Multiple Cameras - Rear Projection

The calibration of multiple cameras goes in several steps.

Aligning the Cameras

The cameras need to be aligned so that they cover the whole screen area. Further, the camera images need to overlap by at least 20cm, so that the computer vision layer can fuse information from multiple cameras.

Adjusting the Configuration File

The configuration file config.txt needs to be tweaked to match your camera configuration. For each camera you need to create a separate Camera section in the file. The sections are usually nearly identical, with the only difference being in the devuid, output-offset, and output-size fields.

The configuration file editing starts off with the same way as for multiple Cells](multiple-cells).

The only difference is that in this case the output coordinates of the cameras need to overlap.

Mirrors

In some cases one can be using mirrors to shorten the distance between the cameras (and projectors) and the screen. This is entirely ok, as the software can flip the output coordinates both horizontally and vertically:

Flipping the output coordinates does not incur any performance penalty, as it is performed as part of the output geometry corrections.

Portrait-Mode Cameras

Cameras can also be set into portrait mode (as opposed to the more conventional landscape mode). In this case needs to use the output rotation feature:

Depending on the rotation (and possible mirrors) one may also need to use the vertical/horizontal flipping mentioned above.