ALTN\ISION / SPATIAL.CRITIC
[ SELECT WORKSPACE ]

What are you here to do?

Choose one focused workspace. You can return here when the session is finished.

No study is currently in progress.

ANONYMOUS WAYFINDING STUDY
ALTN\ISION / SPATIAL.CRITIC
SYNTHETIC SCENE BASELINE v0.8.1
[ RESEARCH PROTOTYPE 001 ]

Can an architectural scene
critique itself?

A transparent navigation baseline that compares the shortest route with a risk-aware alternative. This first build measures path length, turns, decision points and ambiguity exposure — before any claim about human experience.

Run the simulation ↓ RULE-BASED · REPRODUCIBLE · NOT HUMAN-VALIDATED
[ 01 / TEST ENVIRONMENT ]

Reference floor

Compare two 16×12 spatial conditions at a fixed 1.5 m grid scale, then test a metric shortest path against authored ambiguity costs.

FRAGMENTED TEST · 1.5 M/CELL
Shortest Selected Ambiguity
Synthetic architectural navigation scene A floor plan showing walls, decision points, ambiguity zones and the selected route.
FIXED CONTROLS ROUTE READY
PATH LENGTH — grid steps
EFFICIENCY — shortest / selected
DIRECTION CHANGES — 90° turns
AMBIGUITY EXPOSURE — unitless baseline
[ 02 / PROCEDURAL ENVIRONMENT ]

Walk the test scene

A dependency-free ray-cast environment generated from the same occupancy grid. Movement is collision constrained; only anonymous position, heading and elapsed time are recorded.

HUMAN PILOT MODE

Three tasks. One anonymous evidence package.

Participation is voluntary. Stop at any time. No identity, camera, microphone or device data is collected. A separate practice route is not recorded; formal task order is counterbalanced. The study configuration is assigned automatically and remains fixed after practice begins.

01 / Participant P-0000
02 / Consent
03 / Progress
READY · 0 / 3 Awaiting consent
Your browser does not support the canvas-based procedural environment.
READY
X 1.50 Y 1.50 θ 0°
[ 03 / TRACE ANALYZER ]

Compare human routes

Import anonymous session exports from the same task. The analyzer overlays participant routes, measures shared cell use and flags repeated pauses or route splits.

WAITING FOR TRACE FILES 00 VALID SESSIONS
PARTICIPANT EVIDENCE LAYER
Routes Shared use Pause Split
Imported participant route comparison A floor plan overlaying anonymous participant routes, shared-use heat cells, hesitation locations and route divergence points.
NO EVIDENCE LOADED Export sessions above, then import their JSON files here.
Session Duration Travel Changes Samples
Import trace files to compare sessions.
[ 04 / METHOD ]

What the model does

  1. 01

    Parse space

    Convert walls, walkable cells, destinations and decision zones into a navigable graph.

  2. 02

    Run baselines

    Compute a pure shortest path and a weighted path that penalizes predefined ambiguity.

  3. 03

    Expose trade-offs

    Compare steps, direction changes, decision points and accumulated spatial risk.

  4. 04

    Prepare validation

    Preserve the same metrics for a later comparison with human routes on the same tasks.

[ 05 / RESEARCH CONTRACT ]

A prototype should state what it cannot prove.

SUPPORTED NOW
  • Routes are generated from explicit, inspectable costs.
  • Every task is reproducible from the same scene data.
  • Metrics distinguish distance from ambiguity exposure.
NOT YET SUPPORTED
  • The model does not predict emotion or comfort.
  • Risk zones are authored assumptions, not observed behavior.
  • No claim is made about agreement with human navigation.
[ NEXT BUILD ]

From synthetic trial to human evidence.

Run a small consent-based pilot in the procedural scene, inspect the exported traces, then decide whether a public GLB adds research value.

GRID BASELINE → FIRST-PERSON RUNNER → TRACE ANALYZER → PILOT PROTOCOL → PILOT DATA → COMPARE / REFINE