What it was.
This work was driven by a concrete product goal: make cloud-streamed games feel better in Fluid, especially through Xbox Cloud Gaming.
A tracked gamepad gave players a more familiar bridge between flat gaming expectations and spatial interaction, instead of forcing them into an XR-native control scheme for everything.
The hardware constraint was that the VR devices did not support fiducial markers or object recognition for the controller itself, so the 6DOF tracking had to be inferred instead of measured directly.
I used tracked hand positions and reverse engineered where the controller should be in space, then tuned the interaction so that estimate felt stable and believable in use.
I treated it as an experience problem first: the controller had to feel like it belonged in the space, respond naturally, and support transitions without breaking flow.
That made it a practical enhancement to Fluid's cloud gaming support rather than a standalone input experiment.
The result was a controller model with inferred 6DOF tracking, built from tracked hand positions despite the lack of fiducial-marker or object-recognition support, and smoother transitions that made Xbox Cloud Gaming in Fluid feel more natural.
