Actor is rarely recorded in splendid isolation, usually in the hand he holds an object, which is called the prop. This object has its own set of markers, often three is enough, and these markers are moving close to the actor. But the difficulty begins when the actor retarget into a character who has a totally different scale and proportion. And here, if for parts of the body retarget you can do automaticly, then, for all the additional facilities will have to do everything “by hands”. Although if you look at action space property in the character settings, there appears counted on the scale of the character to an actor.
Included bodies are not all that bad. In itself, the body has 6 degrees of freedom, 3 of which – the rotation and it does not depend on the scale (on action space). But there is one important observation that the centers of rotation of the body and source-receiver body should be relatively the same. For example, if an actor holding a sword in his hand and turns to them for support at the palm, then transmit the rotation to select the right pivot point, respectively palms on the character.
There are still 3 degrees of freedom – the movement, for whom, you should adapt the scale factor. As I wrote above, the scale can be viewed as already counted parameter action space. Then select the correct relative pivot point and with a simple Relation constraint pass local translation of the actor’s prop to local translation of the character’s one.
This is a simple scheme, but with a neat placement can give quite a good result. Here Cube – an object of weapon to the actor, driven by markers, and xm8 – the object of weapon from the character on whom the motion is passed.