A 106-camera rig captures people as 3D mesh files

↗

What Microsoft's documentation puts on record

Microsoft's own developer documentation for HoloLens design, Designing Holograms, retrieved 16 September 2026, states that Microsoft operates a Mixed Reality Capture studio in San Francisco and describes footage from 106 cameras in that studio. The same page names the studios Microsoft has licensed the technology to: Avatar Dimension in Washington D.C., Metastage in Los Angeles, Dimension Studios in London, SK Telecom in Seoul, and Volucap in Berlin. That is a specific, dated-as-retrieved account of the capture volume's scale and of who else operates the same method, rather than a marketing summary.

What creates the captured figure

A ring of over a hundred cameras surrounding a subject can reconstruct that person's actual surface from every angle at once, frame by frame, rather than inferring a back or a side that no single camera saw. The same documentation states what comes out of that process: a keyframed mesh, normals and texture, deliverable either as OBJ and PNG files for further editing or as a ready-to-play H.264-compressed MP4 file with an embedded mesh, importable into Unity, Unreal, native and WebXR projects and playable on Windows, iOS, Mac, Android, Magic Leap and PlayStation VR. That is a genuine textured three-dimensional asset of a real, filmed performance, not a rigged character animated afterward, which is the key difference between this method and the Codec Avatar approach used elsewhere for personalised faces.

What the studio is and is not for

This is a production pipeline, not a live communications system: a subject is filmed inside a fixed volume built for the purpose, and the output is a file, edited and shipped later, rather than a real-time stream to a remote viewer. One of Microsoft's own licensees, Dimension Studio, now advertises this kind of capture as part of a wider virtual-production business spanning television and film work; its current site does not restate the camera count or file formats, so those technical specifics rest on Microsoft's documentation alone among the sources checked here.

Questions to bring to a demo

• Is the capture volume fixed, and what is the largest performance area it covers.
• Is the deliverable a live stream or an edited file, and how long does processing take.
• Which playback platforms and file formats does a given studio actually support today.

Read together, the documentation and a licensee's current business describe a mature, if specialised, method for turning a real performance into a reusable three-dimensional asset. That is a narrower and more concrete claim than the phrase volumetric hologram sometimes attached to this work in coverage, and the studio's own numbers are worth citing precisely because they are rarely repeated as exactly as Microsoft states them.