Aims

The ideal workflow would allow us to take existing scans that are in a relaxed A-Pose and be able to automatically rig them through Blender's Mecanim system. Unity would output them in a way that they can be identified and played by an AR app and the unique model is triggered by a custom marker. This could potentially be a custom QR code, which would also correspond to a custom URL (depending on which app it is viewed with).

The models should have the ability to be 're-posed' through the app and/or be able to play a series of animations such as 'idle', run, jump etc.

While Mixamo is good for quick demonstration purposes, it carries a fairly hefty price tag per model if you are looking at anything above the free test examples, plus there is the overhead of having to run each model through an additional online service.

Mecanim will hopefully allow for automated rigging if we have a common A-pose to start with.

Vuforia, that provides the AR library, is also a paid for service though could be used in the short term as it seems robust. It provides the advantages of being able to create custom markers and also use 3D objects as markers.

For the final output we need to consider how do we create a custom marker and link it to an individual's scan?