This is a submission for the Presage SmartSpectra SDK Challenge
What I Built
Airloom
Create an artwork using your breathing and heartbeat. Select a colour scheme, allow the camera to collect data, and breathe freely.
Airloom does not require you to breathe at a certain pace or achieve a score. It uses everyday moments to create unique artworks.
Once the session has finished (after around 90 seconds of useful data, or at any point) the camera will turn off. You can watch the creation of your artwork again, give it a title, and download a high resolution PNG file.
Three different colour schemes can be used: Copper & Ink, Ocean Glass and Sunset Silk. There is a clear simulation to test the artwork without using the camera or an API Key.
Demo
Code
Airloom
A portrait of this moment. A desktop art instrument that weaves a textile from live camera-derived breathing and pulse using Presage SmartSpectra.
Run
Use Node 24 LTS (minimum 22.12) on a supported SmartSpectra platform: macOS Apple Silicon, Windows x64, or Linux x64/ARM64 with glibc 2.35+.
npm install
npm run setup:desktop
cp .env.example .env
Set PRESAGE_API_KEY in .env to your developer key from https://physiology.presagetech.com/. Never commit it. Then:
npm start
The key is read by the desktop main process and passed to the renderer SDK after consent. This source checkout is intended for local developer use with your own key; do not distribute a build embedding your private key. Restart the app after editing .env.
On macOS, allow camera permission for Electron or the application that launches it (for example your terminal). Keep your face and upper chest visible, with even light and a steady camera. Initial measurements…
The repository includes setup instructions, environment configuration, tests, privacy notes, and a live-demo checklist. Users supply their own Presage developer key through .env.
Built with Electron, TypeScript, Vite, HTML Canvas, and the Presage SmartSpectra Node SDK.
How It Responds to Human Signals
Airloom uses SmartSpectra’s chest breathing waveform, breathing rate, and pulse rate.
Live signal How Airloom responds
Chest breathing waveform Changes the curvature and spread of the woven threads.
Pulse rate Changes the spatial rhythm of the thread pattern.
Breathing rate Appears in the live measurement panel alongside pulse.
Validation status and signal freshness Pause weaving when measurement conditions are unsuitable or breathing data becomes stale.
I integrated SmartSpectra through its official Electron main-process, preload, and renderer bridge. The renderer starts sensing after consent, receives metric events, and decodes the SDK’s protobuf packets.
const sdk = new SmartSpectraSDK({
apiKey,
requestedMetrics: [
...breathingMetrics,
MetricType.PULSE_RATE,
],
enableTelemetry: false,
});
The breathing waveform passes through rolling-window normalization before controlling thread geometry. Pulse rate influences the pattern; Airloom does not claim to detect or synchronize with individual heartbeats.
Live mode never substitutes simulated measurements. Missing or stale pulse readings appear as a dash, while unsuitable breathing signals pause the weaving process.
Replay reconstructs the artwork from session samples held in memory, without reactivating the camera.
Prize Categories
Most Creative or Delightful Use
Airloom began with a question: What would a portrait look like if it captured your rhythm instead of your appearance?
Textiles offered a natural visual language: individual threads become a larger pattern, just as small physiological changes become the texture of a moment.
The playful part is becoming the instrument. You do not need drawing skills or continuous mouse input to influence the piece. Breathing naturally contributes to its shape, and your pulse contributes to its pattern.
The finished artwork gives the interaction a tangible outcome: a small keepsake you can name, revisit, or save.
Consent and Data Handling
Camera consent: Live sensing requires an explicit consent checkbox and an Enable camera & begin action. Users are asked to measure only themselves. A camera preview remains visible during capture.
Stopping capture: Finishing a session or choosing Stop & discard stops camera tracks and shuts down the SDK session. Replay and export happen with the camera off.
Application storage: Airloom does not record video or audio, create accounts, or intentionally persist physiological readings. Recent readings and signal-derived replay samples remain in application memory.
Retention and deletion: Discarding, starting a new moment, or closing the app clears session references. This is not a secure memory-erasure guarantee.
Optional export: Saving creates a PNG containing the artwork and chosen title, without numeric measurements or camera images. The image is still derived from personal signals and is not guaranteed anonymous. Exported files remain until users delete them.
Sharing: Airloom has no automatic artwork-sharing or application-upload feature.
SDK telemetry and services: Optional SmartSpectra telemetry is explicitly disabled with enableTelemetry: false. Authentication, model delivery, usage accounting, and other SDK service requests may still occur. Presage-held data follows Presage’s policies; clearing Airloom cannot delete server-side records. SDK network behavior and server-side retention have not been independently audited.
AI services: Airloom does not request SmartSpectra LLM Insights or send session data to another AI service. Codex assisted with development.
Top comments (0)