Demonstrations

We had a couple of interesting demonstrations at CVMP this year.

Lightmap demonstrated their HDRLightstudio in the Atrium this year.

Raytrix will be demonstrating their latest light field camera in the Atrium.
Raytix presented the R11 lightfield camera system that has up to 6 fps live depth calculation and refocusing capability. The camera is based on an 11 megapixel CCD sensor and the raw lightfield images are processed on a GPU using CUDA resulting in images with up to 3 megapixel resolution. The 3D capability of the system is presented with an autostereoscopic display.

i3DLive - Project Results Demo

The TSB/ESPRC project i3Dlive "interactive 3D methods for live-action media" developed methods to facilitate 3D content production and manipulation of video content during post-production. A key area of research during this project was to investigate how a set of static witness cameras can be used to enable stereoscopic content production from the viewpoint of a monoscopic principal camera, thereby eliminating the need for a dedicated stereo camera rig and providing added flexibility. The University of Surrey will demonstrate several advances made towards this aim, such as:

• A tool-kit of algorithms for through-the-lens calibration of a moving/zooming principal camera from a set of pre-calibrated static witness cameras. The calibration process involves the construction of a sparse scene model from the static cameras, with respect to which the moving camera is registered, by applying the appropriate perspective-n-point (PnP) solver.

• A layered-depth estimation approach which can produce a high-quality view-dependent scene representation that can be utilised to automatically convert a monocular input into a stereoscopic stream. The approach is based on a state-of-the-art multi-view graph-cut reconstruction algorithm suitable for use with a small number of witness cameras separated by a wide-baseline.

• Manipulation of material appearance and lighting in video sequences. This combines automated methods for albedo estimation in a video sequence with interactive relighting techniques to facilitate generation of perceptually realistic edited video sequences in post-production.

The capabilities of the presented tools are demonstrated on a variety of datasets produced during the course of the project.