Project:Interactive Real-Time VolumeRendering in VR

Professor Dr. Renato Pajarola

Visualization and Multimedia Lab

Department of Informatics

University of Zurich

1

Introduction

In the current age of big data, notonly methods like machine learning helpus to understand generated or collecteddata. Approaches that focus on visualrepresentation support users in detectingpatterns and getting an overall intuition oftheir data. Volume rendering, as used inscientific visualization, comes, of course,with its specific challenges.

On one side, there are the visualiza-tion related questions. How do we needto display data to allow a domain expertusing our tool to gain new insights andhelp drawing the right conclusions aboutthe data? What are appropriate inter-action methods with the data? On theother side, there are technical limits andengineering challenges. How can we seam-lessly show gigabytes of data in fractionsof a second? How do we need to (pre-)process this data to be prepared to reactto user input in real-time?

VR googles with interaction tools. Source:theverge.com.

Head-mounted Virtual Reality (VR)displays offer a very intuitive way of in-specting spatial data. The head can sim-ply be moved instead of using mouse orkeyboard to direct the gaze to a different

spot of the data. In addition, two trackedinteraction handles with buttons can becarried by the user. This set up allows fora complete focus on the data itself andspending all the attention on visual datainspection. Also it opens up new possibil-ities of interaction, for instance to definearbitrary cutting planes through a volume.Imagine a meteorologist virtually standingin the middle of a hurricane or a physi-cian inside a patient having a look at a3D CT image for which these interactiontechniques can be very beneficial.

Pressure data set of a hurricane with around134 million data points visualized by a volumerenderer.

When it comes to volume rendering,we are facing a computationally heavytask. So we must carefully think aboutwhat to spend the CPU/GPU time on,especially because the rendering shouldbe real-time and interactive. There aremathematical technical methods such asdata compression that can be exploited aswell as cleverly engineered data structuresand implementations, that e.g. can mini-mize expensive data transfers to the GPUor that omit small details, which humansare unlikely to perceive or are not relevant

for the particular problem.

Assignments

The goal of this project is to imple-ment a tool for interactive 3D volumeexploration using VR goggles. Besidesrendering a volume, interaction tools suchas moving cutting planes, changing trans-fer functions, filtering, or simple selectionof volumes are to be implemented. Exten-sions beyond the basic visualization andinteraction can include time-dependentvolumes, e.g. to observe a hurricane de-veloping over time, or using really bigdatasets which do not fit into memoryin raw format or are just too big to beprocessed in real-time.

Project Type

This project can be a bachelor thesisor master thesis through adjusting thetasks and goals.

Requirements

A solid Computer Graphics backgroundis required. Advanced Computer Graphicsskills are an asset. C++ and OpenGL willbe used heavily, good knowledge and/orhigh motivation to learn them is required.

Work Load

• 20% Theory• 70% Implementation• 10% Test

Supervision

• Prof. Dr. Renato Pajarola• Pascal Forny (assistant)

Contact

Write an Email to forny@ifi.uzh.ch tohave a chat if you are interested and wouldlike to know more about this project.