One of the reasons why linear blend skinning is widely used is due to its efficiency and most are willing to trade off the potential artifacts associated with the candy-wrapping effect. What I can speculate is probably favorable for real time computation in video games. However, the proposal of the Dual Quaternion skinning handles artifacts without incurring a significant performance loss. Link to the actual paper:
Then in real-life, if we are using LBS and we realize the joint has to rotate by 180 do we just treat it as an edge case and render that part with another skinning technique?
alvin-xu-5745
Linearity has clear advantages of simplicity, so the creator of Dual Quaternion Skinning actually proposed methods to approximate these nonlinear algorithms with linear methods. They try to show this for not only Dual Quaternion Skinning but for nonlinear blending in general.
One of the reasons why linear blend skinning is widely used is due to its efficiency and most are willing to trade off the potential artifacts associated with the candy-wrapping effect. What I can speculate is probably favorable for real time computation in video games. However, the proposal of the Dual Quaternion skinning handles artifacts without incurring a significant performance loss. Link to the actual paper:
https://users.cs.utah.edu/~ladislav/kavan08geometric/kavan08geometric.pdf
Then in real-life, if we are using LBS and we realize the joint has to rotate by 180 do we just treat it as an edge case and render that part with another skinning technique?
Linearity has clear advantages of simplicity, so the creator of Dual Quaternion Skinning actually proposed methods to approximate these nonlinear algorithms with linear methods. They try to show this for not only Dual Quaternion Skinning but for nonlinear blending in general.
https://users.cs.utah.edu/~ladislav/kavan09automatic/kavan09automatic.pdf