We developed very small and light sensors, each equippedwith 3-axes accelerometers, magnetometers and gyroscopes.Those MARG (Magnetic, Angular Rate, and Gravity) sensors allow for a drift-free attitude computation which in turnleads to the possibility of recovering the skeleton of bodyparts that are of interest for the performance, improvingthe results of gesture recognition and allowing to get relative position between the extremities of the limbs and thetorso of the performer. This opens new possibilities in termsof mapping. We kept our previous approach developed atARTeM : wireless from the body to the host computer,but wired through a 4-wire digital bus on the body. Byrelieving the need for a transmitter on each sensing node,we could built very light and flat sensor nodes that can bemade invisible under the clothes. Smaller sensors, coupledwith flexible wires on the body, give more freedom of movement to dancers despite the need for cables on the body.And as the weight of each sensor node, box included, isonly 5 grams (Figure 1), they can also be put on the upper and lower arm and hand of a violin or viola player, toretrieve the skeleton from the torso to the hand, withoutadding any weight that would disturb the performer. Weused those sensors in several performances with a dancingviola player and in one where she was simultaneously controlling gas flames interactively. We are currently applyingthem to other types of musical performances.