On Apple's ARKit 3 page (https://developer.apple.com/augmented-reality/arkit/) there is fine print reading, "People Occlusion and the use of motion capture, simultaneous front and back camera, and multiple face tracking are supported on devices with A12/A12X Bionic chips, ANE, and TrueDepth Camera."
I read this sentence as a computer scientist, which suggests that the device must have an A12, ANE, and a TrueDepth camera to support any of "people occlusion, motion capture, simultaneous front/back camera, and multiple face tracking." If this is the case, then the only iPad that can use any of these features is the latest iPad Pro, and not an Air, which have an A12, but not a TrueDepth camera. (Sidenote: what is ANE? I can't find documentation on it, but I think it has something to do with the machine learning system.) Is this correct--that only the iPad Pro supports any of these features?
I ask because people occlusion is incredibly important for a multi-user experience around a table.
The ANE is the Apple Neural Engine.
Apple's footnote is a little misleading. People occlusion is performed on the scene captured by the rear camera and so does not make use of the TrueDepth camera.
Multiple face tracking requires a TrueDepth camera, in addition to an A12 chip with ANE.
People occlusion is available on all A12 devices - see this WWDC session so the new iPad Air will support it.