WHEN GEOMETRY MEETS MATTER

Fabrication complexity of ceramic morphologies, through the Stanford Bunny

2022 — 2023 | Harvard GSD | Ceramic, Plaster, Silicone, 3D Print

CONCEPT

The Stanford Bunny is one of the most recognizable test geometries in computer graphics, used for decades to benchmark rendering algorithms, mesh processing, and geometric computation. It represents, in compressed form, every organic shape in the digital world. But what happens when this geometry leaves the screen and enters the physical world through one of the oldest material processes we know?

This project investigates the fabrication complexity of translating computed geometries into ceramic objects through traditional slip casting and silicone molding. Starting from a single mesh, I generated a catalog of geometric variations — low-poly, one-strip, woven, patterned, voided — and systematically tested which could survive the journey from algorithm to clay.

CLASSIFICATION

In consultation with ceramicists at the Harvard Ceramics Studio, the computed geometries were classified into four fabrication categories based on their physical feasibility. Forms with internal structures were deemed unachievable by any existing method. Hollow forms with surface patterns could be achieved by casting around a solid interior. Geometries with significant undercuts required silicone molds rather than plaster. And the simplest variations could be produced through traditional slip casting alone — though "simple" is relative when each piece required over 80 hours of labor and a four-part plaster mold.

MATERIAL

The silicone molding process demanded precise calibration — too rigid and the mold cracked during demolding, too flexible and the geometry deformed under its own weight. Each prototype cost over $350, with silicone alone accounting for $200. The plaster mold had to be separated into four interlocking pieces, and assembling them incrementally introduced cumulative risk: each additional piece could destroy the ones already made.

FINDINGS

1. Resolution works in reverse. In computer graphics, low-polygon models are preferred for faster rendering. In ceramic fabrication, the opposite holds: high-resolution meshes produce cleaner molds and more reliable castings. The faceted surfaces of low-poly geometry introduce undercuts that trap the clay.

2. The complexity of the ceramic is bounded by the complexity of the first mold. If a geometry cannot be successfully 3D printed as a mold positive, it will not survive the subsequent translations into plaster and clay.

3. Each translation between mold stages degrades resolution. By the time geometry passes through 3D print, plaster, silicone, and clay, fine details that were crisp in the digital model become soft, rounded, ambiguous.

4. Small undercuts scale catastrophically. A surface pattern only 1mm deep, repeated across the entire form, made it impossible to separate the bunny from its plaster mold without breaking the mold apart.

5. The weakest geometry governs the whole. The bunny's ears — the thinnest, most cantilevered feature — broke in nearly every casting attempt. The feasibility of the entire form was determined by its most fragile part.

EXHIBITION

View exhibition at Harvard GSD →