What it does
Real-time rendered eyes that react to everything around them.
Face Tracking
OpenCV DNN + MediaPipe detect faces and follow them with multi-layer parallax.
Depth Reactive
Pupil dilates dramatically as you get closer. Constricts when you walk away.
Audio Reactive
Pulses to bass beats, startles at loud sounds, looks toward noise direction.
Motion Detection
Tracks hands, bodies, any movement โ not just faces.
60fps Rendering
Smooth parallax animation with predict-to-vsync pipeline on Raspberry Pi.
Open Source
MIT licensed. Build your own for Burning Man, Halloween, or art installations.
Build Your Own
Everything you need to make a pair of tracking eyes.
Hardware
| Part | Price |
|---|---|
| Raspberry Pi 5 (8GB) | $80 |
| 2ร Round HDMI Displays (1080ร1080) | ~$100 |
| Pi Camera Module 3 NoIR | $35 |
| USB-C Power Bank (20,000mAh+) | ~$30 |
| Pi AI HAT+ (13 TOPS) | ~$77 |
| Pi AI Camera (IMX500) | $70 |
| USB Webcam with Stereo Mic | ~$30 |
Quick Start
01Flash Raspberry Pi OS
Flash Pi OS Bookworm 64-bit to a microSD card. Enable SSH, set hostname to raspieyes.
02Connect Hardware
Plug both round HDMI screens, camera module (CSI ribbon), and power.
03Clone & Install
git clone https://github.com/alevizio/raspieyes.git cd raspieyes && bash setup.sh
04Configure
Edit config.txt โ set RENDER_MODE=parallax, TRACKING=yes, choose your eye color.
05Reboot & Enjoy
Eyes start on boot. Walk in front of the camera.
sudo reboot
Built With
Open source projects that made this possible.
Common Questions
Everything else you might want to know.