We are implementing the VRM specification for 3D avatar models. The VRM is based on GLTF, and is used by extension specifications. Last year, version 1.0 was released. The majority of applications have been made and used in Japanese.
VRM features humanoid bones, face blendshapes, colliders, and other avatar features. Realbits changes its avatar's facial expression using VRM blendshapes and moves its arm using VRM humanoid bones.
In order to detect face land marks and body poses, realbits uses mediapipe from Google and kalidokit's open-source kinematics solver open source. With Google Mediapipe's holistic detection engine, there are lots of pose detecting and converting solutions and open source tools for humanoids. In the future, we may be able to use another solution for that.
Avatar NFT can be used in applications that support VRM format with the help of VRM. Realbits avatar NFT is also available in GLB format. As a result, if the application or service supports that format, avatar NFT content can be used.
Adding an avatar model to such a format will be possible if a metaverse or game service requests it.