Quasi Cycloidal Drive SomethingOrOther

Saw some variable ratio drive mechanism that made me think of this. Figuring out how to have variable radius drive pins wasn't too hard, but figuring out how to vary the radius of the circles they engage into seems to be a stumbling block. (Circles are actually a set radius which is based on an epicyclic drive, but the offset from center has to vary in the same way as the pin offset.) However even if that's worked out it wouldn't be a true cycloidal drive as the outer engagement surface wouldn't