Image Based Reconstruction and BRDF Estimation

DA

Christoph Traxler, Martin Knecht, Michael Wimmer

Content:

Description

The aim of the RESHADE project is to simulate the mutual shading effects between real and virtual objects in mixed reality applications. For that the real part of the scene must be virtualised with sufficient accuracy. That means the geometry of real objects must be captured and the BRDF of their surfaces must be estimated. Image based techniques have been published for both tasks. The goal of this project is a combination of these image based methods so that reconstruction and BRDF estimation can be done concurrently and in near real-time.

Task

The image based reconstruction method presented by Newcombe, et.al. Live Dense Reconstruction with a Single Moving Camera should be implemented and combined with a near real-time BRDF estimation.

Requirements

The students should be interested in some of the following topics: Computer vision, image based techniques, real time rendering, mixed reality.

Environment

We've already developed a C# based mixed reality framework that uses DirectX 10 for rendering (via SlimDX). We integrated PTAM into our system, which is also the basis of the reconstruction method presented by Newcombe et.al. Our framework renders scenes using real-time global illumination based on instant radiosity.

Additional Information

Contact: traxler {ät} cg . tuwien . ac . at