Experiment · Running

Parametrick

A generative vase designer: you load a song or a photograph, the app pulls out their features and builds a parametric form from them — a vase that encodes a personal memory.

Status
Running
Type
Tool
Platforms
Desktop
Capabilities
Sound
Technology
Next.js · three.js · Meyda
Updated
11 February 2026

What it is

The first public test of what later became the project Mnemoform: turning a memory into a physical object. The user brings a song or a photograph and receives a vase whose form is assembled from the features of that material.

How it works

  • Audio. Meyda extracts the spectral centroid, RMS, ZCR, spectral flatness and other features, while a temporal analysis maps the dynamics onto the height of the vase: loud passages give a wider profile, quiet ones a narrow one.
  • Photo. Analysis on a canvas: brightness, saturation, edge density and colour distribution.
  • Everything is computed on the client — neither the audio nor the photograph is sent anywhere.

What you can change and where it ends up

  • A live 3D preview and sliders: twist, wave, bulge, curvature, wall thickness.
  • Export of watertight STL and 3MF with a physical scale, a print-volume check and material presets (PLA, PETG, PLA Silk).
  • The process is assembled from six steps: choosing the material → typology → context → synthesis → tectonic refinement → fabrication preview.

Stack

  • Next.js 16 (App Router), React 19, TypeScript
  • three.js via @react-three/fiber
  • Meyda — audio feature extraction
  • Geometry: a B-spline profile → LatheGeometry → deformations → wall thickness → subdivision

Links

Original (Russian) →