← Learn

Spatial Web Glossary

Software 4 Senses LLC · Last updated

3D Spatial Web Development has a vocabulary of its own, because the flat web's nouns — page, div, scroll bar — stop describing what is actually on screen once a document becomes a place. This glossary collects the field's coined terms in one line each, grouped as category, structure, motion, surface, and measurement vocabulary. The structure, motion, and surface terms belong to the Scene Document Model, 3D elements given document behavior; longer treatments live on the linked definition pages.

The reference implementation grounding these definitions is the Software 4 Senses 3D Spatial Web Studio (below, the Studio). The Studio's tutorial, How to build a scene, uses some of these terms under other names: it calls a slab a card (it also calls the flat plane a picture is printed on a card) and the plane rig a grid.

The category

Six terms name the field itself: the architecture, the model it is built on, the practice, the functional/decorative divide, and the failure pattern the divide was coined to escape. Each links to its definition page.

TermDefinition
3D Spatial Web Interface A web application architecture in which the elements of a real-time 3D scene carry the structure of a document: they form an addressable element tree, the browser's own layout engine can solve their layout, input reaches them as events on that tree, the scene serializes and restores as a document, and the finished product deploys as a self-contained static page.
3D Spatial Web Development The practice of designing and building 3D Spatial Web Interfaces — composing the interface from 3D scene elements, laying them out with CSS, giving them behavior with animation clips, state machines, and events, and publishing the result as a static web page, using the web platform's own layout and animation machinery throughout.
Functional 3D A 3D interface whose elements behave as document content — addressable by id and label, positionable by layout the browser's own CSS solves, reached by input the scene delivers to them as messages, and carrying document state that persists; a document with a camera, not a video with a scroll bar.
Decorative 3D A 3D visual that carries no live document content: it is looked at, not used — its text is pixels baked into textures and its controls are pictures of controls.
The Scroll-Triggered Canvas Trap The dominant web-3D pattern in which a canvas plays a scroll-driven animation of a loaded model and nothing more — functionally a heavy, non-interactive video with no text elements, operable controls, or navigation inside it.
Scene Document Model The model a 3D Spatial Web Interface is built on — 3D elements behaving like a document: solids, lights, groups, and cameras carry a parent/child element tree, events that bubble along it, stable addresses, per-element state, and serialization that round-trips.

Structure

Structural terms name what a spatial site is made of: the content volumes, the machines that arrange them, the places content stands, and the document form the whole thing saves as. The load-bearing idea is that the slab replaces the div. All of them are Scene Document Model nouns — rigs position 3D volumes the way CSS containers position boxes.

TermDefinition
slabThe atomic content-bearing volume — the div's next life: a box that can carry a solid body and a frame of catalog material, and that poses whatever scene element is parented to it; with none of them it is a seat that holds a place in the layout.
rigA 3D arrangement machine that positions slabs in a volume — the 3D analog of a CSS flex or grid container, driven by the author's real CSS via the measurement twin.
measurement twinThe offscreen container of stand-in boxes a rig builds so the author's own CSS can solve slab layout: scene elements have no DOM box of their own, so the browser lays out an empty stand-in carrying each one's declared size instead, and each solved rectangle becomes a slab's world pose. It is a layout instrument only — neither the platform's Shadow DOM nor a hidden state store; state for 3D elements lives with the elements themselves, whose definitions the scene document saves.
the four rig typesPlane (flat page baseline), shelf (rows at distinct depths, near means featured), ring (slabs on an arc, a browse position), and stack (a sequence along z, front is current); a rig animates between types rather than snapping.
cellThe place a slab sits, as opposed to the slab itself; cells never move and never get renamed, so an address like 3:4 stays valid whatever shape the rig is in.
occupantThe content standing in a cell; occupants move between cells while cells stay put — the other half of the two-namespace addressing rule.
bandA whole addressed group of cells — "row 2", the front row of a shelf, the facing arc of a ring — taken as one ask and one re-layout.
labelAn authored, human-readable name on any scene element; it saves with the scene, and lookups accept it in place of the element's id, though a state-machine binding matches only an exact id or '@self'. Labels need not be unique — a duplicate gets a which-one question, never a guess.
stageOne continuous camera-space experience — the spatial equivalent of a page.
sectionA snap point along a stage's scroll rail holding one viewport-full of intent — the spatial equivalent of a <section> element.
propNon-content scenery — lights, ambient geometry, set dressing; never task-critical.
compositeA named, parameterized multi-part structure built from primitives; it serializes as its name and parameters alone, and only declared parameters survive mounting — which is exactly what makes that pair a sufficient saved form.
archetypeA reusable 3D element type: a shape plus its named clips plus a state machine. Same shape, different semantics — a ball and an orb are both spheres; their clips and machines make them different things.
glass UI controlAn archetype from the reference implementation's control library — among them button, switch, checkbox, radio, slider, text field, label, and progress bar — drawn in the scene; each operable one keeps its state in a native, focusable form element placed right after the canvas, so Tab reaches it in authored order.
scene documentThe saved form of a 3D scene: reconstitutable definitions — elements, named clips, declarative state machines — kept deliberately separate from reported live state such as clip progress, so a saved scene reloads as a document, not a snapshot.
world unitThe unitless scene distance, pinned at 50 CSS pixels per world unit, so a 200-pixel card is four units wide.

Motion & travel

Motion terms all hang off one decision — the visitor keeps a one-dimensional scroll input, and everything spatial is authored choreography driven by it. One animation system — browser-computed easing driving engine records of every kind — moves every scene element through the same named clips.

TermDefinition
rails, not WASDThe founding input decision: the visitor never pilots — they keep the 1D scroll they already know, and the world does the moving; the camera is on rails, and the rails are the design.
railThe authored scroll path through a stage and the single source of truth for where the visitor is; scroll moves the visitor — or the world — along it rather than moving a page.
rail cursorThe single scalar s, in rail units of roughly one CSS pixel of wheel travel, of which every camera pose and every bound world translation is a pure function.
detentA snap point on a rail; hard detents always settle, soft ones capture nearby motion, and a "floor" is a detent on a vertical camera span.
scroll grammarA named genre of forward motion that a rail span adopts — the taxonomy names eight: Conveyor, Dolly Rail, Elevator, Turntable, Focus Pull, Page Turn, Semantic Zoom, and Orbit Vignettes.
analog-firstThe rule that every change animates from its current value toward its target — nothing teleports; for visitors who prefer reduced motion, the rail's autonomous moves and layout re-arrangements land as cuts with identical end states.
passageA typed, reversible, interruptible transition between two stages, designed to replace the page-load blink; a taxonomy of eight kinds, from Cut through Shared-Element Morph to Environment Fade.
parkPresent in the scene but deliberately not taking part — not drawn, not lit, not clickable; temporary by definition and never saved with a scene.

Status: the eight scroll grammars and the passage taxonomy are named design vocabulary, not a shipped catalog. The Studio's shipped rail layer carries the Conveyor as working choreography today, along with Elevator mechanics (hard detents on a vertical camera span) and Dolly Rail mechanics (camera spans); the scroll grammars page gives each grammar's status. Passages remain an open build phase. Stage and prop, in the structure table, are design vocabulary as well, with no record type of their own.

Surfaces

Surface terms describe how scene elements meet light and how materials are kept physically honest; the material terms, fade to air and honest approximation, govern every scene element.

TermDefinition
selfLitA 0-to-1 dial for how much a picture's own pixels glow through scene lighting: 0 is fully scene-lit, 1 shows the picture's pixels verbatim.
backdropThe page the scene sits on, as light: the scene's declared page colour or gradient, saved with the scene — the light a reflected or refracted ray picks up when it leaves the scene, so metals reflect it and glass refracts it.
fade to airThe vanish ramp: opacity and index of refraction driven to nothing together, so an object genuinely ceases to matter to light — opacity zero alone is honest clear glass, still refracting.
honest approximationA material-catalog preset badged with the name of the measured property this renderer cannot draw — for example subsurface glow — rather than rendering it silently wrong.

Measurement

Measurement terms come from the discipline the Studio's engine work is graded by: no number without its context, no timing comparison before image identity, and negative results stay on permanent record.

TermDefinition
offers/rayCandidate offers per ray — one iteration of the per-child traversal loop — the census unit for grading traversal structures such as a scene-wide BVH, because ray-marching cost scales with rays times candidates offered per ray.
noise floorA control run with the identical configuration in every timing slot, measured on the same machine the same day; any effect inside that band is ruled a null.
pixel-identity gateByte-for-byte image equality required before any timing comparison — a faster arm that renders differently is doing less work, not the same work faster.
rig (of a measurement)The full context a performance number is meaningless without: scene, pinned camera pose, dispatch resolution, and whether the run was a headless probe or the live product surface.
the graveyardThe standing practice of keeping measured negative results — nulls, refutations, reverted builds — on permanent record in the experiment files themselves, so dead ideas are never re-derived.

Related reading