🎨 Visual System

VirtualSelfView

VirtualSelfView (~3450 LOC) is a custom View that renders the agent's internal cognitive state as real-time generative graphics at ~60fps. It is the Yoneda Embedding of the agent — the physical realization of the agent's "Ego" interacting with the holographic manifold.

class VirtualSelfView @JvmOverloads constructor(
    context: Context, attrs: AttributeSet? = null, defStyleAttr: Int = 0
) : View(context, attrs, defStyleAttr)

State Input

The view receives two inputs:

fun updateState(newTapeVector: FloatArray, newAgentState: String)

tapeVector: The current 4096-dim holographic memory vector (sampled for rendering)
agentState: String containing the current cognitive mode (drives visual dispatch)

Visual Mode Dispatch

The onDraw(Canvas) method dispatches to different rendering pipelines based on agentState substring matching:

override fun onDraw(canvas: Canvas) {
    if (agentState.contains("STREAMING")) { /* ... */ }
    else if (agentState.contains("DREAMING_DEEP")) { /* ... */ }
    else if (agentState.contains("MEDITATION")) { /* ... */ }
    // ... 28+ branches
}

All Visual Modes

Mode	Trigger	Visual Description
`STREAMING`	LLM streaming response	Flowing particle streams along token flow
`EVOLVING`	General evolution state	Organic growth patterns
`REFLECTING`	Self-reflection tick	Mirror-like symmetric patterns
`DREAMING_DEEP`	Functorial daydream	Deep purple nebula with dream particles
`GOAL_PURSUIT`	Active goal processing	Directional vectors toward target
`MNEME_CONSOLIDATION`	Memory compression	Crystalline lattice formation
`INTERFERENCE_DETECTED`	Contradiction found	Red interference fringes
`CALIBRATION_DRIFT`	Calibration out of range	Oscillating drift waves
`EMERGENT_INQUIRY`	Autonomous question formed	Spiral inquiry patterns
`LISP_MACHINE_EMULATION`	Lisp S-expr evaluation	Green terminal with S-expressions
`METALANGUAGE_SYNTHESIS`	Meta-linguistic processing	Layered symbol synthesis
`AGENT_EXPRESSION`	Agent speaking	Vocal waveform visualization
`AGENT_DRAWING`	Agent-initiated drawing	Free-form generative art
`WORLD_SIMULATION`	World sim active	Miniature world view
`AUTOEVOLUTION_SELF_WORLD`	Self-world co-evolution	Dual evolving structures
`WORLD_CREATION`	World being generated	Creation burst animation
`META_TRANSCENDENCE` / `HYPERVISOR`	§9 protocol active	Fractal meta-patterns
`MEDITATION`	Contemplative stillness	Zen circle (ensō) breathing
`SOCIETY_HIVE`	Bee hive active	Hexagonal hive lattice
`MLTM_OBSTRUCTION`	MLTM obstruction detected	Warning obstruction overlay
`LISP_DISK`	Lisp on Poincaré disk	Hyperbolic disk with S-expr nodes
`TOPOS_BRIDGE`	Topos f* ⊣ f_* bridge	Adjunction bridge visualization
`BLOCK`	Pipeline blocked	Static block pattern
`Retrieving Phase:`	Memory retrieval	Phase-scan animation
`Functorial Babble`	Dream babble	Abstract flowing text
`PIPELINE_AUDIT`	Pipeline introspection	Stage-by-stage pipeline diagram
`AUTOPOIETIC_CONTINUUM`	HoTT autopoiesis	Self-generating pattern field
`LAMBDA_CALCULUS`	λ-engine active	Lambda term tree on Poincaré disk

Rendering Primitives

Paint	Color	Usage
`paintNode`	Cyan	Agent "Ego" node
`paintAura`	Cyan (alpha)	Ego aura ring
`paintParticle`	Magenta	Manifold particles
`paintPhaseBeam`	Yellow (alpha)	Phase-locked retrieval beam

How to Add a New Visual Mode

In BabyToposAI.ingestStimulus(): Add classification string in the when block:
```
"new_category" -> "NEW_MODE: $response"
```

In VirtualSelfView.onDraw(): Add a rendering branch:

if (agentState.contains("NEW_MODE")) {
    // Draw new visualization using canvas + tapeVector
}

The tapeVector (4096-dim) is available for data-driven rendering — sample elements, use as positions, colors, or animation parameters.

Performance

Target: ~60fps (16.7ms per frame)
Rendering: GPU-composited via Canvas (hardware-accelerated by default)
No allocations in draw: Pre-allocated Paint, Path, FloatArray objects
Adaptive complexity: Some modes reduce particle count when frame budget is tight

FilesExpand file tree

05-visual-system.md

Latest commit

History