胡宇齐(Yuqi Hu)

Chromasonic Specimen: RGB

Flame · Dream · Bubble

A resin-based installation that preserves AI-generated musical memories as a chromatic specimen of red, green, and blue.

Overview
Chromasonic Specimen: RGB is a small resin-based installation that preserves AI-generated images, pressed flowers, and chromatic fragments as
a material specimen of machine memory.
The work translates three musical inputs into symbolic memory states: Red becomes Flame, Green becomes Dream, and Blue becomes Bubble.
01
02
03
1 / 3
Chromasonic Specimen: RGB (Black-and-White Environment)
Material Context Details
Display Illustration

Material Context

In digital media, images often remain unstable, reproducible, and endlessly transformable. This work reverses that condition by sealing generated images inside resin. The digital image becomes a physical trace, while artificial flowers and botanical fragments turn the object into a specimen box. Rather than presenting AI memory as a screen-based flow, the work treats it as something that can be archived, preserved, and viewed like a fragile biological sample.

Concept

RGB color channels
RGB channels are used as symbolic carriers for three musical memory worlds.

Modern generative systems learn from large visual databases. Inside these systems, words are not only text; they become keys that activate stored visual patterns. This project treats prompt language as a way to recall machine memory.

Each song is mapped to a color channel and a memory archetype. The generated images are then transformed through particle systems, motion fields, and audio-reactive effects. The result is a living environment where memory is unstable, emotional, and constantly reforming.

The term "specimen" refers to the way the work freezes generative images into a physical object, turning computational memory into something closer to an archive, fossil, or botanical sample.

Three Memory Worlds

Red · Flame

Song memory: Rush E

Flame represents high-speed memory intensity. It becomes pressure, rhythm, collision, destruction, and rebirth through motion.

Green · Dream

Song memory: Spring

Dream represents spring-like memory. It carries growth, nostalgia, and unreal landscapes between sleep and awakening.

Blue · Bubble

Song memory: Blue

Bubble represents soft and fragile memory. It is transparent, floating, and easy to break, like emotional fragments suspended in water or air.

Process

The work is built as a pipeline from generative words to visual memory and then to post-processed spatial experience.

1. Song IdentityEach song is assigned a color and a memory form.
2. Generative WordsTitles, moods, and symbolic keywords become prompt language.
3. Visual DatabaseThe model uses learned visual priors to synthesize memory-like images.
4. Post-ProcessingImages become particles, motion fields, and audio-reactive transformations.

Technical Details

The system combines prompt engineering, generative image synthesis, audio analysis, and real-time particle simulation. Text prompts derived from music identity are used to query latent visual memory spaces learned from large-scale image datasets.

Generated outputs are decomposed into spatial particles and continuously transformed by sound energy. This creates a process where language becomes database retrieval, retrieval becomes image synthesis, and synthesis becomes embodied visual experience.

Keywords: generative AI, audio-visual art, computational memory, prompt language, post-processing, particle systems, experimental media.

Reflection

This work asks what happens when generated images leave the screen and become physical artifacts. By sealing AI-generated visual memories in resin, Chromasonic Specimen: RGB turns the fluid process of generation into a fixed object. It treats machine memory not as a stream of outputs, but as a fragile trace that can be collected, preserved, and re-read.