Avatar System [Draft]

The avatar system is a behavioral standard, not a format for arbitrary rigs. Any avatar (human or not) must expose a known set of smooth, adaptive animation hooks:

• Movement (walk, run, idle)
• Facial (expressions, talking)
• Interaction (sit, grab, look-at — with directional/adjective flexibility, the interacted item need to provide a reference point)

Avatar standard

For now, the avatar is a movable model in space. Its movement is triggered by events from the external interface. The avatar can move to different positions and be used to trigger interactions with other artifacts.

In the future, the avatar should support smooth, adaptive animations. A potential solution is to adopt the animation conventions defined in the VRM standard, which provides a structured system for movement, expressions, and interaction semantics.

Implementation of Avatar Customization

Each user can directly use their own avatar, represented as an Artifact whose content conforms to the avatar standard. This approach gives users maximal freedom: the avatar can be humanoid or any other creature form.

To support broader adoption, we will also provide a simple avatar editor. This editor allows users to customize their avatar using a modular configuration interface. Based on the customization metadata, the system can either generate the avatar model at runtime or store a finalized version on-chain.

The primary technical difference across avatar models lies in their rigs and skeleton structures, which are orthogonal to our avatar standard. Artists are free to define their own rigging and skeleton layouts, along with compatible customization options (in their own editor like Blander). And any user can start their customized avatar building in our editor base on any rigging and skeleton layouts that they own or have obtained from an artist. This allows users to build fully customized avatars of any form.

Fashion standard

Each rig/skeleton definition can have corresponding fashion items that can be applied to compatible models. Artists must define how each fashion item integrates with their rig, including attachment points, weighting, layering, etc. To ensure compatibility between fashion items and diverse rigs, we will define a dedicated fashion standard, which, for example, specifies how to:

• Declare the expected rig structure
• Bind fashion items to rig points
• Maximize interoperability across different avatar types and customization sets This enables a flexible and extensible system where creators can define new avatar archetypes and customization kits while maintaining consistency and maximize usability for each fashion item.