embedded by their relational structure, and GloVe word embeddings) demonstrate the same question across domains: what structure survives when hundreds of dimensions collapse to three? An integrated essay with live action triggers guides the reader through the tool as they read.

The topology of the yin-yang, from 2D through the Hopf fibration. A real-time volumetric ray-marcher renders the classical symbol as a helical sweep through a cylinder, torus, and the 3-sphere, exploring the geometric parallels between ancient complementarity and fiber bundles, gauge symmetry, and topological quantization. Responsive sonification maps the topology to sound: winding numbers become harmonic intervals, the cross-section becomes pitch, and the Hopf fibers become a slow counter-phase tremolo.

A sealed glass vessel, a turntable driven by magnets through the glass, and a stereo camera pair watching a mushroom grow, capturing the full 3D geometry every hour while PID loops hold CO₂, temperature, and humidity to a programmed recipe. MPC-1 is a portable robotic laboratory that generates continuous spatial models of biological growth and closes the loop: the system modifies environmental conditions based on the physical shape of the organism, without human intervention. The goal: high-fidelity training data for physics-informed neural networks that model growth dynamics in response to environmental stress. Currently tuned for fungi. Designed to operate autonomously in environments with no internet access.