A-FFECT CAMERA
Reimagining Computer Vision Beyond Facial Recognition
→ A real-time system that translates human presence into non-identifying particle-based visuals
Year: 2025
Medium / Type: Interactive Installation, p5.js
Exhibition: NYU Tisch ITP Winter Show
Tools / Tech:p5.js, WebRTC camera capture, Web Audio API, background subtraction algorithm, motion detection, bitmap font rendering, 3D printed ouroboros casing, Planar PS556T 55" vertical display
View Code
Context
What if a camera could capture your presence without identifying you?
Contemporary camera systems powered by computer vision are designed to identify, classify, and track.
Most cameras are optimized for extraction and labeling.
But they don’t see everyone equally, facial recognition systems have well-documented biases across skin tones.
Contemporary camera systems powered by computer vision are designed to identify, classify, and track.
Most cameras are optimized for extraction and labeling.
But they don’t see everyone equally, facial recognition systems have well-documented biases across skin tones.
Design Challenge
How might we design a camera that:
How might we design a camera that:
- registers bodily presence and environmental energy
- avoids identifying or extracting biometric data
- shifts perception from surveillance → sensation
Approach (Interaction + System Design)
Designed and implemented a real-time computational system that transforms live camera input into an abstracted, non-identifying visual field. Installed as a large-scale vertical display, A-FFECT CAMERA functions as a public-facing magic mirror:
Physical installation included a 55” vertical display and custom 3D-printed enclosure, emphasizing scale and spatial immersion.
Designed and implemented a real-time computational system that transforms live camera input into an abstracted, non-identifying visual field. Installed as a large-scale vertical display, A-FFECT CAMERA functions as a public-facing magic mirror:
- Captures live webcam feed via WebRTC
- Applies background subtraction + motion detection to isolate presence without facial recognition
- Translates input into 25,000+ dynamic particles, forming a continuously shifting representation of bodies in space
- Integrates audio-reactive thresholds to modulate system behavior based on ambient sound
- Overlays generative text triggered by environmental conditions, introducing a secondary interpretive layer
Physical installation included a 55” vertical display and custom 3D-printed enclosure, emphasizing scale and spatial immersion.
- Through live interaction, I observed how people:
adjusted their movement when they couldn’t recognize themselves - became more expressive when freed from visual identity
- engaged longer with the system as abstraction increased
-
A clear shift emerged:
people moved from feeling watched → to feeling felt
Outcome
As visual detail decreased:
Users weren’t concerned with “looking acceptable” they were trying to see what they could become.
As visual detail decreased:
- comfort increased
- self-consciousness decreased
- interaction became more playful and exploratory
Users weren’t concerned with “looking acceptable” they were trying to see what they could become.
Design Insight
Abstraction became a tool for agency.
The less the system defined the user, the more the user defined the interaction.
In a culture saturated with cameras, A-FFECT CAMERA proposes a new visual language:
Abstraction became a tool for agency.
The less the system defined the user, the more the user defined the interaction.
In a culture saturated with cameras, A-FFECT CAMERA proposes a new visual language:
- Visibility without exposure
- Presence without extraction
- Expression without categorization
Bitmap kinetic typography system generated through pixel-based constraints, exploring legibility, distortion, and modular form