VRM: Concept and Implementation of an Application-Interoperable Avatar Format

As the metaverse continues to expand, avatars have become essential representations of personal identity in virtual environments. To enable seamless use of avatars across multiple platforms and applications, the VRM file format was proposed as a platform-independent, runtime-loadable 3D avatar specification. VRM facilitates avatar interoperability by standardizing humanoid model structures and emotional expression (expressive behavior) metadata, allowing users to carry a consistent virtual identity between diverse systems. This paper introduces the design philosophy and technical structure of VRM, including its intermediate layers for emotion abstraction, constraints for animation control, and viewpoint settings optimized for immersive first-person VR usage. Moreover, the format integrates licensing information, including personality rights, enabling creators to govern how avatars are used and represented. Developed originally in Japan and now under active standardization with the Khronos Group, VRM offers a foundation for portable embodiment in the metaverse era.

https://xrstand-standardization-committee.github.io/ISMAR2025--XRStand2025_Workshop/

1. XRStand 2025 in ISMAR 2025
VRM: Concept and Implementation of an
Application-Interoperable Avatar Format
Shinnosuke Iwaki Takeshi Kurata
Virtual Cast, Inc. Research Institute on Human and Societal Augmentation (RIHSA), AIST
ISO/IEC JTC 1/SC 24

2. 2
VRM: Enabling Cross-Platform Connectivity and Collaboration
in Virtual Worlds
•To enable seamless use of avatars across multiple platforms and applications, the VRM file
format was proposed as a platform-independent, runtime-loadable 3D avatar specification.
•VRM was originally developed in Japan and is currently being standardized in collaboration with
the Khronos Group.

3. 3
Intermediate layers for avatar interoperability
•VRM facilitates avatar interoperability by standardizing humanoid model structures and
emotional expression metadata, allowing users to carry a consistent virtual identity between
diverse systems.
•VRM includes its intermediate layers for emotion abstraction, constraints for animation control
Gaze control via the VRM API.Variations of expressive behavior representation
in 3D model structure.
VRMC_vrm: expression in VRM 1.0.

4. 4
Definitions unique to avatars used in virtual spaces
First-person perspective and visibility settings.License conditions with personality rights
for a VRM avatar model.
•Viewpoint settings are optimized for immersive first-
person VR usage.
•The format integrates licensing information with
personality rights, enabling creators to govern how
avatars are used and represented.

5. 5
Toward global recognition: Collaborative efforts byVRM Consortium,
Khronos Group, and ISO/IEC JTC 1
VRMC_vrm-1.0 Avatar-related extensions, including licensing and Humanoid definitions.
VRMC_springBone-1.0 ‘springBone.’ Definitions for secondary motion (e.g., swaying hair, clothing).
VRMC_springBone_extended_collider-1.0Extensions of the springBone collider (plane collider and capsule collider).
VRMC_node_constraint-1.0 Extensions for node constraints.
VRMC_materials_mtoon-1.0 Toon shader extension.
glTF extensions for VRM 1.0
Supported by more than 181 applications and tools.