Codec Avatars Workspace
Designing how users scan, customize, and express themselves through photorealistic digital humans in VR.
Led interaction design for Codec Avatars, Meta's research-driven system for rendering photorealistic digital humans in VR. Defined how users scan, customize, and express themselves through high-fidelity representations in social environments.
Existing VR avatars traded off between fidelity and expressiveness — photorealistic renders often felt static or uncanny, while more expressive avatars sacrificed realism. For social presence in VR to feel genuine, users needed avatars that looked like them and moved with the nuance of real human expression, without a technical setup process that felt like a research demo.
I led interaction design across the avatar creation and expression pipeline — from the initial capture experience through ongoing customization. My approach centered on making a technically complex system (built on research-grade scanning and rendering) feel approachable enough for a mainstream user to complete in minutes, not a lab session.
I worked closely with research and engineering teams to translate the underlying capture technology into a guided, understandable user flow — designing how someone scans themselves, previews the result, and adjusts it before entering a shared space. I iterated on the customization interface to balance user control with the fidelity constraints of the underlying rendering system.
A user starts by scanning themselves through a guided capture flow, which builds their photorealistic avatar. They preview and customize the result, then step into a social VR environment where their avatar renders and animates in real time — expressions, movement, and presence intact — so the person they're talking to sees something that actually looks and moves like them.
Photorealistic rendering at the fidelity Codec Avatars targeted required significant compute, which shaped how much real-time customization and expression range the design could support without compromising visual quality. As a research-driven system, the underlying technology was also still evolving, which meant designing interaction patterns flexible enough to hold up as capture and rendering capabilities matured.
Before: VR social presence relied on stylized or cartoonish avatars that offered expressiveness but not realism. After: Codec Avatars gave users a photorealistic digital identity that captured genuine likeness and expression, moving VR social presence closer to face-to-face interaction.
Led interaction design for Codec Avatars, Meta's research-driven system for rendering photorealistic digital humans in VR. Defined how users scan, customize, and express themselves through high-fidelity representations in social environments — bringing research-grade avatar fidelity into a usable, approachable product experience.
As a research initiative, Codec Avatars continued evolving toward broader deployment — expanding capture accessibility beyond controlled scanning setups, and extending real-time expressiveness as rendering technology matured.