Merge feed from 3+ cameras into single display/pointcloud

[SDay2]

We'd like to use a series of Realsense cameras in parallel to capture tracking info in real-time for a large area.

I've come across several solutions for merging point clouds together, but they seem to be things that are for recordings you've made and then exported. I've also found a whitepaper on it (https://realsense.intel.com/wp-content/uploads/sites/63/Multiple_Camera_WhitePaper04.pdf) but cannot find the demo shown in the PDF. The solution in the paper used for multicam also does not address overlapping camera FOV. Since we're covering a large area we'd have a lot of overlap happening.

So my question is, is it possible to retrieve the data the cameras are seeing into one single stream in real time?

[MartyG · Accepted solution]

If you are referring to the test app in section G of the paper (the one involving a high speed LED panel), it is unfortunately not available. This paper originated as a document only able to Intel's non disclosure agreement (NDA) customers. After the paper was released to the general public, Intel confirmed that the test app would not be made available but hoped to integrate it into the RealSense SDK software's' RealSense Viewer' program someday.

In January 2018 at the Sundance festival, Intel demonstrated a 'volumetric capture' system where four D435 cameras, each attached to its own PC, captured people in real-time and sent all the data from the four PCs to a fifth PC to combine the point clouds and apply post-processing.

https://realsense.intel.com/intel-realsense-volumetric-capture/

More recently, Intel greatly expanded upon this concept in a seminar presentation that is available as a YouTube video.

https://www.youtube.com/watch?v=VSHDyUXSNqY&t=843s