UNREleased project | march 2016

bIG dATA vr

Big Data VR.png
 
 

project description

A simple data visualization tool using virtual reality, that reads .xml files within a specific format and builds an interactive graph. The project ceased development after successfully developing a working prototype for Oculus Rift DK2.

bIG dATA

The past few years have highlighted our ever growing need for large-scale data analysis. data processing softwares have become more popular as companies worldwide adapt to a more data-centric approach on topics such as understanding consumer behavior and increasing the efficiency of production processes.

however, big data graphs are tipically difficult to read by non-experts and graph visualization tools are often little intuitive to the end user. This project was built on the idea of developing a more naturalistic way to visualize and interact with such graphs.

user perspective after expanding a node’s connections

user perspective after expanding a node’s connections


Main features

  • a different way to Analyze data. step into the graph to naturally perceive patterns. follow your intuition by taking a time to freely explore the graph.

  • multiple navigation modes. an Overview mode which shows a static render of the big data graph in its entirety, while A navigation mode allows users to interact with smaller portions of the graph.

  • Read every information attributed to a selected node. when navigating in vr, you can get more information on a node by looking at it and presssing a button on the controller.

technical aspects

The project was developed for oculus rift dk1 using Unreal Engine 4.11. In the working prototype, The .xml file was manually formatted to fit a custom struct in unreal engine, which generated a standartized data table. the graph visualization mainly consists in a heavily optimized blueprint that reads this data table and renders 3d imposter sprites scattered through the playable area. each node in the graph contains an id, name, xyz position, text information and edges (its connections to other nodes). the blueprint then generates persistent “traces” for each one of the edges in the graph.

in the working prototype, we successfully parsed a graph with over 30,000 nodes and 100,000 connections, which was statically rendered for the user. in navigation mode (runtime), we managed to maintain performance in vr while showing up to 5,000 nodes. therefore, the user experience was limited to interacting with smaller portions of the graph. future efforts in this project would have been directed to optimizing the c++ code for parsing the data and increasing visual optimization to maintain performance on larger data sets.

In february 2016, a prototype was developed which FULFILLED the needs for basic graph visualization in the oculus rift dk2.

IN March 2016, development of big data VR ceased due to changes in the year’s schedule.

a node with expanded connections

a node with expanded connections