How to do a gesture
Move hand to the rough position and angle that a gesture starts with (e.g., like you’re about to draw a sword from a sheath)
You’ll see a path showing what to do
Follow the path with your hand
Once you follow to the end, the gesture triggers
How to edit a gesture
Move hand to the rough position and angle that a gesture starts with
Hold the grab button with that hand
While holding grab, move and rotate it to a new starting location. Release to save the new gesture starting location.
While holding grab, hold the trigger button to re-draw the gesture. Release to save the new gesture.
Why?
People love the idea of doing gestures (like a cool pose) to cast spells in games
Everyone has different anatomy, so people do the same basic actions incredibly differently, especially when they’re focused on combat
For example, one player’s punch can end up over 40 cm and 60 degrees away from another person’s punch
So a gesture recognition system built for one person simply doesn’t work for everyone else
The result is that every single system out there causes you to constantly (1) randomly fail or (2) accidentally trigger the wrong gesture
This makes VR games have to choose between (1) only having a few gestures or (2) being unplayably clunky
What does this do differently?
Solve the problem with UX, rather than machine learning
In real-time, show the user exactly what to do to complete a gesture = far fewer mistakes
Let the user modify gestures with 0 UI clicks = users actually customize, instead of blaming the system
Automatically adjust gestures based on the user’s earlier attempts = system gets even better over time
Caveats
I still haven’t added a good way to track the user’s body. This means the system breaks if the user rotates around in real life (which happens very often during real gameplay).
