Leap Motion reveals super-accurate motion control tech, $70 device to change the UI game
252 views
Skip to first unread message
Murilo Saraiva de Queiroz
May 21, 2012, 4:07:00 PM5/21/12
to openn...@googlegroups.com
Does anybody have more information about the new Leap Motion technology?
The good performance seems to come from a much smaller distance and better software (but I doubt their 1/100 mm accuracy claims)... I couldn't find any information about which specific approach (ToF ? Structured light?) they use, though.
muriloq
Murilo Saraiva de Queiroz, MSc
Hardware Engineer at NVIDIA
Lorne Covington
May 21, 2012, 4:56:10 PM5/21/12
to openn...@googlegroups.com
I'm guessing it is not a depth camera at all but more some kind of light-curtain or scanner. It may just return X/Y position of objects in a plane, or narrow volume. Notice all of the demos depend on movement in a plane, not depth from the screen (other than just not being seen). The exception is the hand demo, but that is obviously faked to some extent as the hands have dimensionality on all sides/top, which a device positioned under the hand could not possibly see. The lack of any kind of depth claims makes me think it's mainly X/Y too. We'll see, I'm just talking out of my shorts here!
The problem with this and any kind of air-gesture UI is of course that it is tiring and lacks positive haptic feedback, so I can't see these kinds of things being used for almost any of the applications in their video - mice, joysticks, and touch screens do those jobs pretty darned well. I think until people start focusing on what new kinds of things NUIs can do, not much is going to advance.
Ciao!
- Lorne
--
You received this message because you are subscribed to the Google Groups "OpenNI" group.
To post to this group, send email to openn...@googlegroups.com.
To unsubscribe from this group, send email to openni-dev+...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/openni-dev?hl=en.
M@
May 21, 2012, 5:38:02 PM5/21/12
to OpenNI
I emailed them and asked what they were actually measuring and he
didn't give me a strait answer, but he hinted that it's done
visually. Things that look like they have fingers coming out of them
will register as long as they're not polished too well or jet black.
If I had to guess I'd say they just hooked up a high res digital
camera sensor to a custom machine vision chip that looks for cigar
shapes, and then built out a lot of filters on top.
--M@
On May 21, 4:56 pm, Lorne Covington <mediadogcont...@gmail.com> wrote:
> I'm guessing it is not a depth camera at all but more some kind of
> light-curtain or scanner. It may just return X/Y position of objects in
> a plane, or narrow volume. Notice all of the demos depend on movement
> in a plane, not depth from the screen (other than just not being seen).
> The exception is the hand demo, but that is obviously faked to some
> extent as the hands have dimensionality on all sides/top, which a device
> positioned under the hand could not possibly see. The lack of any kind
> of depth claims makes me think it's mainly X/Y too. We'll see, I'm just
> talking out of my shorts here!
>
> The problem with this and any kind of air-gesture UI is of course that
> it is tiring and lacks positive haptic feedback, so I can't see these
> kinds of things being used for almost any of the applications in their
> video - mice, joysticks, and touch screens do those jobs pretty darned
> well. I think until people start focusing on what new kinds of things
> NUIs can do, not much is going to advance.
>
> Ciao!
>
> - Lorne
>
> On 5/21/2012 4:07 PM, Murilo Saraiva de Queiroz wrote:
>
>
>
>
>
>
>
> > Does anybody have more information about the new Leap Motion technology?
>
> >http://www.engadget.com/2012/05/21/leap-motion-3d-motion-and-gesture-...
> > /Hardware Engineer at NVIDIA/
didn't give me a strait answer, but he hinted that it's done
visually. Things that look like they have fingers coming out of them
will register as long as they're not polished too well or jet black.
If I had to guess I'd say they just hooked up a high res digital
camera sensor to a custom machine vision chip that looks for cigar
shapes, and then built out a lot of filters on top.
--M@
On May 21, 4:56 pm, Lorne Covington <mediadogcont...@gmail.com> wrote:
> I'm guessing it is not a depth camera at all but more some kind of
> light-curtain or scanner. It may just return X/Y position of objects in
> a plane, or narrow volume. Notice all of the demos depend on movement
> in a plane, not depth from the screen (other than just not being seen).
> The exception is the hand demo, but that is obviously faked to some
> extent as the hands have dimensionality on all sides/top, which a device
> positioned under the hand could not possibly see. The lack of any kind
> of depth claims makes me think it's mainly X/Y too. We'll see, I'm just
> talking out of my shorts here!
>
> The problem with this and any kind of air-gesture UI is of course that
> it is tiring and lacks positive haptic feedback, so I can't see these
> kinds of things being used for almost any of the applications in their
> video - mice, joysticks, and touch screens do those jobs pretty darned
> well. I think until people start focusing on what new kinds of things
> NUIs can do, not much is going to advance.
>
> Ciao!
>
> - Lorne
>
> On 5/21/2012 4:07 PM, Murilo Saraiva de Queiroz wrote:
>
>
>
>
>
>
>
> > Does anybody have more information about the new Leap Motion technology?
>
>
> > The good performance seems to come from a much smaller distance and
> > better software (but I doubt their 1/100 mm accuracy claims)... I
> > couldn't find any information about which specific approach (ToF ?
> > Structured light?) they use, though.
>
> > muriloq
>
> > --
> > *Murilo Saraiva de Queiroz, MSc*
> > The good performance seems to come from a much smaller distance and
> > better software (but I doubt their 1/100 mm accuracy claims)... I
> > couldn't find any information about which specific approach (ToF ?
> > Structured light?) they use, though.
>
> > muriloq
>
> > --
> > /Hardware Engineer at NVIDIA/
jeff kramer
May 21, 2012, 5:39:50 PM5/21/12
to openn...@googlegroups.com
One of the posters on HN stated that he had a prototype and it worked with dual stereo IR cameras and structured light.
Grain of salt and all that. =)
Jeff
Victor
May 22, 2012, 6:08:38 PM5/22/12
to openn...@googlegroups.com
This definately looks impressive. I am eager to see the hardware specifications - is it a camera? structured light? normal rgb? is it a sonar-like sensor, a laser?
Good guesses are given in this thread, keep the flow of information please!
I hope that is not just marketing, such a device as the one they describe would be very nice to have!
Good guesses are given in this thread, keep the flow of information please!
It sounds too good to be true, we know. But, that’s what we specialize in around here.
I hope that is not just marketing, such a device as the one they describe would be very nice to have!
Sinan DİNÇ
May 23, 2012, 1:43:09 PM5/23/12
to OpenNI
very impressive! possibly equipped with stereo or quadro IR cameras. i
wish it had full body tracking too,
it could be Kinect killer!
On May 21, 11:07 pm, Murilo Saraiva de Queiroz <muri...@gmail.com>
wrote:
wish it had full body tracking too,
it could be Kinect killer!
On May 21, 11:07 pm, Murilo Saraiva de Queiroz <muri...@gmail.com>
wrote:
> Does anybody have more information about the new Leap Motion technology?
>
> http://www.engadget.com/2012/05/21/leap-motion-3d-motion-and-gesture-...
>
>
> The good performance seems to come from a much smaller distance and better
> software (but I doubt their 1/100 mm accuracy claims)... I couldn't find
> any information about which specific approach (ToF ? Structured light?)
> they use, though.
>
> muriloq
>
> --
> The good performance seems to come from a much smaller distance and better
> software (but I doubt their 1/100 mm accuracy claims)... I couldn't find
> any information about which specific approach (ToF ? Structured light?)
> they use, though.
>
> muriloq
>
> --
> *Murilo Saraiva de Queiroz, MSc*
> *Hardware Engineer at NVIDIA*
Alex Willisson
May 24, 2012, 2:10:48 AM5/24/12
to openn...@googlegroups.com
I contacted them about when they expect to have linux support,
apparently, "[they] should have Linux support by the time it ships."
That said, I think I'm going to hold off until I've seen some reports
on other people using it before I get one for myself. It looks great,
but I'm a little suspicious (mostly related to the points brought up
in this thread).
I'm not sure if I'd mind too much even if it was only good at tracking
points in an XY plane parallel to the screen. If it was really good at
that, it'd still probably be a fun thing to have, at least to play
with.
apparently, "[they] should have Linux support by the time it ships."
That said, I think I'm going to hold off until I've seen some reports
on other people using it before I get one for myself. It looks great,
but I'm a little suspicious (mostly related to the points brought up
in this thread).
I'm not sure if I'd mind too much even if it was only good at tracking
points in an XY plane parallel to the screen. If it was really good at
that, it'd still probably be a fun thing to have, at least to play
with.
Reply all
Reply to author
Forward
0 new messages