Map pose and best rig practices
- Former user (Deleted)
- Ben Stanley
Map pose is used to constrain the target skeleton to the source skeleton. To get the optimum results from the retarget, the source and target poses need to be as close as possible.
When you click Set Map Pose, Shogun Post takes any differences between the two skeletons and stores them as offsets within the VSR file. Any joints with default values or keys are copied onto pre-rotation to match the source Vicon skeleton and stored as the map pose for the solver to use. The original values are maintained in the FBX or USD skeleton, enabling the skeleton to be imported back onto the original rig without any issues.
In addition to supporting the use of skeletons that use pre-rotation to define the base pose (see Retarget with Shogun Post in Getting started with Vicon Shogun), Shogun Post also supports the use of skeletons that have zeros for pre-rotation and use keys or default values to achieve this.
For example, Mannequin.FBX is a commonly used target skeleton. It contains joints like the clavicles which, instead of using pre-rotation, use keys to define a base pose, (or more accurately, the appearance of a base pose, as the actual base pose is not a real base pose.)
When Mannequin.FBX is loaded into AutoDeskĀ® MotionBuilderĀ®, at first glance it appears to be in a base pose:
However, notice that the pose is actually defined by a set of keys on frame 0 and the pre-rotation (as seen in the Properties) is zeros for bones like the clavicle, which have a rotation so the X axis can point to the side even though its parent has the X pointing up.
To see the actual base pose, delete the keys and set all the rotation values for all bones to zero:
This demonstrates why it's better to use the pre-rotation to define the base pose. When all keys are removed, and all bone set to zeros, it's preferable to have a base pose that's a clear T- or A-pose instead of the stack of bones shown above. It's also more sensible because the animation you place on a skeleton via mocap or hand-keying is easier to deal with when poses are relative to zero. For example, it's easier to think of a bend of the elbow as X degrees from 0, not X from 90, which is what you'd have if straight was actually 90 because of the current base pose.
Ideally, you would rebuild skeletons like this to use pre-rotation instead, but this is not always possible. When it's not possible, you can still use the skeleton, but you must set the map pose (see Retarget with Shogun Post in Getting started with Vicon Shogun).
Benefits of using map pose:
- The pose used during mapping is saved and restored when you go back into map mode or load the VSR into the Retarget app.
- The target skeleton matches the source skeleton's joint orientation more closely, resulting in a better solve.
