Between Difference and Indifference― A New Digital Organism
This project is a real-time interactive installation using a depth camera (Intel RealSense D455/D455f), in which visually similar entities—artificial life and visitor avatars—coexist under the same environmental conditions but differ fundamentally in causality and ontology.
The installation intentionally avoids explicit explanation during the experience.
Understanding of the system is optional, deferred, and spatially displaced (e.g., text placed behind or at the exit).
The work is not about interactivity as empowerment, but about indifference, misrecognition, and asymmetric causality.
Core Conceptual Axes
Difference / Indifference
• Artificial life and visitor avatars:
• Are made of the same visual material
• Share the same environmental stress
• Appear nearly indistinguishable
• But:
• Artificial life evolves internally
• Visitor avatars do not evolve
• Artificial life is indifferent to the visitor
• The environment reacts only indirectly to visitor presence
Survival Pressure
• Sound (FFT → turbulence → iFFT) constantly distorts:
• UV coordinates
• Geometry
• Perceptual stability
• This applies equally to artificial life and avatars
Discontinuity
• Visitor contact causes:
• Complete freeze of world parameters (skybox, ray-marching movement, fast objects)
• Duration fixed at exactly 2 seconds
• Artificial life does not react
• After 2 seconds, system resumes previous dynamics
• This creates perceptual confusion and doubt:
“Something reacted—but not what I touched.”
Installation Structure (Exhibition Logic)
• Single-screen projection (front wall)
• Textual explanation placed on:
• Opposite wall, exit, or rear wall
• Viewer position:
• Physically between screen and text
• Ontologically inside the work
• Unable to intervene meaningfully
Not all visitors are expected to read or understand.
Possibility of understanding is sufficient—even if zero people do.
Depth Camera Philosophy (Critical)
Do NOT use skeleton tracking
• No OpenPose / MediaPipe
• No joints, hands, faces, gaze
Reason:
• Skeletons introduce strong human semantics
• Break visual indiscernibility
• Turn the work into a “dance interaction”
D455 is intentionally “insufficient” — and this insufficiency is used.
Data Extracted from Intel RealSense D455
Required Inputs (Minimal Set)
1. Depth Map
• 30–60 fps
• Base material for all computation
2. Foreground Segmentation
• Depth thresholding
• Static background assumption
3. Centroid (Core Parameter)
• 3D center of mass (x, y, z)
• Frame-to-frame delta → velocity vector
• Treated as abstract “root”
4. Spatial Dispersion
• Bounding box or PCA
• Width / Height / Depth
• Change rates
5. Asymmetry Measures
• Upper vs lower depth mass
• Front vs back mass
“Skeleton” Design (Abstract Constraint Structure)
Not a human skeleton.
Minimal structure:
• Root (centroid)
• Axis A (max variance direction)
• Axis B (orthogonal)
• Axis C (depth direction)
Used for:
• Deformation
• Orientation
• Non-anthropomorphic motion
Critical Rule
Artificial life and avatars share the same data structure, but:
• Artificial life parameters are internally generated and evolve
• Avatar parameters are driven by live depth input and do NOT evolve
Visitor Experience (Intended)
• Visitors move; avatar responds
• Touch artificial life → world freezes for 2 seconds
• Artificial life remains indifferent
• Sound pressure continues
• Confusion arises:
• Is this a bug?
• Is this incomplete?
• Is this intentional?
This doubt is productive and encourages searching for explanation elsewhere.
What the Work Is NOT
• Not a responsive character system
• Not a dance visualizer
• Not a participatory empowerment piece
• Not a narrative system
Interactivity is formal, not expressive.
Implementation Priorities (Recommended Order)
1. Depth capture → foreground extraction
2. Centroid + velocity
3. Spatial dispersion metrics
4. Abstract constraint visualization
5. Artificial life with identical structure
6. Shared rendering pipeline
7. Environmental freeze logic (2 sec fixed)
8. Sound pressure integration
Acceptable “Imperfections”
• Lag
• Misalignment
• Partial tracking loss
• Awkward motion
These reinforce:
• Non-human ontology
• Indifference
• Asymmetry
Guiding Principle (For All Further Decisions)
Do not ask: “Is the interaction clear?”
Ask: “Is the causality asymmetric?”
If the visitor can feel response but not control, the system is working.
Temporal and Computational Contingency
The perceptibility of this digital organism depends on computational conditions.
Its visibility, coherence, and moments of identity are not fixed outputs, but emerge under processing pressure.
When computational load increases, discontinuities and momentary clarities appear.
Under more powerful hardware conditions, these gaps may diminish, and the organism may remain in a continuous state of distortion in which identity becomes harder to perceive.
This variability is not treated as a technical limitation but as an existential condition of the work.
The organism does not exist independently of its computational environment — it survives within it.
For this reason, the work is intentionally bound to technological time.
Different hardware conditions do not merely change performance; they change the ontology of what can be seen.
