Use multiple Intel Realsense D435 cameras to create volumetric capture of people

[KD3]

Intel created a demo where they used 4 of the new Intel D435 depth cameras to create 4D volumetric capture:

https://www.youtube.com/watch?v=9oQDz_cfUlo https://www.youtube.com/watch?v=9oQDz_cfUlo

They did it by making custom modifications to the Realsense repository code, and they also used Unity for a little work on post-processing. I believe they stitched the four pointclouds into a common 3D space, and used some kind of calibration method.

Does anyone know how this can be achieved? I'm looking for someone who can repeat the process. From what I've heard, Intel got this to work for the demo in a short amount of time.

[MartyG]

I can't speak for whatever custom adjustments they made to the RealSense Viewer in this demo. I did grab a still-frame of the Viewer's settings from the video though, enhanced it and compared it to the actual Viewer to see some of the changes they made to the settings. These changes seem to involve:

1. Expanding open the RGB Camera' section of settings

2. De-selecting the 'Enable Auto Exposure' setting.

3. Setting manual values for settings such as brightness, contrast, exposure and gain.