Microsoft Mesh promised photoreal selves, shipped avatars

↗

What the March 2021 announcement promised

On 2 March 2021 Microsoft announced Microsoft Mesh, describing it as an Azure-based mixed-reality platform for shared holographic experiences across HoloLens 2, a range of virtual reality headsets, smartphones, tablets and PCs. Technical Fellow Alex Kipman is quoted saying people would be able to feel like they are in the same place as someone sharing content. The announcement placed Mesh explicitly in the lineage of Microsoft's own Holoportation research, and its stated roadmap went further than avatars: it described a future in which participants could project themselves as their most lifelike, photorealistic selves, a capture-based aim closer to Holoportation than to a cartoon avatar.

What creates the shared-space effect today

Microsoft's current support documentation for immersive spaces, Get started with immersive events in Microsoft Teams, retrieved 16 September 2026, describes a narrower mechanism than that roadmap: attendees join as avatars that can be personalised in the Teams Avatars app, positioned in a three-dimensional scene so that directionality and distance let separate small-group conversations happen in the same virtual room, viewed on the Teams app for PC and Mac or a Teams Immersive app on Meta Quest 3. That is an avatar-driven positioning and spatial-audio system, not a system that captures a participant's actual face and body live. The gap between the two documents is the point: the 2021 language reached for photorealistic capture, and the shipped feature is a customisable figure placed in a shared scene.

Why an avatar meeting is not telepresence

This distinction matters for a field guide that treats presence claims carefully. A telepresence system, in the sense used elsewhere in this archive, generally captures a real, current image of a person, whether by matched-room video, light-field display or volumetric capture, and reproduces it live. An avatar-based platform substitutes a pre-built or customised model, driven by tracked motion or a microphone, for that live image. Both can support a sense of co-presence and directional conversation, but only one of them lets a viewer see what the other person actually looks like at that moment, and Mesh's shipped avatar system is the latter category, whatever its 2021 announcement aspired to.

Questions to bring to a demo

• Does the platform show a live capture of the participant's face, or a driven avatar model.
• Which devices are required to see the shared three-dimensional space, and what does a participant on a flat screen see instead.
• How much of the 2021 photorealistic-capture roadmap has actually shipped, as of the date the claim is made.

Mesh is a legitimate example of shared spatial computing, but its own history is a caution against reading a platform announcement as a description of what ships. The avatar meetings available today are one specific, more modest thing, and this archive treats them as such rather than as telepresence by another name.