BLANK CANVAS

An Interactive Audiovisual Installation Powered by Generative AI

What happens when a single brushstroke triggers an entire world? Blank Canvas is an interactive installation that augments freehand drawings in real time using a stable diffusion model — transforming minimal strokes and text prompts into continuous, evolving imagery while the act of drawing itself shapes a procedurally-generated soundscape delivered through a quadraphonic speaker array.

About the Project

The system invites participants to draw freely on a custom iPad app, where each gesture simultaneously steers AI-generated visuals and influences spatialized sound. The result is a generative playground where illustration, language, and music converge into a single multimodal experience — one shaped entirely by the participant's imagination.

Blank Canvas was built as a provocation: rather than treating generative AI as a replacement for creative labor, the system positions it as a collaborator — amplifying human gesture instead of automating it. The project celebrates post-human creativity, engaging audiences in a direct, felt conversation about what AI-augmented art-making can actually look like.

Exhibited at the Neo Arcade showcase and the Goat Farm Arts Center.

How It Works

A custom Swift-based iPad app captures Apple Pencil strokes and text prompts, routing them over a local network via OSC. A TouchDesigner patch integrates a stable diffusion model that continuously generates imagery from the drawing input. Simultaneously, pencil data drives a VCV Rack patch producing procedural audio spatialized across four speakers. The full pipeline — drawing, generation, and sound — runs locally in real time over NDI and OSC.

Collaborators

Rafael Collado — Custom iPad app (Swift, OSCKit)
Ishaan Jagyasi — TouchDesigner integration, stable diffusion pipeline, procedural audio (VCV Rack)
Kyle Smith — Local infrastructure, model implementation, spatialization, quadraphonic hardware

Developed under Dr. Jeremy Muller · MUSI 6201 – Integrated Multimedia · Fall 2025

Previous
Previous

Spores_WIP

Next
Next

Real-Time Visuals