Ik modifier 3d

[fire]

For review

[fire]

Yay the ik is passing cicd.

[fire]

@TokageItLab How does chainik3d work?

[TokageItLab]

I plan that ChainIK3D works to process no-branched chains but uses ManyBoneIK solver info struct, and I think we can define BranchIK3D with extending ManyBoneIK for multi-effector to organize intersection set of the joints. I am currently implementing TwoBoneIK, and once that is complete, I will attempt to implement BranchIK3D.

[fire]

The basic idea behind Eron's decomposition algorithm is:

From each effector, traverse down to the skeleton root to find other effectors. As you do this maintain a list of bones_encountered, and current_weight

1a. Check if the bone is an effector other than this one. If it is, multiply current_weight by 1-effector_opacity of that effector.

1b. If current_weight is 0, stop descending for that effector.

1c. Otherwise. Add the bone to the list of bones_encountered and continue to its ancestor bone.

You now have a list of effectors, each containing a list of the bones that solve for them.

2b. Find any subsets of identical runs of bone sequences among the lists. Consolidate these into effector-groups, which contain the sequence of bones common to the origin bones_encountered lists, as well as the effectors those origin bones_encountered lists were for.

You now have effector-groups. Which are sequences of bones that all solve for the same effectors, and those bones all know how much to prefer each of the effectors they solve for.

Create an array of effector-groups, reverse sorted by the distance of the rootmost bone of each list from the skeleton root.

Create a solve_order array by traversing through the effectorgroups array, and appending each of its bones to the solve_order array.

You now have a solve_order array containing the bones in the appropriate order in which they should be solved, and with a constant set of effectors per run which you can use to minimize branching when solving, since you'll know that you're always solving for the same number of targets

[TokageItLab]

Superseded by #54