Are scene graphs specified by people or are there automated scene graph creation that balance multiplication reuse and scene graph depth?
soohyun3
@evelynk I am not sure if there are automated scene graph creation techniques that get applied to any arbitrary scenes, but in general scene graphs are manually created by people. This is because the hierarchy between parents and children is essentially defining a dependency between them, and how would automated processes understand arbitrary dependencies the creator might want?
That being said, there are techniques used to save computation from repeated matrix multiplications, which you can look into separately.
Koke_Cacao
They are generally specified by people. There are vertex animation and skinned animation. Transformation is typically applied to the bones of characters. And artists specify vertex weights followed to specific bones, but they don't have their own transformation matrix. However, some film industries do also store transformations for individual vertex, (very costly, so typically for facial expression) although they are generally captured from real-world data (face capture).
Are scene graphs specified by people or are there automated scene graph creation that balance multiplication reuse and scene graph depth?
@evelynk I am not sure if there are automated scene graph creation techniques that get applied to any arbitrary scenes, but in general scene graphs are manually created by people. This is because the hierarchy between parents and children is essentially defining a dependency between them, and how would automated processes understand arbitrary dependencies the creator might want?
That being said, there are techniques used to save computation from repeated matrix multiplications, which you can look into separately.
They are generally specified by people. There are vertex animation and skinned animation. Transformation is typically applied to the bones of characters. And artists specify vertex weights followed to specific bones, but they don't have their own transformation matrix. However, some film industries do also store transformations for individual vertex, (very costly, so typically for facial expression) although they are generally captured from real-world data (face capture).