Nemosyne is an experimental framework for investigating immersive data visualization. We're testing a hypothesis: does embodied, navigable 3D space improve data comprehension compared to 2D? Help us find the answer.
This interactive VR scene demonstrates our experimental spatial data framework. We're testing whether the Datumplane mapping (X=relationships, Y=hierarchy, Z=time) improves comprehension compared to traditional 2D layouts.
Research Question: Does embodied navigation through 3D data space improve recall and pattern recognition? See our methodology โ
Create immersive data experiences through spatial computing for deeper understanding and intuitive manipulation.
GOALS
Bind real-world datasets to interactive 3D objects. Manipulate complex data through natural hand gestures and spatial interactions.
Stream live data into VR environments. Watch as your data updates in real-time, responding to changes in the physical world.
Multi-user VR environments where teams can simultaneously interact with shared data artefacts, regardless of physical location.
Built on open web standards. Run in modern OSes & browsers without plugins. Compatible with Linux, Mac & Windows; Quest, HoloLens, and mobile AR.
Extensible architecture with reusable artefact components. Build custom data representations or use pre-built templates.
Connect to REST APIs, GraphQL, WebSockets, or IoT streams. Visualize everything from financial data to sensor networks.
Create interactive VR objects with declarative syntax. Bind data, define interactions, and render in minutes.
Nemosyne makes design assumptions that need validation. We're treating the framework as a platform for answering seven critical research questions.
Question: Does embodied, navigable 3D space produce better data comprehension than 2D alternatives, and for which task types?
Hypothesis: 3D advantages emerge for topology-reading tasks (finding bridges between clusters), not value-reading tasks (precise comparisons).
Question: Are the X/Y/Z axis assignments (relationships/hierarchy/time) congruent with human spatial intuition, or arbitrary?
Experiment: Present the same dataset in multiple axis configurations and measure comprehension speed and accuracy.
Question: Can we reliably distinguish structural-level topology (scale-free networks, continuous fields) from schema-level patterns?
Challenge: Extend constraint programming approaches (Draco system) to the 3D design space.
Question: Does encoding information in a self-constructed, navigable VR space improve recall compared to 2D representations?
Hypothesis: Self-generated spatial contexts produce stronger encoding than imposed ones (generation effect + contextual reinstatement).
Question: When multiple users navigate the same data space simultaneously, how do shared reference frames and private viewpoints interact?
Problem: In VR, users can have literally inverted spatial reference frames. "Look at this" becomes ambiguous.
Question: At what point does node density in 3D space cause the visualization to become illegible regardless of frame rate?
Focus: Effective field of relevant perception, depth as a data channel (vergence-accommodation conflict), clutter thresholds.
Question: When is a 3D form a better encoding than a colour channel or a label? What conditions make arbitrary shapes read as meaningful icons?
Framework: Apply semiotics (Peirce's icon/index/symbol) and affordance theory (Gibson, Norman) to 3D glyph design.
These domains represent our hypotheses for where immersive data visualization may provide advantages. We're seeking collaborators to help validate (or refute) these use cases.
Monitor factory floor sensors as interactive 3D models. Spot anomalies by observing colour and scale changes in real-time.
Navigate market data as spatial landscapes. Walk through trading volumes and manipulate time-series in 3D space.
Handle molecular structures or astronomical data at scale. Collaborate with remote teams in shared virtual labs.
Manipulate MRI and CT scans as volumetric artefacts. Annotate and discuss cases with specialists worldwide.
Urban data visualizationโtraffic flows, energy grids, and population density as manipulable city models.
Interactive learning artefacts. Students manipulate historical artifacts or mathematical concepts in VR.