You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Interactive Graphics, Data Visualization, 3D and Spatial Roadmap
Goal: make research tangible through direct manipulation. Every canvas is an enhancement over meaningful HTML/SVG content and follows an explicit quality tier.
ID
Deliverable
Effort
Depends on
Status
IF1
Graphics feasibility spike: persistent singleton canvas versus isolated lazy islands; record ADR and memory profile
M
UI6, IT6
✅
IF2
Capability-gated 3D hero model with orbit/pointer response, pause/reset controls, static SVG fallback and context recovery
Lower DPR/geometry, no post-processing, event-driven rendering, short transitions
Full
capable device and visitor preference
60 fps target, adaptive DPR, enhanced lighting/effects, continuous render only while visible
Acceptance criteria
IF1–IF2 — renderer and first 3D experience
Compare memory, route persistence, context count, first interaction cost and chunk size before choosing the renderer lifecycle.
The 3D chunk loads after intent/visibility, never blocks hero copy or CTAs, and reserves its layout space.
Pointer drag and arrow keys offer equivalent model control; reset/pause buttons have accessible names and visible focus.
Handle webglcontextlost/webglcontextrestored; dispose geometry, material, texture, listeners and animation frames on teardown.
Clamp device pixel ratio and suspend rendering when hidden/offscreen. Record fps and heap/VRAM proxy measurements.
IF3–IF4 — data interaction
Data and visual encodings are separate typed modules; invalid/empty datasets return a descriptive state.
Nodes/series are reachable by keyboard, selection is announced, color is never the only category encoding, and an equivalent textual list/table is present.
Filtering/highlighting maintains 60 fps for the documented reference dataset and completes within 100 ms on the main thread.
URLs can link to a selected entity and the visualization does not overwrite browser navigation semantics.
IF5–IF7 — spatial and audio stories
Assets use glTF/GLB with mesh/texture compression where supported; budgets and licenses are stored beside the asset manifest.
Annotations remain DOM content with collision-aware positioning and a non-spatial index.
Panorama controls constrain pitch, avoid motion sickness, and provide discrete “previous/next hotspot” navigation.
Audio starts only after a user gesture, exposes playback state, releases AudioContext resources, and never uploads local media.
IF8–IF12 — advanced effects
Each effect is individually feature-flagged and removable without affecting content flow.
WebGPU code checks adapter/device limits instead of assuming buffer sizes; compilation is warmed during idle time with visible progress.
Experimental XR/splat features are opt-in and excluded from default route payloads.
IF13 — component catalogue
Every catalogue component documents its input contract, data source, accessibility equivalent, loading behavior, and whether it owns a canvas/audio/video resource.
Components are independently importable so a future route can lazy-load only the required domain chunk; there is no growing components/interactive catch-all.
R2 research-derived backlog
The research report turns the catalogue into evidence-led packages: RR2 fleet playback, RR3 solver/Pareto comparison, RR4 ML replay/model cards, RR5 audio explanation, RR6 media/reading timeline, RR7 game prototype, RR8 360 room, RR9 optional WebGPU aggregation, and RR10 worker replay protocol. Each package must ship a semantic/list fallback, provenance, reduced-motion mode, and performance measurements before it can be marked Done.
Document history
Date
Revision
Change
2026-08-08
R2
Added research-derived interaction packages and evidence gates.