I recently purchased the Leap Motion for exclusive use with the Oculus Rift DK2. I was convinced to try it out due to its recent developments in skeleton tracking and optimisations for use as a hmd forward-facing depth camera. I have always believed that your hands are the most flexible and most natural controls and would be ideal for use in a 3d virtual environment. So the Leap Motion should be a suitable device and become the missing component to VR - the controls.
At this time, there are not many VR demos that come with the Leap controller; here is a summary of the 5 that are available:
Flocking VR - a simple demo of being surrounded by fish, when your hands come into view, the finger ends become lights and attract the fish. A nice demo to play around with for a few minutes.
Isolated Hands - this demo runs at about a frame a second. Runs fine without the Leap Motion input so must be the computation on the leap data. The demo attempts to show just the hands by identifying the hands from the raw video output and using it as a mask. Maybe this will work with future SDKs.
Oculus Passthrough - Another slow demo - I think Unity demos that use the video feed from the Leap Motion suffer from slow frame rates. A simple demo of interacting with cubes whilst also viewing the video feed from the Leap Motion controller.
Plasma Ball VR - similar to the Flocking VR demo - you can have limited interaction with a wobbly ball in front of you.
VR Intro - perhaps the best demonstration for the Leap, this demo allows you to access the quality of the Leap controller by seeing the video input and a skeleton mesh on top. Hint: fade out the video input so you only see the skeleton - doesn't it feel weird controlling virtual hands. There are numerous modes in this demo too - like interacting with balls surrounding you and putting your hands forward
Does it work?
The Leap controller works but only for a limited distance - reliably at about 30cm from the device. This is unfortunate as this distance is far too short to be practical; we need around a full arm's length (over 100cm) to use the device properly in a virtual environment. I think this is due to the limited resolution of the camera in the Leap - you can still see your hands on the raw output but they have not been matched to the skeleton.
At this point in time, the leap motion is probably not worth buying - unless you are a developer and want to look into using just the raw data from the leap and manually matching parts of your hands.
There are some things I want to do with the Leap Motion. Although it is not currently accurate enough to be of any use in a game there are a couple of things that could still be done.
Render your hands - Like the Isolated hands demo, rendering the hands would still add a layer of immersion to any game. Rendering an avatar adds immersion to a vr environment and rendering your hands on that avatar would make it feel even better.
Match simpler gestures - The skeleton that is used to match the hand in most of the demos is complex, we could instead attempt to match just the ends of the hands to the base. Thus gestures like a gun or a bird could be matched more easily - perhaps at a longer, more useful range.
Alternatives to the Leap Motion
Whilst the Leap does not live up expectation, there are alternative devices that could be used in a similar manner to the Leap - providing the ability to track your hands more reliably:
Creative Senz3D - this device is like a short-range Kinect and suitable for depth tracking.