Corrective Systems
Corrective blend shapes are added to maintain volume and create distinctive shapes within a deformation. However, the goal is to paint the weights as best as possible first, not to rely on the shape to ‘fix’ any skinning issues. Corrective shapes adjust geometry using the (blend) shapes deformer system.
Fall-off joints: Positioned throughout the rig to help extend and blend the twist of geometry. For example, a forearm fall-off joint to help when the wrist twists side to side
On-site shapes: For example, using the Pose Space Deformer system in Maya instead of duplicating the mesh to create a blend shape
Off-site shapes: Duplicating the mesh and creating corrective blend shapes or for additional rigging effects
Influence joints: A series of joints positioned and animated (SDK) to work with the movement of the rig – Maintaining volumes or for adding additional effects.
Influence objects: These can be a simple as a single plane of geometryor full muscle systems
A blend shape is a duplicate of the model which has been edited, and when you target it in and assign it a slider, when the slider is on 1 the original becomes the target.
Behind the scenes the blend shape node simply stores the change sin coordinates of each vertex. After a blend shape is made, the duplicate model (target) can be exported of deleted from the scene
All targets under one blend shape node work together (morph / blend). Two blend shape nodes plugged into one model. This will conflict with each other
For skinned models (characters) the blend shape node needs to be below the skin cluster node so the blend shapes morph onto of the movement (deformation) caused by the rig
The change in geometry is triggered by the rotation of the joint using a blend shape
The shape that is used to correct model is the result of a subtracted shape (positive / negative shapes) to accommodate the skinned mesh.
Corrective shape:
The model is posed and then duplicated twice. First to capture the problem (negative) and second to fix the problem (positive. Both are then connected as a blend shape to the original skinnedmess. Both duplicate shapes are then connected to the original using a ‘parallel blend shape’. The negative is set to –1, the positive to +1, and the rig is reset. A new duplicate is created of the original model and the parallel blend shape deleted. This is the corrective shape that will be used.
Fall-off joints example:
I have created a correct the wrist movements. One option is to insert a joint. After I have inserted a joint in between the elbow and the wrist joint, in order to mirror the joint on the other arm, I have created a locator of which the origin corresponded to the joint origin and mirrored to the there arm to use as a guide to insert the joint.
Add influence to the skinning
create the IK handle but with limiting the movement of the joint so it can rotate on one axis
add expression to make the joint follow the rotation of the inserted one
insert the joints on the legs as well (by duplicating the knee joint and deleting the ankle duplicate and parented it to the knee joint) and add the expression on rotate y to make the ankle and the inserted joints rotate along
skin weight the arm new joints replace the influence of the elbow with the forearm joint with 1 percent influence
after in order to maintain the mesh volume when the forearm joints rotate The influence on the mesh between the elbow and the forearm needed to be shared (by selecting the vertices from the edge loop of the forearm area and either replace the influence with the elbow to add it to the forearm, or using the hammer tool)
same skinning process for the knee and the shin joints on the leg
An alternative to the blend shapes are joints that are parented to the “problematic joints that are skinned to the mesh to correct the shape of them when doing a specific movements.
Corrective shapes example:
first example:
Blendshapes are a process that can be used to correct the Character mesh shape in a specific pose. one the character pose is created the mesh should be duplicated twice: one to use as a positive and one as a negative.
By editing the positive mesh duplicate and selecting the three meshes now a parallel blend shape can be created. in the attributes id the positive is set to 1 and the negative to -1 now the pose would be correct.
Once the character is pose back in the default pose, however, the mesh would be changed and in order to edit it the mesh has to be duplicated one more time (this duplicate would be used as blend shape) and with the set driven key editor with the character set as the driver and the bleanshape as the driven now the pose is finally correct.
Second example:
The second process involves the pose editor. First select a joint and after add poses for each movement with the edit button turned on. for each individual joint every rotation where the problems in the mesh appear should be added.
Once each joint has been added on one side the pose can be mirrored on the other side using the pose editor mirror options set to either object or topology.