videocube

videocube · sources

VIDEOCUBE — the VIDEO version of the audio CUBE oscillator. It ingests THREE 60-frame video rings — A (FLOOR), B (WALL/connector), C (CEILING) — each generated LIVE from a connected video input or loaded from a .frametable.png atlas, and combines them exactly the way audio CUBE combines its three wavetables: an OCCUPANCY-WEIGHTED trilinear morph. For every output pixel it reads a colour from each ring, derives an occupancy position from the WALL luma, and blends the three by the SAME cube-dsp occ weights — MORPH cross-fades A↔C through B, CONNECT (+ CONNECT STRENGTH) reshapes/swells the interior binding, CRUSH posterizes colour depth, SPACE CRUSH mosaics the sampling grid, SPACE DIFFUSE smears the coords, MATERIAL switches soft-blend↔hard mosaic and WRAP mirror-folds — so video_out is a recognizable MORPHED IMAGE of the sources. It ALSO emits AUDIO: the same three rings are reduced to luma heightfields and scanned by the IDENTICAL cube-dsp surface-height slice the audio CUBE plays, so audio_out is a mono wavetable-oscillator drone whose timbre is driven by the SAME knobs (MORPH/CONNECT/CRUSH/SPACE/slice Y·ROT/WRAP/MATERIAL) — the picture and the sound move through one shared field structure. TUNE/FINE set pitch, LEVEL the gain, FOLD a west-coast wavefolder on the audio (audio-only). A global READER (SMOOTH/MORPH/CHAOS) reads each ring's 60-frame history, FREEZE holds the live rings and LIVE forces the real-time read. Beyond video_out + audio_out it exposes SIX dedicated slice-visualizer video jacks — SCOPE (the 256-sample surface-height wave the audio plays), SLICE (the 2-D cutting-plane cross-section), DEPTH (the intersection-depth heightmap, bright=deep/loud), and a SMOOTH/MORPH/CHAOS triptych (the cross-section forced through each reader treatment) — each a different readout of the SAME field/slice, each PER-PORT gated so an unpatched jack costs ZERO GPU, and the SLICE VIEW toggle (TEXTURED/XRAY/WEIGHTS) recolours slice+triptych. Rendered at half engine resolution with a renderer-gated tap count (the FrameTable budget ×3); the audio slice is GPU-reduced + recomputed off the audio thread on a throttle. v1 ships MONO-DRONE-first (poly/ADSR = follow-up). The occupancy combine + colour blend + luma reduction are a 1:1 CPU mirror unit-tested in $lib/video/videocube-core, reusing the audio-CUBE field math (cube-dsp) wholesale. All ports live on the yellow drill-down PATCH PANEL. Look-affecting new WebGL shader — held for owner visual preview.

VIDEOCUBE is the VIDEO version of the audio CUBE oscillator, and it works the SAME way: it stacks three video sources into a GENUINE volumetric 3D solid and lets you fly a cutting plane through it. It ingests THREE 60-frame video rings — A (FLOOR), B (WALL / connector) and C (CEILING) — each either GENERATED LIVE from a connected video input (video_a/b/c) or LOADED from a .frametable.png atlas file. Each frame the reader picks ONE frame from each ring as a SURFACE, and those three video-luma surfaces are stacked into a real 3D scalar field over (x, y, z): the depth axis z is a genuine connecting dimension, and the occupancy curve (the SAME cube-dsp occ math the audio CUBE uses) fills SOLID DENSITY between the three surfaces — so the three videos join THROUGH SPACE, exactly as three wavetables join in the audio cube. video_out is a VOLUMETRIC RAY-MARCH of that solid: every output pixel casts a camera ray and marches through the cube, compositing the occupancy-weighted colour of the three source videos (SMOOTH = a soft blend, HARD = one surface wins per voxel) with opacity set by the field density, and it draws the CUTTING SLICE PLANE (the exact plane the audio reads, tinted by the density it cuts) and a 12-edge wireframe for orientation. An orbit CAMERA (VIEW ZOOM / VIEW X / VIEW Y / VIEW Z) flies around the solid so you can look THROUGH the three videos from any angle. MORPH cross-fades the FLOOR fill toward the CEILING fill through B, CONNECT reshapes how B binds them (rounded ↔ hard ramp), CONNECT STRENGTH swells B's mid-band, CRUSH posterizes the colour and amplitude-crushes the density, SPACE CRUSH voxelizes the field lookup, SPACE DIFFUSE pulls the sampling toward the field's lowest-information face, MATERIAL switches SMOOTH↔HARD, WRAP mirror-folds out-of-range coordinates, and Y / ROT X / ROT Y / ROT Z position the cutting plane. Crucially it ALSO emits AUDIO from the SAME field: the reader-selected frame of each ring is reduced to a luma heightfield and the IDENTICAL cube-dsp surface-height slice is flown through the stack along the SAME plane (Y / ROT), so audio_out carries a mono wavetable-oscillator drone whose timbre is driven by the SAME field knobs ('isomorphic in all cases'). Pitch is set by TUNE/FINE (and the tune CV), level by LEVEL, and FOLD is a west-coast wavefolder on the derived audio (audio-only — it has no image analog); SPREAD is a FRAMETABLE-style TEMPORAL WINDOW — the sampling size, in time, of the reader: at 0 each ring surfaces a single frame (crisp), and as you open it a Hann-weighted window blends a wider span of the ring into every surface, so a FROZEN table OOZES through its captured frames (the picture AND the drone flow in lockstep — the same window feeds the audio reduce). SCAN is the FRAMETABLE MORPH partner to SPREAD: where SPREAD sets the WIDTH of the reader window, SCAN sets its POSITION — it moves the reading CENTRE through the whole 60-frame (~2-second) ring, wrapping at the seam, so on a FROZEN table you can scrub the crisp read (or the widened window) THROUGH the captured content and it oozes through the ~2 seconds exactly as FrameTable's MORPH + SPREAD do (SPREAD alone can only widen the blend at the fixed centre; SCAN is what walks it through the ring). SCAN feeds BOTH the picture surfaces and the audio reduce (image + drone in lockstep, the same as SPREAD), and at 0 it is byte-identical to the pre-scan read — the default look/sound is unchanged. A global READER selector (SMOOTH default / MORPH / CHAOS) sets which frame each ring contributes as its surface (SMOOTH = a trailing frame, MORPH = the newest frame, CHAOS = a per-pixel dithered frame), FREEZE holds all live rings so you can scrub a held window, and LIVE forces the real-time (no-lag) read. The ray-march runs at quarter engine resolution and LINEAR-upscales to the half-res video_out, with a renderer-gated step count; the audio slice is recomputed off the audio thread on a throttle, never per sample. VIDEOCUBE v1 ships MONO-DRONE-first (no poly / ADSR yet — a follow-up). All ports live on the yellow drill-down PATCH PANEL (no raw side jacks). The field, the per-voxel colour blend and the luma reduction are a 1:1 CPU mirror unit-tested in $lib/video/videocube-core, reusing the audio-CUBE field math (cube-dsp) wholesale. Look-affecting new WebGL shader — held for owner visual preview.

the faceplate

22 inputs · 8 outputs · 30 params

about this module

Hover a control, or open the patch panel and hover a jack.

inputs

id	cable	what it does
`video_a`	video	The LIVE source recorded, frame by frame, into ring A (FLOOR — one of the two morph ends / the bottom surface of the 3D solid). Ignored when slot A is loaded from a file. Unpatched (and no file) leaves ring A black. RGB video stream
`video_b`	video	The LIVE source recorded into ring B (WALL / connector — the surface that binds A and C through the middle of the solid). Ignored when slot B is loaded from a file. RGB video stream
`video_c`	video	The LIVE source recorded into ring C (CEILING — the other morph end / the top surface of the solid). Ignored when slot C is loaded from a file. RGB video stream
`morph_cv`	cv	CV that modulates MORPH (the FLOOR↔CEILING cross-fade of the 3D fill, A ↔ C through B), swept linearly over 0..1. Drives both the volume and the audio field. control voltage (CV); modulates morph_fc (additive offset — ±1 CV sweeps the full range, centered on the knob)
`connect_cv`	cv	CV that modulates CONNECT (how the WALL binds A and C — rounded soft connector ↔ hard linear ramp), swept linearly over 0..1. control voltage (CV); modulates connect (additive offset — ±1 CV sweeps the full range, centered on the knob)
`connect_strength_cv`	cv	CV that modulates CONNECT STRENGTH (over-emphasises B in the mid-band, swelling the connector through the solid / bulging the audio field), swept linearly over 0..1. control voltage (CV); modulates connect_strength (additive offset — ±1 CV sweeps the full range, centered on the knob)
`crush_cv`	cv	CV that modulates CRUSH (posterize the colour + amplitude-crush the field density AND the derived audio — the same crush levels), swept linearly over 0..1. control voltage (CV); modulates crush (additive offset — ±1 CV sweeps the full range, centered on the knob)
`space_crush_cv`	cv	CV that modulates SPACE CRUSH (voxelize the field lookup — a chunky 3D mosaic in the picture and the identical voxelization of the audio field lookup), swept linearly over 0..1. control voltage (CV); modulates space_crush (additive offset — ±1 CV sweeps the full range, centered on the knob)
`space_diffuse_cv`	cv	CV that modulates SPACE DIFFUSE (pull the sampling toward the field’s lowest-information face — a smear in the volume and the same coord warp in the audio scan), swept linearly over 0..1. control voltage (CV); modulates space_diffuse (additive offset — ±1 CV sweeps the full range, centered on the knob)
`slice_y_cv`	cv	CV that modulates Y (the cutting slice plane height through the solid — the same plane the audio surface-height scan reads), swept linearly over 0..1. control voltage (CV); modulates slice_y (additive offset — ±1 CV sweeps the full range, centered on the knob)
`slice_rx_cv`	cv	CV that modulates ROT X (Euler tilt of the cutting plane about X — the same plane the audio reads), swept linearly over -pi..pi. control voltage (CV); modulates slice_rx (additive offset — ±1 CV sweeps the full range, centered on the knob)
`slice_ry_cv`	cv	CV that modulates ROT Y (Euler tilt of the cutting plane about Y), swept linearly over -pi..pi. control voltage (CV); modulates slice_ry (additive offset — ±1 CV sweeps the full range, centered on the knob)
`slice_rz_cv`	cv	CV that modulates ROT Z (Euler tilt of the cutting plane about Z), swept linearly over -pi..pi. control voltage (CV); modulates slice_rz (additive offset — ±1 CV sweeps the full range, centered on the knob)
`view_x_cv`	cv	CV that modulates VIEW X (orbit-camera ELEVATION — the HORIZONTAL axis of the VIEW joystick pad), swept linearly over -pi..pi. Picture only — reprojects the volumetric view without touching the sound or the cutting plane. control voltage (CV); modulates view_rot_x (additive offset — ±1 CV sweeps the full range, centered on the knob)
`view_y_cv`	cv	CV that modulates VIEW Y (orbit-camera AZIMUTH — the VERTICAL axis of the VIEW joystick pad), swept linearly over -pi..pi. Picture only — orbits the volumetric view without touching the sound or the cutting plane. control voltage (CV); modulates view_rot_y (additive offset — ±1 CV sweeps the full range, centered on the knob)
`fold_cv`	cv	CV that modulates FOLD (the west-coast wavefolder on the derived audio — audio-only, no image effect), swept linearly over 0..1. control voltage (CV); modulates fold (additive offset — ±1 CV sweeps the full range, centered on the knob)
`spread_cv`	cv	CV that modulates SPREAD (the temporal reader window width — how far a frozen table oozes through time), swept linearly over 0..1. control voltage (CV); modulates spread (additive offset — ±1 CV sweeps the full range, centered on the knob)
`scan_cv`	cv	CV that modulates SCAN (the reader-centre POSITION — moves the reading centre through the 60-frame ring / scrubs a frozen table through its ~2 seconds), swept linearly over 0..1 (wraps at the ring seam). Drives both the picture surfaces and the audio reduce. control voltage (CV); modulates scan (additive offset — ±1 CV sweeps the full range, centered on the knob)
`tune_cv`	cv	CV that modulates TUNE (the derived oscillator pitch in semitones), swept linearly over -36..36. Affects the audio only, not the picture. control voltage (CV); modulates tune (additive offset — ±1 CV sweeps the full range, centered on the knob)
`chroma_depth_cv`	cv	CV that modulates CHROMA DEPTH (the overall colour→sound intensity of the CHROMASTACK hue morph), swept linearly over 0..1. Audio-only — the picture already shows full colour; this drives how much the colour reshapes the DRONE. 0 = the pure luma carrier (today’s sound). control voltage (CV); modulates chroma_depth (additive offset — ±1 CV sweeps the full range, centered on the knob)
`motion_cv`	cv	CV that modulates MOTION (how much frame-to-frame picture change adds "alive" energy/loudness to the derived audio), swept linearly over 0..1. Audio-only. 0 = pure content (a static colourful frame still sounds rich). control voltage (CV); modulates motion (additive offset — ±1 CV sweeps the full range, centered on the knob)
`hue_mode_cv`	cv	CV/GATE that toggles the HUE MODE bank (the colour→timbre character): low = MUSICAL (tonal), high = INSTRUMENT (analog↔digital). Swept linearly and rounded, so it acts as a gate. Audio-only. control voltage (CV); modulates hue_mode (additive offset — ±1 CV sweeps the full range, centered on the knob)

outputs

id	cable	what it does
`video_out`	video	The PRIMARY output: a VOLUMETRIC RAY-MARCH of the 3D occupancy solid, textured by the three source videos joined across a real depth axis, under an orbit camera — you look THROUGH the three videos. Includes the cutting slice plane (the plane the audio reads) and a cube wireframe. The card preview shows this output; rendered at quarter engine resolution and LINEAR-upscaled to half res. RGB video stream
`audio_out`	audio	The derived audio: the reader-selected frame of each ring reduced to a luma heightfield and scanned by the audio-CUBE surface-height slice (the SAME field, the SAME plane as the picture), played as a mono wavetable oscillator at TUNE/FINE pitch and LEVEL gain. A continuous drone whose timbre is shaped by every field knob; silent until the audio worklet stands up (a moment after spawn). audio signal
`scope_out`	video	A SCOPE TRACE of the exact 256-sample surface-height wave audio_out plays (cube-dsp.sampleSlice of the cutting plane, after FOLD) — the literal picture of the sound. A dedicated small video readout; PER-PORT gated (renders nothing until patched). Because it visualises the audio-derived wave, it is flat until the audio worklet stands up and the rings fill (the same warm-up as audio_out). Route it into a video mixer/monitor to SEE the drone move as you fly the slice. RGB video stream
`slice_out`	video	The 2-D CROSS-SECTION where the cutting plane cuts the 3-D solid: for each pixel a point (a,b) on the tilted plane (positioned by Y / ROT exactly as the audio's slice ray) reads the three ring surfaces + the cube-dsp occupancy, coloured by the SLICE VIEW flavour — TEXTURED (the occupancy-weighted source videos on the plane), XRAY (grayscale occupancy density) or WEIGHTS (false-colour R=floor / G=ceiling / B=wall occupancy shares). Y and the rotations slide/tilt the cut through the solid; it honours the card's READER mode. PER-PORT gated. RGB video stream
`depth_out`	video	The intersection-depth HEIGHTMAP: for every plane point (a,b) a ray marches the solid along the plane normal and accumulates how much material it crosses (the 2-D extension of the audio's 1-D surface-height scan), colour-mapped so BRIGHT = DEEP = LOUD. The picture of the audio's loudness field across the whole plane. The z-march step count is renderer-gated (SwiftShader-safe). PER-PORT gated. RGB video stream
`smooth_out`	video	The slice cross-section (as slice_out, honouring SLICE VIEW) forced through the SMOOTH reader treatment — a trailing frame of each ring — regardless of the card's READER selector. Routed alongside morph_out + chaos_out it lets all three temporal treatments be compared/mixed at once. PER-PORT gated. RGB video stream
`morph_out`	video	The slice cross-section forced through the MORPH reader treatment — the newest (crisp) frame of each ring — regardless of the card's READER selector. Differs from smooth_out (which trails) under motion. Honours SLICE VIEW. PER-PORT gated. RGB video stream
`chaos_out`	video	The slice cross-section forced through the CHAOS reader treatment — a per-pixel dithered frame across each ring's 60-frame window — regardless of the card's READER selector. The noisiest of the triptych; differs from smooth_out/morph_out under motion. Honours SLICE VIEW. PER-PORT gated. RGB video stream

params

id	label	range	default	curve
`tune`	tune	-36..36st	—	linear
`fine`	fine	-100..100¢	—	linear
`morph_fc`	morph	0..1	—	linear
`connect`	connect	0..1	—	linear
`connect_strength`	cnct str	0..1	—	linear
`crush`	crush	0..1	—	linear
`space_crush`	space crush	0..1	—	linear
`space_diffuse`	space diffuse	0..1	—	linear
`fold`	fold	0..1	—	linear
`spread`	spread	0..1	—	linear
`scan`	scan	0..1	—	linear
`chroma_depth`	chroma	0..1	—	linear
`motion`	motion	0..1	—	linear
`slice_y`	y	0..1	—	linear
`slice_rx`	rot x	-3.1416..3.1416	—	linear
`slice_ry`	rot y	-3.1416..3.1416	—	linear
`slice_rz`	rot z	-3.1416..3.1416	—	linear
`level`	level	0..2	—	linear
`view_zoom`	view zoom	0.3..3	—	linear
`view_rot_x`	view x	-3.1416..3.1416	—	linear
`view_rot_y`	view y	-3.1416..3.1416	—	linear
`view_rot_z`	view z	-3.1416..3.1416	—	linear
`wrap`	wrap	0..1	—	discrete
`material`	material	0..1	—	discrete
`screen_on`	ray-march	0..1	—	discrete
`reader_mode`	reader	0..2	—	discrete
`freeze`	freeze	0..1	—	discrete
`live`	live	0..1	—	discrete
`slice_view`	slice view	0..2	—	discrete
`hue_mode`	hue mode	0..1	—	discrete

controls

control	what it does
chroma	CHROMA DEPTH (0..1, default 0.6): the CHROMASTACK colour→sound intensity — how much the picture’s COLOUR reshapes the derived drone on top of the luma cube carrier. The reduced RGB is read as a hue histogram that blends 8 band-limited harmonic archetypes (red = dark/hollow → violet = bright/rich) with a hue-centroid spectral tilt, so a colour change AUDIBLY sweeps the timbre. 0 = the pure luma carrier (byte-identical to the old, static sound); a GRAYSCALE picture always collapses to that carrier regardless of this knob (the clean fallback). Audio-only. CV via the chroma depth input.
connect	CONNECT (0..1, default 0): reshapes how the WALL (B) binds A and C through the solid interior — 0 = a rounded/soft connector (circle occupancy), 1 = a hard linear ramp toward B (sawtooth-V). Same occ profile drives the picture and the audio. CV via the connect input.
cnct str	CONNECT STRENGTH (0..1, default 0): over-emphasises B's connector in the mid-band of the solid (B swells through the volume / bulges the audio field). 0 = the exact CONNECT shape. CV via the connect strength input.
crush	CRUSH (0..1, default 0): posterize the ray-march colour (RGB quantized) AND amplitude-crush the field density (opacity) and the derived audio (the same crush levels). 0 = clean. CV via the crush input.
fine	FINE (-100..100 cents, default 0): fine pitch trim of the derived audio between the semitone steps of TUNE. Audio-only.
fold	FOLD (0..1, default 0): a west-coast wavefolder applied to the derived audio waveform (adds harmonics). Audio-only — it has no image analog. CV via the fold input.
freeze	FREEZE (0/1, default 0): stops all LIVE rings from advancing so the held surfaces can be scrubbed with the slice/view controls. File-loaded rings are always frozen. Picture + (via the ring content) audio.
hue mode	HUE MODE (0/1, default 0 = MUSICAL): which colour→timbre CHARACTER bank drives the chroma morph. MUSICAL = tonal harmonic archetypes (red = root/dark/hollow → violet = bright/rich); INSTRUMENT = warm colours sound analog-ish, cool colours digital-ish. A front-panel toggle; CV-GATED (a gate high selects INSTRUMENT). Audio-only.
level	LEVEL (0..2, default 1): output gain on the derived audio (applied after FOLD). 1 = unity. Audio-only.
live	LIVE (0/1, default 0): forces the real-time / no-lag ring read (the newest frame) for the SMOOTH reader (MORPH is already newest; CHAOS keeps its per-pixel dither). Applies to BOTH the picture and the audio surface selection.
material	MATERIAL (0/1, default 0 = SMOOTH): SMOOTH renders the solid translucent, blending the three videos by their occupancy weights; HARD renders a binary solid where one surface wins per voxel (a hard-cut mosaic), and the same in the audio field.
morph	MORPH (0..1, default 0): cross-fades the FLOOR fill (surface A) toward the CEILING fill (surface C) of the 3D solid through the WALL (surface B) — 0 biases toward the floor, 1 toward the ceiling. Drives both the volume and the audio field. CV via the morph input.
motion	MOTION (0..1, default 0): how much frame-to-frame picture CHANGE adds "alive" energy/loudness to the derived audio — still = quiet, moving = alive. BLENDED with content, not a replacement: a static but colourful frame still sounds rich, and MOTION 0 is pure content. Audio-only. CV via the motion input.
reader	READER (0..2, default 0 = SMOOTH): which frame each ring contributes as its surface (global — all three rings), in BOTH the picture AND the derived audio — they read the SAME temporal frame (the unified-field promise). 0 = SMOOTH (a trailing frame, sub-frame smoothed), 1 = MORPH (the newest frame, crisp), 2 = CHAOS (a per-pixel dithered frame across the ring window). Because CHAOS is per-pixel in the picture, the 1-D audio scan reads its window-MEAN representative frame.
scan	SCAN (0..1, default 0): the reader-centre POSITION — the FrameTable MORPH partner to SPREAD. It moves the reading CENTRE through the whole 60-frame (~2-second) ring: 0 = today’s per-mode centre (byte-identical to the pre-scan read), and turning it up shifts the centre scan·(N−1) frames back, wrapping at the ring seam. Where SPREAD widens the blend at a fixed centre, SCAN walks that centre (crisp or widened) THROUGH the ring — so a FROZEN table can be SCRUBBED through its captured frames and oozes through the ~2 seconds exactly like FrameTable’s MORPH. Drives BOTH the picture surfaces and the audio reduce (image + drone in lockstep). For CHAOS it shifts the per-pixel dither base. CV via the scan input.
ray-march	SCREEN (0/1, default 1 = on): perf gate for the ray-march. When off AND video_out is unpatched the render is skipped (the rings keep capturing and the audio drone keeps running).
rot x	ROT X (-pi..pi, default 0): Euler tilt of the cutting plane about X — the same plane the audio reads and the render draws. CV via the rot x input.
rot y	ROT Y (-pi..pi, default 0): Euler tilt of the cutting plane about Y. CV via the rot y input.
rot z	ROT Z (-pi..pi, default 0): Euler tilt of the cutting plane about Z. CV via the rot z input.
slice view	SLICE VIEW (0..2, default 0 = TEXTURED): the colorize flavour for slice_out AND the smooth/morph/chaos triptych (it does NOT change video_out, the audio, or the field). 0 = TEXTURED (the occupancy-weighted source videos where the plane cuts solid), 1 = XRAY (grayscale occupancy density), 2 = WEIGHTS (false-colour occupancy shares — R = floor-fill, G = ceiling-fill, B = wall/connector). Picture-only; no CV.
y	Y (0..1, default 0.5): the height of the CUTTING SLICE PLANE through the solid. It is the exact plane the audio surface-height scan reads AND the plane drawn (tinted) in the volumetric render. CV via the y input.
space crush	SPACE CRUSH (0..1, default 0): voxelize the field lookup — snaps the (x,y,z) sampling coordinates to a chunky 3D voxel grid (a blocky solid) and the identical voxelization of the audio field lookup. 0 = transparent. CV via the space crush input.
space diffuse	SPACE DIFFUSE (0..1, default 0): pulls the field sampling toward the cube’s lowest-information face (computed from the field, latched on the field, matching the audio scan) — a smear/gravity in the volume and the same coord warp in the audio. 0 = off. CV via the space diffuse input.
spread	SPREAD (0..1, default 0): the TEMPORAL sampling window WIDTH of the reader (FrameTable-style). At 0 each ring surfaces one crisp frame (byte-identical to the pre-window read); opening it Hann-averages a ±window of ring frames into every surface — a wider window blends more of the ring, so a FROZEN table oozes through time. The reader mode (SMOOTH lag / MORPH newest) + SCAN pick the window CENTRE; SPREAD sets its WIDTH. Feeds BOTH the picture surfaces and the audio reduce, so image and drone flow together. CHAOS keeps its per-pixel single frame. CV via the spread input.
tune	TUNE (-36..36 st, default 0): coarse pitch of the derived audio oscillator in semitones. CV via the tune input. Audio-only.
Videocube a file input {n}	FLOOR table - loads a .frametable.png atlas (a lossless 10x6 = 60-tile contact sheet, the format FRAMETABLE saves) into ring A, replacing live capture and holding the ring frozen so the loaded content can be scanned with SCAN and SPREAD. Session-only in v1: nothing is written to disk or to the patch, so a reload returns the slot to LIVE.
Videocube a live {n}	FLOOR live - points ring A (the FLOOR surface, one of the two morph ends) back at whatever is patched into video_a, resuming live frame-by-frame capture. This is the way back from a loaded table: a slot stays on its file until you press this. Unpatched, the ring fills with black.
Videocube b file input {n}	WALL table - loads a .frametable.png atlas into ring B, the connector surface. Session-only in v1.
Videocube b live {n}	WALL live - points ring B (the WALL / connector surface that binds FLOOR and CEILING through the middle of the solid) back at video_b for live capture. The same reset as FLOOR live, on the ring that decides how the other two join through space.
Videocube c file input {n}	CEIL table - loads a .frametable.png atlas into ring C, the ceiling surface. Session-only in v1.
Videocube c live {n}	CEIL live - points ring C (the CEILING surface, the far morph end) back at video_c for live capture. The same reset as the other two.
view x	VIEW X (-pi..pi, default 0.6): orbit camera ELEVATION. Picture only — rotates the volumetric view up/down without touching the sound or the cutting plane.
view y	VIEW Y (-pi..pi, default 0.7): orbit camera AZIMUTH. Picture only — rotates the volumetric view left/right.
view z	VIEW Z (-pi..pi, default 0): orbit camera ROLL (a genuine roll of the view). Picture only.
view zoom	VIEW ZOOM (0.3..3, default 1): the orbit camera distance (camera dist = 2.6 / zoom). Picture only — a deliberate divergence from audio CUBE’s viz-only view: here it shapes the OUTPUT frame.
wrap	WRAP (0/1, default 0): OFF reads each source video with a single clamped sample (the plain look). ON extends the sampling domain and mirror-folds it, so the videos MIRROR-TILE across the cube — a kaleidoscopic fold at the faces/mid-planes in the volume, and the matching mirror-fold of the audio slice samples that reach beyond the cube. Governs BOTH the picture and the derived audio.

source

videocube.ts on GitHub.