MotionInput.

Tuesday

Abid brought an almost completed implementation of the joystick aim hand box with some minor bugs. After debugging with the team we finally managed to complete the joystick hand box module which operated almost perfectly. Later on in the week, Taha optimised even further by introducing dead zones to the aim box as well as introducing a simple algorithm for aim sensitivity to ramp quadratically instead of linearly to closer represent the aim in many games where you have to aim in.  

We also decided on Tuesday to re-align our efforts with the core of our original brief, which was to allow for custom gesture use in superhero games. We started from scratch as the gesture recorder in the current MotionInput implementation hasn’t been used in years and has nearly no documentation to help us. In the session we made our own static gesture recorder which uses the ratio between hand landmarks to record single frame hand gestures (poses) with high flexibility where you can rotate your hand and the recorder will still detect the gesture.  

Wednesday-Thursday

Throughout the week Abid took this idea even further by making a dynamic gesture recorder for both hand gestures and full body gestures, so that players can make a movement over time to be detected, allowing for a much more immersive experience in play. It uses multiple algorithms that normalises coordinates in a 2D space to allow for gestures to be recorded from a mp4 file but work with people of all body types.  It works in real time to track points of bodies during play.  

Friday

On Friday we presented the new prototype gesture recorders with Professor Dean Mohammedally who was very impressed with the work. For next steps we need to continue to optimise current implementations of the games that we have as well as implementing our gesture recorders into MotionInput.