For more detailed informations check Inochi Creator and Inochi Session.
On vtubing, content creator uses model as avatar to present themselves. Model can be 2D or 3D. On 2D models, there are static models (.png) and moving 2D models (react png, realtime facetracking).
Motion tracking is method to capture human motion. It can use webcam and specific hardware for to do so. On vtubing, there are face tracker, hand tracker and full body tracker.
Toggles uses shortcut-like mechanism to trigger an expression on model or to play pre-made animation
Shortcut and hotkeys on Inochi2D's software.
The terms of “Bones” and “BlendShapes” are an abstractions that comes from the digital animation world. Bones are the elements used to define the orientation of 2D/3D models parts. A bone has a position (X, Y, Z) and an orientation (Yaw, Pitch, Roll). On the other hand, blendshapes are used to express mesh deformations, usually through the use of key frames and interpolation.
In the context of camera tracking, tracking software usually creates an internal 3D model that is deformed to match the shape of the body seen by the camera. Then both bone and blendshape information is extracted from this internal 3D model to be sent as data. For example, a parameter that most camera tracking software abstracts as a bone is the Head, which has both a position and orientation. More advanced tracking solutions can also send hand and finger data as bones. In the case of blendshapes, the parameters carry numeric information on how much a specific part of the tracked model has deformed.
In short, Blendshapes are a way to represent facial expression data. You can use them to input those expression to your vtuber model.
Example: