FluidDiagrams: Web-Based Information Visualisation using JavaScript and WebGL

Much attention has been focused on the provision of information graphics and visualisations inside a web browser. Currently available infovis toolkits produce graphical output by either injecting SVG nodes into the DOM or using the JavaScript Canvas 2D API. FluidDiagrams is a prototype information visualisation framework written in JavaScript which uses the WebGL 3D JavaScript API for its output, falling back to Canvas 2D as necessary, via the Three.js library. Six visualisations are currently implemented: bar chart and line chart, scatter plot and parallel coordinates for multidimensional data, and cone tree and hyperbolic for hierarchies. Anecdotally, visualisations using SVG nodes in the DOM for output can become rather sluggish when displaying more than a few dozen items. Visualisations using Canvas 2D exhibit similarly slow performance. WebGL utilises hardware acceleration where available and promises much better performance for complex visualisations, potentially in the order of many thousands of items without becoming unresponsive. A comparison of parallel coordinates visualisations with 100 records in 20 dimensions compared three implementations: FluidDiagrams (WebGL), FluidDiagrams (Canvas 2D), and D3 (using SVG nodes). They achieved 62, 6, and 10 frames per second respectively. The FluidDiagrams (WebGL) implementation was able to render 1,000 records in 20 dimensions at 18 frames per second, compared to 1 and 6 respectively.