Skip to content

Home

xPRIMEray
Where physics performs.
A curved-ray renderer that solves, not simulates. Every ray is integrated through a refractive field. Every pixel is classified. The validator tells you whether the render is real.
Observation precedes explanation. Plausible ≠ Correct.

Observer Disagreement — 23.8% of pixels classify differently

23.8% of pixels classify differently
Same scene. Same camera. Two transport models. Curved rays miss geometry that straight rays hit — not randomly, but directionally: 9 misses for every 1 new contact.
A straight-ray renderer would report 33% more geometry hits for this scene. The difference is not a visual effect. It is the result of solving the transport equation, not a lens-shader effect.
Ch 2 — Observer Disagreement →

Hermetic Closure — budget=32 (0% closure) vs budget=700 (100% closure)

0% closure at budget = 32
Two renders that look identical. Left panel: every pixel is unresolved budget noise. Right panel: every pixel is a real transport result. The images are indistinguishable by eye.
A render can pass visual inspection while being completely wrong. The Validation HUD is the only reliable detector. Plausible ≠ correct.
Ch 3 — Hermetic Closure →

Coherence Basin — 960×540 instability map with risk overlay

276 instability nodes, all at the same precision floor
Two symmetric horizontal bands that no integration budget can eliminate. The oracle finds them at every precision level. Their uniformity is the finding — it points to topology, not noise.
Some transport failures are structural. The field geometry creates a convergence floor the integrator cannot cross. Scheduler decorrelation reduces banding; it does not remove these zones.
Ch 4 — Coherence Basin →

Three ways to enter

Entry What you will find
xPRIMEray Runtime Curved-ray observatory and diagnostics
Portable Observatory Public ontology: Scene → Transport Lens → Plate → Evidence
Project Glowing Heart Protocol and artifact trail
Observation Atlas Observer field guide, not a ranking

Observatory Atlas is the chapter tour. Observation Atlas is the observer map.


What xPRIMEray Is

A null-geodesic integrator where curved rays are first-class primitives. Transport is solved, not faked with lens shaders or post-process distortion. GRIN and Gordon-metric fields define an effective spacetime; the renderer solves the correct eikonal transport through it.

A visual diagnostics platform. The engine finds hits, seams, high-curvature regions, unstable domains, and boundary-layer events from scene data — then reveals their structure through an expanding library of overlay modes: film overlays, heatmaps, contact sheets, domain maps, curvature contours, and transport ownership graphs.

A structured diagnostic infrastructure connecting renderer behavior to theoretical frameworks. Six-layer Cathedral Probe overlays, scheduler resonance DOE, domain telemetry, oracle reference comparison, and transport island microscopy separate independent failure layers that naive per-pixel analysis conflates.


Current Renderer State

Category Count Examples
Ready to ship 20+ systems GRIN fields, boundary layers, hit detection, overlays, 46 fixture scenes, observatory scripts
In progress 4 systems WormholePrototypeRig, TestBench recipes, atomic orbital observatory fixture
Research / diagnostic only 5 systems ReferenceTransportOracle, SceneTransportMemory, MetricHeuristicIntegrator (4 open TODOs)
Active overlays 21 modes FilmOverlay2D, step budget heatmap, domain ownership map, curvature contours, observatory contact sheets
Proposed overlays 13 modes Celestial Boundary Overlay, Bulk-to-Boundary Dual View, Curvature Domain Map, S-Matrix Event Ledger

Full Feature Index · Release Readiness Audit · Overlay Master List


Observatory & Audit Navigator

Document Purpose
Overview / Transport Observatory How all systems fit together; gallery; recommended next actions
Feature Index Master feature map with quick-reference tables
Release Readiness Audit Ship-readiness classification; cleanup list; packaging exclusions
Optical Transport Feature Map Transport system completeness; engine feature candidates; gap analysis
Overlay Master List All 34 overlay modes — existing, partial, and proposed
Inspiration Cards Thinkers and concepts linked to engine features (Pasterski, Noether, MTW, Maxwell, Feynman)

Project Glowing Heart

Project Glowing Heart is the active Core extraction and bridge track for xPRIMEray. The current public-safe state is metadata-first: Core emits standalone preview artifacts, the Godot side has static fixture exports, and the bridge has shared fixture, schema, observer, and reconciliation packets. It does not claim parity, runtime equivalence, closure equivalence, or pixel comparison readiness.

Latest bridge artifact Status
Shared fixture schema and instance Preview draft / candidate
Shared observer contract Preview contract
Core and Godot observer instances Metadata generated
Observer reconciliation Pixel comparison not ready
Public demo posture Safe with limits; no parity or validation claims

Project Glowing Heart v1.8.1 Observer Reconciliation · Shared Observer Contract · Public Demo Readiness


What This Is

  • A null-geodesic integrator — curved rays are first-class primitives. Transport is solved, not faked with lens shaders or post-process distortion.
  • A GRIN and Gordon-metric renderer — refractive-index fields define an effective spacetime; the renderer solves the correct eikonal transport through it.
  • A hermetic validation harness — every render run is classified: source hit, background hit, portal, absorbed, escaped. escaped_no_hit = 0 is the contract.
  • A multi-scene wormhole system — two causally isolated overspaces joined at a topological throat, each rendered with full curved-ray physics.
  • A transport diagnostics platform — six-layer Cathedral Probe overlays, scheduler resonance DOE, domain telemetry, oracle reference comparison, and transport island microscopy.

Current Milestone: Curved-Field Validation Ladder

Curved-field diagnostic quad panel

Four-panel diagnostic layout: rendered frame with curved transport active · hit-normal vector overlay · camera cross-section minimap showing field geometry in the vertical camera plane · transport/field overlay with ownership seams and oracle comparison context. Oracle replay failures: 0. All 64 sampled pixels sealed at step 0.02.

The curved-field validation ladder is the first end-to-end validation packet for curved ray transport in xPRIMEray. It runs the complete diagnostic stack — ReferenceTransportOracle step-size convergence, six-layer Cathedral Probe overlay, camera cross-section minimap, and curved-vs-control storyboard — against a GRIN field scene with curved transport active. The result confirms that the oracle, diagnostic infrastructure, and transport ownership machinery operate correctly under non-trivial geodesic curvature.

Measurement Value
Oracle step 0.0015625 (8× finer than production floor)
Sampled pixels 64
Oracle replay failures 0
Sealed at step 0.02 64 / 64
Mean decision-risk delta (0.00625 vs 0.003125) 0.000090
Comparability status warning — control scene differs by camera pose key

All 64 sampled pixels achieve Stable classification at the coarsest tested step, with near-zero risk delta between the two finest steps. No transport topology anomalies were identified in this scene. The comparability warning records that the baseline used the domain_resolver_stress scene while the curved run used curved_minimal_backdrop; a matched-pose control was not available at run time — the storyboard is evidence that the ladder infrastructure wires correctly under this constraint, not a head-to-head geometry comparison.


Core Capabilities

Transport

Capability Status
RK4 null-geodesic integration (eikonal ODE) ✅ Production
GRIN field sources (FieldSource3D, spatially varying n(x)) ✅ Production
Gordon effective metric framing ✅ Production
Tiered metric hierarchy (Tier 0 GRIN → Tier 3 exotic) ✅ Production
Morris-Thorne wormhole topology (causal observer ladder) ✅ Production
Dual-scene overspace composition ✅ Production
Derivative-aware adaptive step control ✅ Production

In-Game Overlays (live, no render pass required)

Overlay Toggle
Dual-reality straight-reference inset EnableDualRealityResearchMode
Curvature heatmap (5 metric modes) DualRealityOverlayMode
Semantic wireframe glyphs (portals, fields, BLVs) WireframeReferenceOverlay
Collision radar (projected AABB/sphere bounds) DualRealityCollisionRadarOverlayEnabled
Hit normal vectors + film gradient normals FilmOverlay2D
Debug ray polylines DebugOverlayOwnedByFilm
Top-down / oblique research overlay WormholeResearchOverlay

Experimental Feature Flags

  • EnableDomainTelemetry — exports per-pixel renderer diagnostics: domain_id, domain_confidence, boundary_confidence, selection_flip, normal_discontinuity.
  • EnableDomainAwareFirstHitResolver — experimental domain-aware first-hit heuristic. Off by default. Requires EnableDomainTelemetry.
  • EnableTileMetricsScaffold — tile-metrics subsystem: reorder simulation, execution, and persistent-priors scheduling.
  • EnableObjectSeededTileScheduler — tile ordering seeded from projected scene object centroids.

Validation and Diagnostics

The hermetic fixture rule enforces complete pixel classification on every run. Fixture 011 (six-checkpoint wormhole observer ladder) is the canonical validation sequence.

The bridge (post-throat backstep) is the confirmed transport anomaly: 366 segments/crossing vs. 50–153 at all other checkpoints (z-score 4.40). Three independent anomaly detectors agree. Domain-aware analysis separates three transport regimes (near-side, bridge anomaly, far-side) via PCA and k-means clustering (k=3, ARI=0.595).

The Cathedral Probe framework — six passive diagnostic layers composited over a single render — separates scheduler-induced global banding from localized topology failure. The key finding: transport instability is topological and localized, not globally smoothable. Scheduler decorrelation (tile traversal) eliminates horizontal banding; it does not eliminate local geometry seam instability. Those are two independent failure layers.

The ReferenceTransportOracle measures transport stability against a fine-step reference (0.0015625) without feeding results back into the renderer — a guardrail enforced in code, not by convention. Oracle microscopy surfaces transport topology that phase-space diagnostics miss.


Current Research Frontier

Cathedral Probe — Scheduler Resonance and Dual-Layer Transport Failure

Cathedral Probe contact sheet

Six-layer Cathedral Probe diagnostic contact sheet — domain resolver stress scene, step_length=0.015, row traversal. From left: beauty render, geometric wireframe, transport ownership map, risk probe markers, spacetime transport diagram, transport continuity vectors.

Scheduler resonance stride plot

Scheduler stride sweep (56-cell DOE). Stride 1: ~31% band coverage across all step lengths. Stride 4: < 0.7%. Traversal cadence — not physics precision — is the primary amplifier of row-global banding.

Four-mode traversal comparison

Traversal mode comparison at step_length=0.015. Scheduler decorrelation reduces banding across modes. Local corner instability persists unchanged — two independent failure layers confirmed.

Cathedral Probe architecture paper

Transport Island Microscopy — Oracle-Guided Precision Closure

Following scheduler decorrelation, a ReferenceTransportOracle ROI sweep identified a compact unresolved transport island (x=36..44, y=31..37). Dense island microscopy (289 samples) confirmed precision closure: all pixels seal at step 0.00625, with zero oracle replay failures. The island was not independently flagged by the Cathedral Probe continuity vectors — demonstrating that oracle microscopy surfaces transport topology that phase-space diagnostics miss.

Island measurement Value
ROI sweep unresolved pixels 54 / 320 comparisons (16.9%)
Island bbox x=36..44, y=31..37
Dense pass samples 289
Sealed at step 0.00625 true (289/289)
Mean decision-risk delta (0.00625 vs 0.003125) 0.000189
Oracle replay failures 0

Transport Island Microscopy paper


Architecture

xPRIMEray uses a tiered transport hierarchy: Tier 0 GRIN ray integration → Tier 1 metric parameter extraction → Tier 2 Gordon Metric bridge → Tier 3 exotic metrics. The multi-scene wormhole system joins two causally isolated overspaces at the wormhole throat. The hermetic fixture rule (escaped_no_hit = 0) enforces complete pixel classification.

Architecture overview — pipeline, stored-hit system, domain emergence, Gordon metric math → Architecture subsystems — subsystem contracts and data-flow diagrams


Research papers

Paper Description
Cathedral Probe architecture Scheduler resonance DOE, dual-layer failure model, six-layer overlay methodology
Transport Island Microscopy Oracle-guided precision closure, island identification, convergence ladders
Paper 001 — Causal Observer Ladders Six-checkpoint wormhole fixture, transport anomaly z-scores, regime clustering
Paper 004 — Hermetic Throat Validation Coverage contract, bridge anomaly evidence, throat-depth maps

Diagnostics and analysis

Document Description
Phase coherence field Per-pixel coherence scores correlating banding with domain-boundary transitions
Domain ownership analysis Transport regime decomposition, spectral ruling-out of oscillatory model
Feature maturity matrix What is live in-game vs. harness-only vs. post-process only
Visual milestone inventory Chronological archaeology of all rendered output artifacts

Reference

Document Description
Glossary Null geodesic · GRIN · Gordon metric · domain boundary · phase coherence
Hermetic fixture rule What 100% pixel classification means and why it matters
Spec index All active specifications

Repository

github.com/AetherTopologist/GD_xPRIMEray

License: MIT — academic, commercial, and creative use welcome. Citation templates: papers/shared_bibliography.bib