Portable identity
People should not lose their avatar when a platform changes direction.
VRMDN is built around portability, self-hostable options, and formats people can carry between products instead of renting identity from a closed stack.
Open VRM avatar infrastructure
Portable avatars for teams building VR on Quest today and a wider XR ecosystem tomorrow.
VRMDN is an open effort to make VRM avatars easier to build, host, moderate, and extend across engines, devices, and communities without trapping people inside a single vendor ecosystem.
- Unity-first workflows
- Quest-aware defaults
- Self-hostable services
- Portable avatars by design
Why this exists
VRM needs an end-to-end developer experience, not just a file format.
The format is promising, but the turnkey developer templates, moderation tools, hosting layers, and headset-native creator flows are still fragmented. VRMDN aims to close those gaps while keeping the ecosystem open, expressive, and welcoming to a wider range of communities and aesthetics.
Inclusive expression
Default avatars should support more than one visual culture or genre.
The project explicitly leaves room for different aesthetics, tones, age ranges, and creator communities so teams can choose styles that fit their players and spaces.
Safer access
Open tooling still needs moderation, accessibility, and high-quality UX.
VRMDN treats curation, safety controls, headset-friendly flows, and approachable onboarding as core product requirements rather than afterthoughts.
Roadmap: What is missing from VRM?
The pieces developers still have to stitch together by hand.
VRMDN is scoped around concrete gaps that developers still have to solve by hand. The goal is not another abstract standards page. The goal is practical infrastructure, reusable templates, and reference experiences that help teams ship.
01
Drop-in networking templates
Hover or tap to flip for the roadmap note.
01
Drop-in networking templates
Unity prefabs and community examples for avatar sync across networking stacks such as Normcore, Photon, Mirror, and Unity Netcode.
02
Cross-headset reference builds
Hover or tap to flip for the roadmap note.
02
Cross-headset reference builds
Canonical examples that run out of the box on Quest, PCVR, Vision Pro, Android XR, Pico, and other OpenXR-adjacent targets.
03
Headset-native avatar builder
Hover or tap to flip for the roadmap note.
03
Headset-native avatar builder
A visual avatar builder that works standalone in-headset and on the web, supported by liberally licensed starter assets.
04
Selfie-to-VRM pipeline
Hover or tap to flip for the roadmap note.
04
Selfie-to-VRM pipeline
A one-click flow from reference photo to a usable VRM, designed as a practical starting point rather than a locked black box.
05
Feature-rich open avatar stack
Hover or tap to flip for the roadmap note.
05
Feature-rich open avatar stack
Opinionated examples with pluggable pieces for IK, lipsync, auto-height, LODs, sit and stand variants, and full-body support.
06
Accessory compatibility
Hover or tap to flip for the roadmap note.
06
Accessory compatibility
An open clothing and accessory system so creators can make modular parts with a dependable baseline level of compatibility.
07
Reference hosting stack
Hover or tap to flip for the roadmap note.
07
Reference hosting stack
An open server and storage layer for hosting models, with a reference implementation for teams that want a usable starting point.
08
Moderation and safety controls
Hover or tap to flip for the roadmap note.
08
Moderation and safety controls
Curated repositories, age-appropriate subsets, and policy hooks so broad avatar expression can coexist with safer application spaces.
09
Future-facing format extensions
Hover or tap to flip for the roadmap note.
09
Future-facing format extensions
Longer-term exploration of richer VRM extensions, including pathways toward streaming avatar formats built on newer glTF capabilities (Gaussian Splats).
End-to-end flow
From reference photo or modular parts to a portable avatar people can actually use.
VRMDN is more than a rendering demo. It is a delivery network: creation, validation, packaging, hosting, moderation, and in-app delivery. That end-to-end view is what makes VRM practical for live products.
- Build or customize avatars from reusable asset packs, selfies, or creator-authored modular parts.
- Validate for target runtimes, curate for safety, and publish through open hosting and moderation layers.
- Deliver the right avatar variants into Unity-first products now while leaving room for browser, Unreal, Godot, and future runtimes later.
Where it starts
Begin with a stack that ships, then widen the circle.
VRMDN starts from a concrete internal need: Unity, Normcore, Quest, and Vision Pro, with the explicit intent to package the work for reuse and to expand support over time with community participation.
Phase one
Ship a practical Unity-first reference stack.
Build reusable VRM delivery patterns around Unity and Normcore, target Quest and Vision Pro first, and support avatar building both in-headset and on the web.
Phase two
Broaden support without collapsing into lock-in.
Add more runtimes, networking stacks, aesthetic packs, and community-maintained examples while keeping the defaults open source, modular, and portable.
Open ecosystem
Built on top of existing VRM work instead of replacing it.
Important ecosystem foundations already exist, including UniVRM, vrm.dev, VRoid, and three-vrm. Adjacent products and SDKs have also helped shape the direction of the effort. VRMDN complements that surrounding work with developer templates, builder UX, moderation layers, hosting patterns, and delivery infrastructure that a production team can adopt.
Foundation resources
Projects we studied while shaping VRMDN
About VRMDN
Who is behind VRMDN?
VRMDN is intended to be shaped with independent VR studios and other people across the ecosystem. We are currently building the effort through Boxtree, Inc., and we want the initiative to grow into a wider collaboration over time.
Current steward
Boxtree is driving the initial formation work.
The present goal is to turn internal product needs into reusable open infrastructure for VRM-based avatars, then widen participation without compromising portability, safety, or developer usability.
Get involved
Reach out if you want to help shape the initiative.
If you would like to be part of the formation of the VRMDN initiative, email hello@boxtree.gg.
Audience
Built for VR software developers who need to ship now and keep their options open.
If you build primarily in Unity for Meta Quest, this is the starting point. If you want that work to grow into a healthier, more portable ecosystem for avatars across XR, this is the direction of travel.