Leap Motion Goes Mobile
blog.leapmotion.com
blog.leapmotion.com
It seems that the new version of the software, Orion (https://developer.leapmotion.com/get-started), made things better. But it has been in beta, and Windows-only, since at least last February... and the list of "known issues" doesn't seem to shorten over the months.
The gadget is now again in its box, where it has been for some time now. It's sad to see that they pivoted into VR based interaction, completely abandoned their current users in the process, and are now only working on the new VR version.
Eg conjuring cubes from finger tips and bouncing them away in the demo. I can see it being a better UI for many games / apps than the controllers
On the other hand, the "known issues" section is Orion-specific, not a previous version list that Orion improves upon.
- The company was being overly secretive about the tech behind the hardware, and as a result was exposing very little useful stuff in the API.
- There is something inherently dissatisfying about waving your hands in thin air to control something. The lack of any haptic feedback and no way to ascertain the vision frustum of the device made it very uncomfortable to use.
- Like a lot of startups, Leap Motion wanted to become a platform instead of being a hardware company. This made using it a total pain, having to go through their app store to accomplish even basic tasks. This also leads to a rather disingenuous customer-company relationship, as the company is not interested in just selling you their hardware, but also wants to control the way you use the hardware.
The entire thing would be far more appealing and hackable if they opened up their API and allowed the developer to access the raw data - depthmaps, point clouds, motion vectors - whatever's available internally. Sadly, this does not seem to be happening.
It was included in laptops. It was terrible and useless.
A classic example of potentially great technology with very poor implementation and application.
It was pretty bad. I was fighting the API the whole time because its post-processed palm and finger positioning data was noisy and and error-prone, and it didn't have a raw mode. The API would frequently lose track of individual fingers. You also couldn't do any hand gestures that involved your palm pointing horizontally instead of up/down because it would lose track of your occluded fingers. It also had issues tracking fingers when you held them together (as if you were going to karate-chop something). It's just too many quirks to deal with if you're trying to do hand position and gesture input.
Maybe it's improved since then, but after one day back around the initial launch I basically gave up on the tech for hand tracking, deciding that it was just not a good approach.
It's clearly a much more mature company now, so I'm genuinely curious even if I had reasons to give up on the original device.
A lot of these improvements were diminished when we went to VR -- new angles, complex backgrounds, different ambient lighting conditions. So we made the Orion software, which was an even bigger step up:
- lower latency
- longer range
- better and faster hand recognition
- vastly improved robustness to cluttered backgrounds and ambient light
It's hard enough to come up with compelling interactions using it already, adding technical issues and limitations meant enthusiasm died down. Now it's filed as "these fancy things that never really worked right" and attention has shifted to newer gadgets. I think it would be quite hard after that to convince manufacturers to integrate it again, since the first attempt didn't result in much outside the VR use case.
I've been meaning to come back to it and check out what's changed, but there is a long list of things I've been meaning to check out ;)
Only downside I see is in regards to not having something physical to grip/grab for many games. Would this work with a fake gun controller in a shooting game, for example? I doubt it could track my trigger finger when obscured.
This sounds pretty neat.