Research

Stuth’s research praxis is grounded in systems theory and intersects the worlds of process design and product design. We are fascinated by wicked problems and following the questions that surface in our work, guided by our ethos of sustainable, localised production through digital distribution.

Our research brings ideas into processes that become material explorations, forms and solutions. Moving between system, code and material, Stuth’s research has developed both the parametric models that generate form and the fabrication methods that make it physical. 

But it’s in the gap between these that most of the interesting problems live.

Parametric Configurator

The way we design, manufacture and purchase products has rapidly transitioned to digital platforms, and the next phase in product design is parametric modeling. With the capability for completely unique and customizable product offerings, parametric modeling redefines ‘made to order’ products to support a new standard for what consumers can expect in designed outcomes. Simply put, parametric modeling enables us to incorporate customer values and needs directly into the design and making of products.

Stuth's long-term platform research project has been creating a tool that enables real-time customer configuration of parametric products. Form, pattern, size, and colour can each be selected and previewed in-browser before production.

Research architecture: Grasshopper definitions executed server-side via Rhino.Compute, geometry delivered to a Three.js viewer in the browser, integration with the shop for direct ordering.

The challenge isn't developing the technology, it's creating an interface for a 588-combination system that appears simple to use without collapsing it into a dropdown menu.

Wireless Infrastructure

Evaluating LoRa mesh networking for studio communications and monitoring. Firmware comparison across Meshtastic, MeshCore, and Reticulum — each with different trade-offs between maturity, resource requirements, and privacy architecture.

Current hardware: Heltec WiFi LoRa 32 V4 (ESP32-S3 + SX1262). Investigating custom node designs using ESP32-C3 with external LoRa modules for specific applications.

The research interest is decentralised, low-power communication that doesn't depend on WiFi infrastructure or cloud services.

Hardware Research

Print Platform Development
Investigating mechanical and firmware improvements to FDM platforms for production reliability.

  • Z-axis synchronisation. Four-point Z systems require precise tramming. Research into mechanical approaches (set screws with defined torque sequences) and firmware-level active correction. Custom G-code implementations (M915, M916) for automated tramming and probe offset calibration.
  • Bed probing. Optical probe concepts for non-contact bed levelling. Investigating accuracy limits and repeatability compared to mechanical probes.
  • Modular architecture. Designing printer frames where the XY motion system is reusable across different Z-height configurations. Same precision, different build volumes.

Closed-Loop Drive Systems
Early-stage research into replacing stepper motors with geared brushed DC motors using magnetic linear encoder feedback. Steppers waste significant energy as heat and have no position confirmation. Closed-loop DC with linear encoders on the actual carriage — not the motor shaft — could improve efficiency, torque density, and positional accuracy.

Production-Scale FDM

Six networked FDM printers running continuously. The research here is operational, how to run a small print farm reliably, efficiently, and without manual intervention at every step.

  • Fleet management. Custom dashboard for monitoring all six machines from a single interface. Real-time temperature, job progress, filament tracking, and health diagnostics. Built on Flask with OctoPrint APIs.
  • File synchronisation. Decentralised sync across studio machines and printers using lsyncd and Syncthing. G-code generated once, distributed automatically. No central server dependency.
  • Batch processing. Automated slicing pipeline — model files are processed through SuperSlicer templates without manual interaction. Post-processors handle sequential printing geometry. Reduces per-file handling to minutes.
  • Failure detection. AI-assisted print monitoring flags failures in progress, reducing wasted material and machine time.

The problem being studied: what does reliable, small-scale digital manufacturing actually require? Not the idealised version, but the one that runs on a Wednesday when something jams.

The Language of Patterns

Each of the fourteen surface patterns carries a Scottish Gaelic name signifying its origin. The name describes what the pattern is, not what it looks like.

Water patterns derived from fluid dynamics at different scales:

  • loch — concentric displacement from a point disturbance in still water
  • cuan — rolling periodic displacement with variable amplitude (open ocean)
  • mar — parallel flow lines with directional bias (tidal movement)
  • uillt — high-frequency ridges from constrained fast-moving flow (stream)
  • lon — low-amplitude undulation with irregular periodicity (marsh)

Textile patterns derived from thread behaviour under tension:

  • tana — fine parallel striations at high density (thin fabric)
  • tiugh — bold ridges with pronounced depth (heavy weave)
  • cotan — interlocked texture from warp-weft intersection (cotton)
  • anart — crisp folded geometry with angular transitions (linen)
  • sioda — continuous low-friction flow with minimal surface interruption (silk)

Organic patterns derived from biological and material deformation:

  • fas — branching displacement from biological expansion (growth)
  • neonach — asymmetric deformation with no repeating unit (strange)
  • pasgadh — coiled surface geometry from continuous winding (wrapping)
  • fidheall — dense interlocking carved pattern from interlaced paths (weaving)

The naming system extends to forms (named for their character in Gaelic) and printers (named for water features).