On the desk / platform watch 2026
Roblox gives avatar faces a makeup drawer
Brows, lashes, lips and face paint become separate choices. The interesting bit is how they combine.
People, performance & virtual identity
A new face.
A real person.
New faces, virtual stages, and the people making them work. Follow the creative details, the platform changes, and the occasional very good rabbit hole.
Open the dispatches ↗100source-backed stories
in the working archive
in the working archive
Fresh on the desk.
Explore dispatches →
research 2026
A nod has more moving parts than you'd think
One research team tested the timing, speed and shape of an avatar's nod. Looking attentive is still different from understanding a conversation.
Read the story
research 2026
Who does an AI teammate choose to hear?
An audit crosses avatar style and voice presentation to examine how models make social choices from synthetic character cues.
Read the story
research 2026
A convincing face is not a validated experiment
A 2026 theory paper proposes a way to study generated identity cues. Its most useful boundary is the one between a demonstration and evidence.
Read the storyChoose a thread
Before the latest update.
Explore dispatches →
research 2023
VTuber self-presentation is built with a character and a crowd
A qualitative study of Chinese-speaking VTubers treats the avatar as a medium for identity work, performance and gender expression. Its interview and observation boundaries matter.
Read the story
platform watch 2023
Warudo Made a Node-Based Stage for 3D VTubers
Warudo’s July 2023 Early Access release combined tracking, scenes, audience interactions and programmable blueprints in one 3D performance tool.
Read the story
platform watch 2023
Teams Avatars Made Camera-Off Presence Visible
Microsoft’s May 2023 release brought customizable animated stand-ins into eligible Teams meetings.
Read the storyInto the details.
Explore research →research 2026
A nod has more moving parts than you'd think
One research team tested the timing, speed and shape of an avatar's nod. Looking attentive is still different from understanding a conversation.
Read the storyresearch 2026
Who does an AI teammate choose to hear?
An audit crosses avatar style and voice presentation to examine how models make social choices from synthetic character cues.
Read the storyresearch 2026
A convincing face is not a validated experiment
A 2026 theory paper proposes a way to study generated identity cues. Its most useful boundary is the one between a demonstration and evidence.
Read the story