Forgot your password?
typodupeerror
Apple Build Hardware Science

Glasses-Free 3D On iPad (Sorta) 96

Posted by CmdrTaco
from the where-all-the-blue-women-at dept.
johkir writes "The Engineering Human-Computer Interaction (EHCI) research group has created an apparent 3D effect on the iPad 2. Called the Head-Coupled Perspective, it uses the front camera to track the relative changes in the position of the user's head."

This discussion has been archived. No new comments can be posted.

Glasses-Free 3D On iPad (Sorta)

Comments Filter:
  • But I wonder how it'll react to being touched. They should put out an API for this sort of thing. Could truly change the industry.

    • Dunno about that. I still believe that 3D is one of those things that is interesting for a while, and then you move on. The only thing that remains for me is a headache.
      • Something very similar showed up also on DSi [youtube.com]. It generally feels almost as great as when you hold a "proper" hologram (unfortunately only static ones have good quality so far; for good holographic dynamic display we need pixels smaller than the wavelength of light + processing and memory we're nowhere near yet - but once there, a screen could be felt essentially like a window or mirror)

        Stereoscopy feels very flat to me in comparison - yes, there's depth of course, but the usual very deep focus makes ever
      • Re:Smart... (Score:5, Informative)

        by JustinOpinion (1246824) on Tuesday April 12, 2011 @10:15AM (#35793036)
        To recap [slashdot.org] the reason why current 3D implementations are lacking (and thus give some people a headache)... Humans use at least 5 tricks to determine the three-dimensional makeup of a scene:
        1. Focal depth: based on how much the eye's lens has to focus
        2. Convergence: based on the slight differences in pointing of the two eyes to a target
        3. Stereopsis (static parallax): based on the slight differences between the left and right eye images
        4. Motion parallax: based on the different displacements/motions of objects at different distances (e.g. as you move your head)
        5. Visual inference: reconstructing using cues like occlusion, lighting, etc.

        In the real world, all 5 of those systems work in concert, giving you a consistent understanding of your environment. The problem with modern 3D implementations is that they only trick you using only two, or maybe three, of the above. For instance "3D glasses" are showing you different left/right images, creating fake stereopsis, but the focal distance is still "to the screen" and doesn't match the apparent parallax-based distance to objects. So your brain is rightly confused because the various systems are giving conflicting answers. Amazingly our brains have no problem looking at two-dimensional images like pictures and conventional movies: in such cases 1-4 don't work, and our brains instead just use #5 to fully reconstruct/guess at the three-dimensional nature of the scene. A few optical illusions notwithstanding, this works remarkably well.

        What's presented in the above video is fooling your brain using motion parallax. This is neat, but because the image is still flat, your other brain systems (1-3) will be giving a different answer and so the illusion won't be perfect.

        In principle we could combine techniques to make for a more convincing sort of 3D. E.g. combine motion parallax (eye tracking) with stereopsis (3D glasses). But it won't be truly convincing (and thus headache-free) until we fully reconstruct the three-dimensional light-field that should properly be emanating from the virtual objects. Doing this requires some very good holography, to fully reconstruct the required light waveforms, or something like anisotropic pixels that can control their emission as a function of viewing angle. In the meantime, 3D will remain a bit of a gimmick outside of some niche applications.
        • by sznupi (719324)

          For instance "3D glasses" are showing you different left/right images, creating fake stereopsis...

          A really "fake" one at that - as in, working in a "wrong" (not merely "incomplete") way: when the time comes to look at an object which is at a different depth (which is pretty inevitable - that's the point of "3D"), there's none of the "dance" of translucent scenery in front of and behind the momentary focus plane.

          Also, in "2D" images and movies, a shallow focus can give a nice illusion of many effects... so it's also partly #1 and #3 (this shallow focus often resulting in a kind of translucence of obje

        • But it won't be truly convincing (and thus headache-free) until we fully reconstruct the three-dimensional light-field that should properly be emanating from the virtual objects.

          A big part of the problem is that your inner ear stays motionless while the 3D movie/game depicts motion. Most people's brains have difficulties processing those conflicting information streams, resulting in motion sickness (headache, nausea).
          So even holographic 3D would be problematic once it starts moving...

          • Most people's brains have difficulties processing those conflicting information streams

            Well, you could always start killing everyone who experiences the problem then after a few generations everyone will be able to process conflicting information streams easily.

            • by sznupi (719324)
              Huh, "killing"? Helloooo, what about the established practice of sterilization?

              (so crude, such lack of subtlety and finesse... you have no chance of becoming the hegemon. Good, less competition for me)
          • by sznupi (719324)
            From what I've seen, motion sickness in trains doesn't really happen when they roll slowly on perfectly smooth track, when you can't feel movement - but when the scenery outside the window (very 3D) moves.

            I don't get headache, nausea with stereoscopy; can easily see it instantly with "rough" methods, the "crossing eyes" one for example, also for such videos. It just looks weird (even unreal & more flat) & wrong to me, and I can hypothesize why (other responses under top GP); the stimuli is confli
        • by sh00z (206503)

          But it won't be truly convincing (and thus headache-free) until we fully reconstruct the three-dimensional light-field that should properly be emanating from the virtual objects.

          I've seen that 3D box. It's called a "puppet show."

      • by egamma (572162)
        Flashback:

        I still believe that these 'talking' movies is one of those things that is interesting for a while, and then you move on.

        (a few years later)

        I still believe that these 'color' movies is one of those things that is interesting for a while, and then you move on.

        (2011)

        Dunno about that. I still believe that 3D is one of those things that is interesting for a while, and then you move on. The only thing that remains for me is a headache.

    • OpenCV already provides you the necessary building blocks.
      And the best part : it's a library in C, and can be used on almost any device under the sun (not restricted to iOS).

  • Hey! (Score:3, Informative)

    by Addict7 (2024042) on Tuesday April 12, 2011 @09:44AM (#35792630)
    2008 just called, he wanted to share this link http://www.youtube.com/watch?v=7QoaklieUho [youtube.com]
  • So... Holotoy? (Score:4, Insightful)

    by MBCook (132727) <foobarsoft@foobarsoft.com> on Tuesday April 12, 2011 @09:44AM (#35792638) Homepage

    So it's the same thing as Holotoy [kode80.com], which has been out since slightly after the iPhone 4 release? This isn't new, the only story here is that it's on the iPad.

    Now Holotoy's effect isn't perfect, and the larger screen may help. I would be interested in seeing this in person, but it's not unique.

    • by greeze (985712)

      Holotoy uses the accelerometer and calculates the viewing angle based on the iPad's orientation. This uses the camera and calculates the viewing angle based on the viewer's location in space. The main difference is that with this, you can put the device on a table and it'll adjust the angle as you move around it. You don't have to be holding and moving the device for it to work.

      • by MBCook (132727)
        Since an update just after the iPhone 4 was released, Holotoy has a similar mode where it can use the front camera to do face tracking, simulating 3D the same way this program does.
  • ...but this is cool. Disclaimer: I'm not a fanboi
  • Making something that looks 3D and making something that either does, or appears to project a 3D image into space are two different things. This is nothing more than drawing a picture with perspective... next, please.
    • by GooberToo (74388)

      Not to mention, there is absolutely nothing new or novel presented in the video.

      • huh, did you watch the video? yes it is a picture with perspective, but the perspective changes as you move your head.
        • by GooberToo (74388)

          And? Some of the first tech demos I became aware of were done on Android over a year ago and even then, I *seriously* doubt Android was the first implementation. I strongly suspect this was first done a decade ago or more elsewhere on non-commodity hardware. Hell, the android tech demo I saw was even more interesting because they were combining a projected building with real-time video from the on board camera and aligning the rendered building with the video feed.

          Really, the only thing first here is its fi

        • by Anonymous Coward

          huh, did you watch the video? yes it is a picture with perspective, but the perspective changes as you move your head.

          There was a Nintedo DSi downloadable game that used the DSi's camera to track the head movement and display a very similar 3D effect. I don't know the name of it, but it only came out in Japan. Here's a link to a video of it in action.

          http://gamevideos.1up.com/video/id/28532

  • There was a game on the ipod touch years ago called Word Fu that did something similar without a camera, when you tilted the ipod your view of the dice would change so you could see the sides...it was subtle and not even an advertised feature but it became a rather mind blowing demo to show off to friends. It was accelerometer based only but combined with head tracking it could be an amazing new way to play games if it gets support.

  • by Godai (104143) * on Tuesday April 12, 2011 @09:53AM (#35792754)

    There was a story on Slashdot a while ago (last year? later?) about how someone did something similar on the Wii: http://www.youtube.com/watch?v=Jd3-eiid-Uw [youtube.com]

    I notice the 3D target thing is the same (I mean, EXACTLY the same) but maybe that's some kind of pre-packaged 3D test everyone uses? Though I noticed that the first example in the iPad used what were clearly Mii faces; is this just an adaptation of the Wii guy's work? Granted, the Wii guy had to strap the controller on his head, so they'd have to use the facial locating software. Ironically, that's what Nintendo's had since the DSi (and greatly improved in the 3DS). In fact, I *think* that's how the 3DS works -- using this trick. I might be mistake on that, but I'm pretty sure it uses facial location; I just can't remember if that's all it needs to get the 3D effect or if it uses other stuff as well.

    • by redkcir (1431605)
      Johnny Lee use to work for that company (EHCI), but he doesn't seem to be on the roll anymore. You are right, old tech in a new box.
      • It's a research group under the publicly funded research institute CNRS. This is French people tax euros at work (with likely a sprinkle of European funds).
        • by redkcir (1431605)
          As this is targeted at for the Ipad and doesn't seem to be open source and was used on the wii first (a Nintendo product), it looks like a ripoff of Johnny's work by the French for profit. Sounds like a company to me. Just saying.
      • by ^BR (37824)
        Also Johnny Chung Lee used to work at the HCII lab of CMU, not there.
        • by redkcir (1431605)
          Woops! Looks like I got some of my letters wrong. Still, it was done on a wii first, so still old tech on a new box.
    • No, the 3DS shows an interlaced image, and has a polarized layer over top that points the right lines at the right eyes.
      • by Godai (104143) *

        Ah, ok. I knew they had facial recognition (well, location) software form the DSi. Someone told me it was doing the trick outlined in the article, but I guess not. I did wonder why that would produce the double-vision you get from the side :)

        Though I *think* someone had this trick working on the DSi. I could have sworn there was a Slashdot article where some game company in Japan had used the 'locate the face' trick to create the effect. If I'm remember right, that might be why I got it confused with the 3D

        • I think you may be right, I think someone DID do this on the DSi, whether it was an actual game company or a homebrewer, I'm not sure.
    • by sznupi (719324)
      Nope, "3D" in 3DS is just the old trick of stereoscopy (old: barely younger than "2D" photography... so largely ignored - with few "golden eras" - for quite a while now); the barrier which shows each eye a separate image is just in the screen - not in, say, glasses. Which... makes it useless with the discussed trick.
    • by buback (144189)

      John Carmack did this ~ten years ago, but never released it, saying it was too slow (at the time) to be used in any ID games.

      it's the parallax effect, which was also used in the 2D Duke nukem games 20 years ago.

      but yeah, Johnny Chung Lee made it one of the best implementations i've seen so far, using a wii sensor bar and some IR LEDs.

      it is a much more convincing form of glasses-free 3d, and it doesn't give you headaches from eye strain

    • by sootman (158191)

      I would pay so much if someone would make a game like Time Crisis [imageshack.us] with this tech. I've been waiting for a game to be made like this ever since I first saw that Johnny Lee demo years ago but no luck yet. That kind of responsiveness plus a wall-size projection would be so awesome.

  • The wooden looking Labyrinth game that came out years ago on the iPhone/iPod Touch did this with just the accelerometers. It's a really cool in game effect, and I'm glad they made the change to use the cameras, but without some API worked into the next OS release, this seems to not be too useful.

    Also, what's the CPU load with this running? Is it the kind of thing that takes a lot of processing power, or is it something that can be done in the background of a game?

    • I'm sure it's perfect for the iPad 2 and iPhone 5 as they are dual-core so there should be plenty of background CPU cycles to add in the head-tracking.

  • http://www.youtube.com/watch?v=lEMkgVnzvdE [youtube.com]

    This game also uses an embedded camera to track your head relative to the system, enabling you to "look inside" the display. It's very fun, though it requires a near perfect environment for your head's background to ensure the tracking works properly.

  • One basic method used by almost all animals with stereoscopic vision is to look at the scene using two eyes and use the difference to perceive depth. Human babies apparently achieve this remarkable feat between three and nine months of age. (Determined by placing babies face down on a glass ledge and checking to see if they feel alarmed. When the babies have mobility they checked to see if a baby would be wary of an apparent chasm and stay on the right side of the ledge. No babies were harmed in the researc
    • We also constantly perceive parallax (hence we perhaps use it - at the least, I would be disappointed if our brains didn't evolve to use such handy info) - the "doubling" of objects which are not currently on the focus plane. The rates of it very dependent on the distance from focus plane (so I suspect it might be a more useful info than the parts on which "3D" / stereoscopy depends on - allowing the brain to feel the depth of some object before focusing on it is, to know how to "set" the eyes is... kinda u
    • by daid303 (843777)

      We don't need depth perception for games and stuff. We need this.

      I fondly remember Doom and Duke3D, and all the times that I stood up just to look down into my monitor to see how deep a pit was only to break the illusion of 3D.

    • by itsdapead (734413)

      One basic method used by almost all animals with stereoscopic vision is to look at the scene using two eyes and use the difference to perceive depth

      ...but if you try to fake this with stereoscopic images (e.g. never-to-be-sufficiently-damned 3D movies*) you set up conflicts between the stereoscopic sense and all the other methods (e.g. focal distance, head motion) and give people headaches. Plus you have to wear glasses.

      From what I've seen of the similar accelerometer technique used in the Labyrinth app, this technique might be a bit easier on the eye...

  • I have decided. I am going to get a cat and an iPad with that software on it, and then I'm going to watch the world burn.

    • by sznupi (719324)
      Simpler: hollow-face [wikipedia.org] & three dragons [grand-illusions.com]

      Actually, inspired by how one preschool-theater costume (of a...cat; with proper ears) supposedly induced a panic attack in the kitten of my buddy, I essentially reworked the dragon once, to be more "danger! Possible unknown big cat!"-like. Yup, panic attack also in my cat.

      (quick google search for the above wiki page even revealed one with a cat design [blogspot.com]... I can't vouch for how convincing it is, though)
      • by AvitarX (172628)

        I had a cat, and my girlfriend put on a cat ear headband for a lazy Halloween costume, the cat growled, then ran away and hid for hours.

        Otherwise the cat was fearless of people.

        • by sznupi (719324)
          I guess one can say her costume was quite unique & unusually good... actually inducing some sort of fear, for once ;p

          And hey, no reason why cats shouldn't use such cognitive shortcut as the presence of cat ears.[1] I'm fine with it as long as cats are (supposedly; nvm the means) unusually good in generally realizing that humans are not cats; in comparison to, say, dogs. Even how the cats are almost certainly more stupid from them overall (and no wonder, cats don't have to function in groups), is more
  • Anything done on an Apple product is new and innovating....

    Actually....no.

    http://www.youtube.com/watch?v=Jd3-eiid-Uw [youtube.com]
    (Mind you. You would look pretty stupid with a wiimote taped to the side of your head!)

    I imagine most devices with a front facing camera can do this.

    There is also this: http://johnnylee.net/projects/wii/ [johnnylee.net]

    Curb you enthusiasm fanbois nothing to see move along.

    • by ^BR (37824)
      Well the innovation here is obviously not the 80s video game 3D rendering but the fact that it tracks the user movement using just the camera, without the user needing to wear a ridiculous looking "tracking bar" mounted on safety goggles, if you don't see the innovation there well, can't help you.
  • FaceTrackNoIR have been doing this on the PC with just a webcam for a while. It's especially good in simulation games. http://facetracknoir.sourceforge.net/home/default.htm [sourceforge.net]
  • Someone 3rd party jailbreaking hacker scum just stole Steve's "One more thing..." for iPad3. Release the app mods! Rape their goats and kill their women!
  • It looks really cool* and I'd love to get this App on my new iPad 2 :)

    The next step which I've been looking for but haven't seen yet is some means of a user "grasping" objects in 3D space. While the technology doesn't exist for a true haptic interface, you would imagine someone could use two cameras to calculate the 3D position of a user's fingers (or other body part!) or some sort of 3D depth sensor (how does the proximity sensor on an iPhone work?). So while this may not work on this generation iPad pre

  • by Anonymous Coward

    1) Set device on a hard surface (a desk or a table).
    2) Look at the very top of the device -- the very, very top edge of the device.
    3) Keep moving eyes up. There, the whole world should now be in 3D.
    4) If the 3D world is blurry do not adjust your monitor -- go see en eye doctor.

  • So, when I lean one way or the other in FPS's trying to look around a corner, it might be useful instead of just funny looking?

    more likely it'd be useful and funny looking, I'd guess....but at least some improvement ...
  • Apparently "GoPro" camera's have made it possible to shoot your own 3D film's, on their website in states: "The 3D HERO System allows you to combine two 1080p HD HERO cameras into a single housing to record 3D video and photos while simultaneously recording in 2D. A synchronization cable plugs into the rear HERO Port on both cameras to join them together, enabling both cameras to record video and photos in perfect synchronization." So 2 of their 1080p HD camera's would run you around $270 each, plus the 3D
  • I wonder, though, just how important head or eye tracking is? When you're viewing an iPad, I'd think that your head and eyes are pretty fixed. I noticed that in the demo, it was the iPad that was moving, not the viewer's head. Of course, that might just be because that's the best way to shoot a demo video - we might get seasick if the camera was moving around to mimic head movement!

    But, given that a typical user isn't likely to move their head around while viewing an iPad, why not simply use the gyroscope (

  • Actually I think that might be the place where they got the idea, it looks very similar: http://xk3d.xkcd.com/ [xkcd.com] More seriously, this will not solve the headache problem caused by the eye wanting to bring stuff in- and out- of focus and failing because it's all already focused.
  • if this system can locate your eyes in 3d space in front of the monitor, then it should be able to compensate for the distortion of looking at your monitor at a angle. that requires not only a change in the viewpoint like you see in this video, but also distortion to compensate for the plane of the monitor being closer to your eyes at some corners. thats why most of this video dosent diverge from a head on view much. they didnt bother to code this in. it just changes viewpoints.
    • Agreed. It would work as is if the camera was detached from the display and the display rotated to always be in the normal direction.

The flow chart is a most thoroughly oversold piece of program documentation. -- Frederick Brooks, "The Mythical Man Month"

Working...