Works in --palm mode but are these the correct gestures?

[matanox]

This is great stuff. +1 for the two hand approach which is much more ergonomic. 💯
I'm not sure about the gestures though, tilting a palm left or right nicely drives the pointer on the axis, but for moving the pointer up or down I seem to use a gesture that's way too original (no matter the orientation of the device, I need to tilt my hand/palm down to move the pointer up, and tilt my hand/palm up to move the pointer down.

Is it supposed to work much differently?

[matanox]

Never mind, I forked and switched the gesture.. let me know if you want a patch, or in case I had totally misinterpreted the gesture.

[wyager]

Could you describe a little more what you mean? The idea is that, if your screen were flat on your desk, the back of your palm (excluding the vertical axis) would point in the direction the cursor was meant to go.

This code comes from a pretty early alpha of the leap API, so it's possible they switched something.

[matanox]

Oh not at all, in that case it works as planned. I just prefered the usability metaphor of having one's fingers point up or down, which translates to reversing the direction effect of the same gesture. The code still 'works' by the way, just the design needs a lot of evolution. For example, when a hand is withdrawn or introduced into the scene, it tends to cause very random behavior and this is just one design challenge working with the Leap.

[wyager]

Hmm. Yeah, this code was designed on the v0.6 SDK or something, so I'm not surprised. I see there are quite a few forks of this project; if anyone wants to submit a pull request with changes such that the program works properly with the new API, I'll be very amenable to merging.

Matanster, if you want to submit a pull request with a switch to enable the behavior you're describing, I could merge that as well.

[matanox]

Well actually, the main problem with designing desktop control for the Leap Motion, is the aspect of hand withdrawal and introduction into the scene. Given that there's little to no use in engaging a desktop solely with the Leap (consider switching to your keyboard, or using your mouse to click as the Leap isn't that well catered for click safety - and inadvertent mouse clicks are a very big deal) - then there is the problem of retracting a hand or both hands from the scene - short of mind reading, any code will trigger a myriad of events during the retraction process rendering the usefulness of any Leap desktop quite close to none beyond initial developer enthusiasm.

The only solution may seem to be introducing a delay in desktop impact so that a withdrawal is only recognized as a withdrawal, which limits the design. (plus, the Leap SDK doesn't recognize a withdrawal same as it recognizes gestures so one has to capture this gesture on their own).

An equivalent or similar problem occurs in introduction of a hand to the scene.

It seems with the current API, the Leap is somewhere between limited to ill-suited for desktop control that can be used by humans and all of these attempts are really a naive waste of time, unless one really has the time to artfully bypass these problems while still being left off with a convention of usage that is usable and convenient to human beings.