As I was developing a video game, I actually implemented inverse kinematics for dynamic animations and the issue of IK solving for unnatural limb rotations was indeed a huge hurdle to overcome.
Rishiparikh
For me I also have ran into similar inverse kinematics issues when working with robots. In these situations, the robot would be unable to perform a task that seemed very reasonable to me. At the time, I did not know why this was happening, but now it makes a little bit more sense as to why we get kinematics that are not always consistent with what I expected them to be.
ethanyxchen
Is the reason why it is time consuming for artists because inverse kinematics is so difficult to solve? Do many artists/graphics programmers have to rely on computationally inefficient methods to solve inverse kinematics problems? Or do they avoid inverse kinematics altogether?
micahtyong
Does this pros and cons list apply to both forward kinematics and inverse kinematics? I was under the impression that IK lacks direct control and is increasingly difficult to implement as the number of links, N, increases. Then, if we use a data-driven approach with inverse kinematics, animation is more consistent with physics if we frame it as an optimization problem. Then, most of the time is spent training a model rather than wasting an artist's time.
As I was developing a video game, I actually implemented inverse kinematics for dynamic animations and the issue of IK solving for unnatural limb rotations was indeed a huge hurdle to overcome.
For me I also have ran into similar inverse kinematics issues when working with robots. In these situations, the robot would be unable to perform a task that seemed very reasonable to me. At the time, I did not know why this was happening, but now it makes a little bit more sense as to why we get kinematics that are not always consistent with what I expected them to be.
Is the reason why it is time consuming for artists because inverse kinematics is so difficult to solve? Do many artists/graphics programmers have to rely on computationally inefficient methods to solve inverse kinematics problems? Or do they avoid inverse kinematics altogether?
Does this pros and cons list apply to both forward kinematics and inverse kinematics? I was under the impression that IK lacks direct control and is increasingly difficult to implement as the number of links, N, increases. Then, if we use a data-driven approach with inverse kinematics, animation is more consistent with physics if we frame it as an optimization problem. Then, most of the time is spent training a model rather than wasting an artist's time.