NoisyThingy was the most intelligent version of SpringyThingy with a totally redesigned software architecture. Si Li and Mikey Fujihara spent a considerable amount of time on a new data structure which allowed 'Context Responsive Modeling' and 'Behaviour Stacks'. For instance the surfaces produced could react (transform itself) to different contexts like 'sun' and 'wind' along with reacting to gravity. NoisyThingy interface was built to be totally immersive. The user no longer had to point at a model space - but be in it using wireless sensor driven wands. Also multiple users could draw in the same space.
The drawing environment or the model space was also non-uniform and dynamic - ie, other than gravity we had dynamic forces in the space. We could attach behaviour stacks with forces as well - over time the force field would transform itself as well. To make this dynamic force field more vivid we introduced the notion of 'snixels' - sound pixels. The forces in the model space would have snixels attached to it - they are invisible - but if a user gets close to it, it would make a small sound, the frequency of which is based on the amount of force in the position of a snixel in 3d-space.

Carnevan Chiu (SMarchS, MIT) and Carl Yu(SMarchS, MIT) built a system with IR Emitter + Sensors that allowed the user to draw in 3D space. It was a paradigm shift for Springy Thingy - from users pointing at a model space to physically being in the model space. Instead of a pointing device, we used a positioning device. This was one the first intensions of the entire project - making drawing a perfomance.