• 

  Final Painting Presentation

• 

  Building Process

• 

  Natural Eye Lighting Change

• 

  Early Eye Development

• 

  Facial Detection to get the point the eyes should look at

• 

  Eye Tracking Tests

• 

  Some Early Painting Progress

• 

  End of Hackathon Painting Progress

• 

  Pre-event Soldering

Peeping Painting

The painting the stares back at you

Inspired by 1984.

The Art - Painting

We went for George Orwell as the main subject of the painting to stay in the theme of being watched. I started with a rough sketch in watered down acrylic and layered up paint to create the main subject and the figures around him. The rough strokes and pale colours add to the sense of discomfort.

The Eyes - Robotics/Raspberry Pi

We started with an existing system for displaying eyes and reverse-engineered it to achieve our goal. The eyes are 3D models rendered with OpenGL which rotate to look around. The webcam tracking program provides data for the eyes to follow you.

We were held back by performance. The Raspberry Pi 4 struggled to run the face tracking and eye rendering concurrently so we had to optimise. One major improvement was by compiling the Python scripts with Cython. We also tried smoother interpolation functions to get better movement.

The Tracking - Computer Vision

Opencv is used in a thread on the pi to track faces live. To make the eyes look like they are staring at people the point between the closest person's eyes is calculated.

Built With

• a-canvas
• acrylic-paint
• cython
• opencv
• opengl
• python
• raspberry-pi