Holoportation turned a body into a live 3D stream

↗

What the 2016 paper put on record

In March 2016, researchers at Microsoft published Holoportation: Virtual 3D Teleportation in Real-Time, presented that year at the ACM Symposium on User Interface Software and Technology. The abstract states plainly what the system does: high-quality 3D models of people are reconstructed, compressed and transmitted anywhere in the world in real time, then viewed through a mixed-reality headset such as HoloLens so a remote person appears to occupy the viewer's own space. That is the claim worth holding onto: this is a capture-and-transmission pipeline married to a see-through headset display, not a projector, a panel or a laser system.

What creates the image, and what it needs

Microsoft's continuing project page, Holoportation, describes the capture side as an array of depth-and-colour sensors, now built around Azure Kinect units, arranged to cover a person from multiple angles so no side of them is a flat cutout. That multi-camera coverage is what a single webcam cannot supply: it is the difference between a flat image with a depth guess and an actual reconstructed surface a viewer can walk around. On the display side, the viewer needs a mixed-reality headset capable of registering the model in their own room, not a screen; a person without one cannot see the effect at all, a constraint any comparison with ordinary video calling has to state up front.

The bandwidth question, and what is and is not documented

The same project page reports that when the system was adapted for use inside a moving vehicle, the team achieved a ninety-seven per cent reduction in bandwidth requirements while holding quality, and lists a telemedicine collaboration with surgeons in Glasgow and integration into Microsoft Mesh as later applications. Those are separate, later deployments rather than details of the original 2016 demonstration, and the page does not restate what the 2016 system's own bandwidth cost was, so the two figures should not be read as a single before-and-after claim. The published abstract likewise does not state a camera count or a specific data rate, so this entry does not invent one.

Questions to bring to a demo

• How many capture cameras cover the volume, and what happens to the reconstruction at its edges.
• Does viewing require a headset for every participant, or can some join by flat video.
• What data rate does a live session actually sustain outside a research network.

Holoportation is best read as an early, well-documented answer to one narrow question: can a body be reconstructed and streamed in real time for a headset viewer. It is not, on this record, evidence about cost, comfort or bandwidth at ordinary broadband scale.