next up previous contents
Next: Fixed plane approximation Up: Lightfield modeling and rendering Previous: Lightfield modeling and rendering   Contents

Rendering from recorded images

Our goal is to overcome the problems described in the last section by relaxing the restrictions imposed by the regular lightfield structure and to render views directly from the calibrated sequence of recorded images with use of local depth maps. Without loosing performance the original images are directly mapped onto one or more planes viewed by a virtual camera.

To obtain a high-quality image-based scene representation, we need many views from a scene from many directions. For this, we can record an extended image sequence moving the camera in a zigzag like manner. The camera can cross its own moving path several times or at least gets close to it. To obtain a good quality structure-and-motion estimation from this type of sequence it is important to use the extensions proposed in Section 5.2 to match close views that are not predecessors or successors in the image stream. To allow to construct the local geometrical approximation depth maps should also be computed as described in the previous section.


next up previous contents
Next: Fixed plane approximation Up: Lightfield modeling and rendering Previous: Lightfield modeling and rendering   Contents
Marc Pollefeys 2002-11-22