Follow Slashdot blog updates by subscribing to our blog RSS feed


Forgot your password?
Graphics Input Devices

Portable, Super-high-resolution 3-D Imaging 31

An anonymous reader writes "At SIGGRAPH 2011, a team of researchers from MIT presented a clever method for measuring microscopic surface structure using a rubber sensor, a camera, and a set of lights. This technology could have applications to industrial inspection, dermatology, and even forensic ballistics."
This discussion has been archived. No new comments can be posted.

Portable, Super-high-resolution 3-D Imaging

Comments Filter:
  • by jkcity ( 577735 ) on Thursday August 11, 2011 @09:06AM (#37055024) Homepage

    You can find some more demonstration videos here [] while they are intresting to watch. I am not sure I'd class the device as portable looked quite bulky and looked like it needed alot of power.

    • by plover ( 150551 ) *

      The device they were using at the end of the video in TFA showed a cylinder about the size of a soda can. Looked plenty portable to me.

      • It also looked like that was only necessary for 3D - it looks like a purely 2D version could be made much smaller, given it'll consist of the gel, a fixed light source, and a camera.

        • by siglercm ( 6059 )

          Parent comment was moderated "underrated?" Underrated??? Do you not know a sardonic remark when you read one?!

    • by Anonymous Coward

      It doesn't need a lot of power. The camera and the LED inside of the portable device could connect to the computer with a simple USB cable. How much power do you think the USB port could give out? Also, Comparing to a lot of devices measuring the surface structure, GelSight portable device is really small and easy to carry.

  • by JoshuaZ ( 1134087 ) on Thursday August 11, 2011 @09:11AM (#37055080) Homepage
    According to TFA even older, cruder versions of this equipment can "detect the raised ink patterns on a $20 bill." TFA also notes that the way this works is to make the optical properties of the surface of the object essentially uniform. To some extent this is an extension of existing technologies. There are a fair number of imaging techniques that involve doping the surface of something before trying to image it. For example, when one is using an electron microscope one frequently dopes the target with some helpful metal first. But this is a lot easier to run than that and works on the optical range. And the cameras used themselves seem to be not that expensive. This seems a like a really potentially very groundbreaking technology. And it is a good example of how sometimes the technologies that show up and change things aren't technologies that any futurist or science-fiction writer has anticipated. I'll be very interested to see where this technology goes in ten years.
    • by AJH16 ( 940784 )

      Agreed, sometimes people have an idea that is genuinely a new thought and direction that nobody really saw coming. Sometimes they work out sometimes they don't, but they are certainly critical to innovation and it is cool to see something like this. I wish I could see the youtube videos while at work.

      • I tried to create some simple 3D video footage a while ago using webcams and a toy video projector projecting De Bruijn sequenced colored bands. It occurred to me that things would be a lot easier if everything in the world were a uniform dusty-white. I'm thrilled that they've discovered a way to make the universal dream of a uniform dusty-white world a reality.
  • This technology could have applications to industrial inspection, dermatology, and even forensic ballistics.

    And, don't forget counterfeiting.

    • This technology could have applications to industrial inspection, dermatology, and even forensic ballistics.

      And, don't forget counterfeiting.

      Also Anthropology, and specifically archaeology. You could take a perfect image of every artifact at the dig site.

  • Some what similar (Score:4, Interesting)

    by pieisgood ( 841871 ) on Thursday August 11, 2011 @10:34AM (#37056270) Journal

    This is somewhat similar to how some artists obtain normal maps for textures.
    They go and take source pictures of materials (bricks, grass, what ever), then for each photo they bring a light and shine a light on the scene from four different directions. This allows you to obtain the height information later on back at the PC. This gives you a realistic normal map for the surfaces you've been taking references of. Now, the problem being that this only works with surface deformations of a certain height. Also the angle you place the light at determines how "bumpy" the normal map is going to be. I presume those issues are worked out in the MIT device, but this works for nearly flat surfaces best.... anything else and you'll get "shadows" of objects overlapping each other.

    The resolution and scale are what most impress me here though. :)

    • Shininess also wrecked the joint up when I tried to play with it. A series of book spines, a wonderful map to have for all occasions, looked alright on older fabric books, but anything with metallic lettering or made of slightly shiny synthetic material produced terrible distortions.
    • You can easily compensate for light placement if you've got a target object of known shape (and optical properties) in the scene. I use a three-sided pyramid with a right-angled apex; it makes the maths almost trivial, and it's very easy to fold a decent one out of printer paper if you haven't got a premade one handy.

  • Why so quiet? (Score:4, Insightful)

    by djlemma ( 1053860 ) on Thursday August 11, 2011 @10:40AM (#37056354)
    This is so damn cool I don't know why there aren't more comments. I guess it's because there's no real controversy here- it's just cool tech. Quick, somebody claim that this is going to create an invasion of privacy! Or that it will cause climate change!
  • Considering how detailed the fingerprint are, I suspect this could be a boon for police and similar when it comes to getting fingerprints from people. No more messy ink - instantly digital copy.

    But I wonder how well it'd work for picking up fingerprints from objects.

    • They already have inkless fingerprint scanners. Heck there is a fingerprint scanner on most thinkpads for years that doesn't require ink.

      This may be used to get fingerprints from a crime investigation though. However the way it works one shake of your arm you might rub it out. It is cool but it also right now requires a lot of extra physical contact with what it is trying to get a picture of.

      • [...]one shake of your arm you might rub it out.

        Look, I realize this is /., but I'm certain all of us here can last longer than one shake!

    • That's actually mentioned in the fine article. They have been approached by fingerprint specialists, but this stuff is on a scale way smaller than you'd need for fingerprints.

      • Although, there is a small portion of the population that doesn't have fingerprints.. and this scanner has so much detail, it might still be able to capture identity for such people. Also, I think I remember a story about fingerprint scanners being used in a school to take attendance, and some kid figured out how to fake a fingerprint using gummi bears (or similar). This would probably allow for enough additional accuracy that forged fingerprints would be much more difficult.

        Yep, here's the gummy bear t []
  • What I would like to see is something consumers could use that would scan rooms or objects and create them in 3D space ala some of those Kinnect videos, except you know, usable by the common man (or is there already?). Something comparable to what Johnny Chung Lee did with the Wiimote. Put this into the hands of the public and you will have some pretty interesting uses. I'd think it would be perfect for those working in real estate or remodeling, quickly and easily getting a 3D model you could edit.
  • by Richard Kirk ( 535523 ) on Thursday August 11, 2011 @11:31AM (#37057184)

    Gels have been used to capture very fine metallurgical detail. However, the normal route was to take a print of the surface, coat the gel in gold, and then view it in an electron microscope. This was handy when you had a large sample in some sort of fatigue test apparatus, and you wanted to monitor the growth of a crack in nanometers. This GelSight process will not be able to rival electron microscopy, but you might be able to measure sub-micron features with microscope optics.

    This seems to be just the range of scales for looking at human-made objects. If you find a bit of bone or a bit of pottery, this can give you detailed information on how smooth the surface was, how it was broken, and stuff like that. You could capture the accurate shape of coins, and work out which ones probably came from the same die. The portable version could be used on a dig, or in a museum. There would be little risk of contaminating the object if it was solid enough to be picked up with gloves. This could show up all sorts of things. Neat.

    NB: fingerprints are usually pictured using similar optics, but using a rigid piece of glass rather than a prism. Light will reflect internally off the glass surface, but the reflection is interrupted where the finger touches the glass, giving you a good high-contrast image. I have worked a bit on these.

    • The interesting part here about fingerprints would be that this could allow for better imaging of them over a much larger range of pressures. Meaning that more accurate results of searches could be done. While that doesn't solve anything new it could make things go much smoother than I've seen in the past. i.e. User puts finger onto glass too hard thinking it'll help and then get frustrated when it doesn't work and keep trying harder when only a mild touch is needed.

  • Go for the money shot .. as it were. I wonder though, would patients prefer a hard camera tube, or clammy gel shoved you know where ?
  • What the metal coated gel does is that it temporarily paints the 3D surface with metal.. Which is then photographed by camera from different angles. The 3D contours could anyway be photographed from different angles without the temporary metal paint that the proposed gel adds. And the multiple photographs could be combined to generate a 3D version. Is it some special kind of metal, that supports/enhances 3D imaging? Are the cameras special that work only with metal. Independently, we also loose the color

The rich get rich, and the poor get poorer. The haves get more, the have-nots die.