Scaling up beyond 24 cameras: Usb hubs vs Raspberry Pis

To scale up your bullet-time or photogrammetry scanners beyond 24 cameras, the best option is to use a Raspberry Pi architecture (4 cameras per Pi). In any case, the main software would still run on a Windows based computer.
Full documentation about Raspberry Pi client installation:
Here are the advantages:
  • Faster download time: by using fewer cameras per unit (typically, 4 cameras on a single pi), you get much faster download time to bring your images from the cameras to the main server. Check out the benchmark page for a few insights:
  • More stable architecture: the OS on the Pi runs on the memory. That means that you can always cold reboot the pis without risking any corruption on the microSD cards.
  • No configuration needed: the OS is installed through the provided image (card flashing is made directly through our software). On Windows, you'll always be stuck with updates, manual configuration, and unexpected behaviors. In other words, Linux is made to be operated in the background, while Windows is more about user interaction.
  • Proven solution: this is what we and all of our clients use when we go beyond 12 or 24 cameras. This has been used in a lot of various photogrammetry and bullet-time installations up to 220 cameras.
What about 24 cameras or less?
We still use the Pis sometimes for installations between 12 and 24 cameras, but only when the priority is about speed optimization. There are a few seconds that can be gained when using Raspberry Pis instead of USB hubs: