Skip to content

Research Division · Dream Team

Dream Engine

Visualize the overnight signal.

Turn decoded sleep dynamics into sensory-safe 3D environments — a research narrative layer for Dream Team, not a mind reader.

Open previewDocsRoadmap

Motion field

Synthetic demot=0 · N1
Autonomic ridge
0 / 119
Magnitude0.36
Turbulence0.12
Attractor0.02
Direction

(0.00, 0.09, 1.00)

Simulated multi-modal sleep signals → features → motion parameters → vector field. Not a reconstruction of private mental imagery. Hover a region for local vector + features.

Product

Neural 3D storytelling for sleep-stage research

Dream Engine sits at the end of the Dream Team stack: after decode and optional localization, it projects features into continuous axes, builds a portable scene graph, and renders inspectable environments for collaborators.

  • Sensory-safe by default

    Intensity, motion, and chroma are clamped. Flash is forbidden. Research narrative — not entertainment deepfakes.

  • Portable scene graphs

    JSON-first SceneGraph v1 travels from builder to Three.js (and future Babylon.js) without locking the stack.

  • Neurodecode handoff

    Bridge latents.npy into DreamAxes, then export interlace.dream-team.demo-bundle/v1 for lab demos.

Viewer

Subconscious motion field

Interactive reconstruction demo: scrub time, gate modalities, and inspect attractor basins derived from synthetic sleep-state signals.

Motion field

Synthetic demot=0 · N1
Autonomic ridge
0 / 119
0.75
0.55
0.45
Magnitude0.36
Turbulence0.12
Attractor0.02
Direction

(0.00, 0.09, 1.00)

Simulated multi-modal sleep signals → features → motion parameters → vector field. Not a reconstruction of private mental imagery. Hover a region for local vector + features.

Pipeline

Signals to motion field

Subconscious Motion Field Reconstruction — a transparent, step-wise method from multi-modal sleep proxies to an inspectable vector field. Research demo, not a black box.

motion-field · client synthetic

  1. Step 01

    Signals

    Synthetic EEG-like bands (δ θ α β), HRV index, breath cadence, and sleep stage over discrete time

  2. Step 02

    Features

    F(t): band powers, HRV volatility, breath irregularity, stage encoding — pure extraction

  3. Step 03

    Motion params

    Interpretable map → direction, magnitude, turbulence, attractor strength (modality-gated)

  4. Step 04

    Motion field

    Vector field on a grid + attractor basins (REM cluster, deep sleep, autonomic ridge)

  5. Step 05

    Viewer

    Time scrub, modality toggles, intensity sliders — inspect local vectors and attractors

flow · Signals → Features F(t) → MotionParameters → Field(grid, attractors) → Viewer