keijiro / Rcam2

Remote depth camera with iPad Pro with LiDAR sensor
The Unlicense
383 stars 40 forks source link

Rcam2

gif gif

gif gif

Rcam2 is my second attempt at a real-time volumetric AR VFX system with Unity (the first attempt is Rcam). This time I used iPad Pro with a LiDAR scanner.

The Rcam2 system consists of two software components: RcamController and RcamVisualizer. RcamController runs on an iPad device and sends a video stream and metadata (camera position, control data, etc.) to RcamVisualizer, which runs on a desktop computer and renders VFX. It uses NDI to communicate between these two components, so it doesn't require any special hardware but only a network connection

I used Rcam2 in a Boiler Room stream on 24th September 2020. You can watch the recorded video on Vimeo.

System Requirements

RcamController

RcamVisualizer

How To Run

  1. Build and run the RcamController project on the iPad device.
  2. Play the RcamVisualizer project on the desktop system.
  3. Select a controller from the dropdown list. You can hide the UI by clicking an empty area of the screen.

TIPS

FAQ

I can't find the iPad device on the dropdown list.

Try turning off the Windows firewall. It solves the problem in most cases.

I chose the right device from the dropdown list, but it's unresponsive.

If you're using a virtual network device, try turning it off. Note that WSL2 implicitly creates a virtual network device (vEthernet). You have to turn it off to establish a connection correctly.

Does it work with iPhone/iPad?

No. Rcam2 requires a LiDAR scanner. iPad Pro 2020 is the only device that supports LiDAR at the moment.

Is it possible to make it run on an iPad Pro device as a standalone app?

It's technically possible but requires lots of changes. You might have to remove some HD features (like the DoF effect) and reduce some performance-related numbers like particle capacity counts in VFX. You also have to tackle several device issues like thermal throttling, battery life, etc.

Don't expect that I do anything in this direction. The controller-visualizer design of Rcam2 is the most convenient way to avoid these problems.