“3D Capturing using Multi-Camera Rigs, Real-time Depth Estimation and Depth-based Content Creation for Multi-view and Light-field Auto-Stereoscopic Displays” by Kovacs and Zilly
Conference:
Type(s):
Entry Number: 01
Title:
- 3D Capturing using Multi-Camera Rigs, Real-time Depth Estimation and Depth-based Content Creation for Multi-view and Light-field Auto-Stereoscopic Displays
Presenter(s):
Description:
The wide variety of commercially available and emerging 3D displays – such as stereoscopic, multi-view and light-field displays – makes content creation challenging as each displays technology requires a different number of views available of the scene. As consequence, the content creation pipelines differ considerably and involve different camera setups such as beamsplitter rigs with small baselines and high quality cameras used for stereo 3D productions or camera arrays for auto-stereoscopic displays which usually use small lower quality cameras in a side-by-side arrangement. Converting content shot for a specific display technology into a different format usually impairs the image quality and is very labor-intensive.
Against this background a generic method for capturing and rendering live 3D footage for stereoscopic, multi-view and lightfield displays is presented. The system consists of a widebaseline multi-camera rig, a camera assistance system, a realtime depth estimator, a real-time view generation and rendering engine, and multiple displays, one multi-view auto-stereoscopic and a light-field display. The system features several innovative components: the professional-grade multi-camera assistance and calibration system; a real-time depth estimator producing convincing depth maps; a real-time and generic depth-image based rendering (DIBR) engine that is suitable for generating imagery for a range of 3D displays; and the largest autostereoscopic light-field display to date.
Acknowledgements:
The demonstrated system is based on work partially supported by the MUSCADE European FP7 project (EU-FP7-247010).
