Hardware Architecture Question

I’ve read many posts that folks throw a mini-pc (like a NUC) at the kinect, then run cat5/6 into routers which allows their main PC to be located anywhere with network access.

Question: Is there any benefit in buying a slightly bigger computer with a dedicated graphics card to replace the NUC for this task? Any extra processing that could be done by this secondary input-computer and remove overhead either processing or network load?

Or is the only pro, that this larger input-computer, placed near the projectors, could also be utilized for output too?

It looks like for Body Tracking the Azure Kinect recommends a host pc with GTX1070, so maybe NUCs and mini-pc’s won’t be a viable future alternative unless they can be used as a data pass-through to a larger computer…

However the specs do mention that lower-end pc’s might still work…

" Body tracking host PC hardware requirements

The body tracking PC host requirement is more stringent than the general PC host requirement. Recommended minimum Body Tracking SDK configuration for Windows is:

  • Seventh Gen Intel® CoreTM i5 Processor (Quad Core 2.4 GHz or faster)
  • 4 GB Memory
  • NVIDIA GEFORCE GTX 1070 or better
  • Dedicated USB3 port

The recommended minimum configuration assumes K4A_DEPTH_MODE_NFOV_UNBINNED depth mode at 30fps tracking 5 people. Lower end or older CPUs and NVIDIA GPUs may also work depending on your use-case."

Interesting Intel RealSense Camera whitepaper. They list stats on how many camera’s max out their NUC + quality/resolution + bandwidth. USB hubs pretty much max out after 1.