Follow Slashdot blog updates by subscribing to our blog RSS feed

 


Forgot your password?
Close
typodupeerror
×
Technology

Using Lasers And Range Finders To Digitize Objects 50

Posted by CmdrTaco from the now-isn't-that-clever dept.
esoteric0 writes: "Those boys at Stanford are at it again: They created some new algorithms for 'combining multiple range and color images, allow us to reliably and accurately digitize the external shape and surface characteristics of many physical objects.' " It's not just a mouthful -- they've created a cool digitized version of buildings, maps, and Michelangelo's David. Ever wonder what his toe looks like when digitized at .05mm?
This discussion has been archived. No new comments can be posted.

Using Lasers and Range Finders to Digitize Objects More

Using Lasers and Range Finders to Digitize Objects

Comments Filter:

  • Flashback... (Score:3)

    by Paradigm Lost ( 179221 ) on Saturday July 08, 2000 @03:24AM (#949341) Homepage
    Whoa! Did anyone else have a sudden flashback to Tron? I can almost see Michelangelos David in one of those glowing blue outfits.

  • If we only could... (Score:4)

    by The Grammar Jew ( 208575 ) on Saturday July 08, 2000 @03:25AM (#949342) Homepage
    digitize certain young actres...and then release the data under an open source [slashdot.org] license...the possibilities are, shall I say, endless...
  • Yes! Finally I can get digitized and take on
    Master Control myself! But really, this technology
    has been around since 1977. Disney brings
    us one of those excellent "historical documents"
    refered to as Tron. There we see digitizing
    people has been around for a lot longer then
    some stafford punks would like us to believe.

    Anyhow, I am off to suit up and get ready for
    a cycle race or two and hopefully I can slam
    my disc right through the MC. Remember, the
    users have all the power!

  • Pragma: nocache & don't email it (Score:3)

    by h2odragon ( 6908 ) on Saturday July 08, 2000 @03:30AM (#949344) Homepage
    From the "more images [stanford.edu]" page:

    "The images of Michelangelo's statues that appear on this web page are the property of the Digital Michelangelo Project and the Soprintendenza ai beni artistici e storici per le province di Firenze, Pistoia, e Prato. They may not be copied, downloaded and stored, forwarded, or reproduced in any form, including electronic forms such as email or the web, by any persons, regardless of purpose, without express written permission from the project director Marc Levoy. Any commerical use also requires written permission from the Soprintendenza. "


    Hmph. Let's email the picutes to HIM. :)
  • Of course, we all know what will end up being one of the primary uses of this technology: pr0n.
  • Sorry, that was the wrong link. Can't find the right one now, but believe me, it's funny.
  • Yea, Ars [arstechnica.com] ran it. It's old enough to be #2 in the top row.

  • Ars Technica has a feature (Score:3)

    by Segfault 11 ( 201269 ) on Saturday July 08, 2000 @03:41AM (#949349) Homepage
    Ars Technica [arstechnica.com] has a feature on this subject entitled: `Michelangelo Goes Digital' [arstechnica.com]
  • Here it is: Something ate my child [passagen.se]. Did I promise too much?

  • Re:Pragma: nocache & don't email it (Score:3)

    by The Grammar Jew ( 208575 ) on Saturday July 08, 2000 @03:51AM (#949351) Homepage
    Yes, they own the right to those electronic reproductions of Michelangelo sculptures because they produced those creproductions. It did cost them time and money. Please don't expect they'll share these reproductions with you.

    If you want your own reproduction, go and make one. Michelangelo's copyright, if any, has expired. No one is hoarding David. It's here, open for everyone to copy.

    OTOH Harry Potter's hasn't, so you can't OCR his books. Sorry.

  • Wow, I'd better not even go see that page - cause if I did, technically I'd be d/l'ing the images which is strictly verboten. Especially since my box accesses the web through a Squid proxy server, and I just can't help it if it stores it on there afterwards. Sheesh, I think they made it illegal to go to their site :)

  • Re:Ars Technica has a feature (Score:3)

    by tcomeau ( 114361 ) on Saturday July 08, 2000 @03:58AM (#949353) Homepage
    And NPR did a wonderful interview with Marc Levoy last month, available on their web site as David's Eyes [npr.org] which includes things that the laser scan missed, and the news that all of the art history books are wrong.

    tc>

  • Buy it, then destroy it (or do whatever you damn please with it). Thank you.

  • Applications in Space Exploration (Score:2)

    by Anonymous Coward
    Maybe this process would be useful in mapping extraterrestrial terrain. It might allow detailed analysis of a planet's surface from great distances.(?) I wonder if using a different frequency for the laser could cut (like X-rays) through thick, vaporous atmosphere allowing scientists to get data from hidden features on cloud-covered planets. This process has some interesting applications...
  • Interesting and cool, but will need orders of magnitude better accuracy and precision before it could be a viable replacement for commercial coordinate measuring machines such as the ones built by Brown and Sharpe [e-metrology.com]. They say they can "split the micron" which is pretty impressive.


    Their CMMs are for a completely different market though...more manufacturing oriented.

  • Well, this is certainly kind of cute, but isn't it just adding some sophistication to the process done by those companies (e.g. tomra [tomra.no]) using lasers to scan things for pattern recognition? Just wondering, I'm not too into this kind of stuff... :-)
  • The Univeristy of Mass. (Amherst) has been doing some cool stuff with 3D modeling of images too. Check it out.

    Ascender II Project [umass.edu]
  • 3D scanners using radar, lidar, the works have been around for a long time now...so what's new here?

    The visual effects industry regularly makes models and puts them in a 3D scanner to get a basic mesh to work from. Also, the vfx industry regularly takes full lidar sweeps of outdoor sets to more easily do match-moves, make mattes, and such. For example, see: http://www.vfxpro.com/.getarticle/.772 954741 [vfxpro.com] and http:/ /www.digitalpostproduction.com/Htm/Features/ScanMa ster/ScanMasters.htm [digitalpos...uction.com]

    Heck, do a search for 3D scanner, and you come up with tons of hits...here's a couple:

    So can someone please tell me what the big news is? Is is the resolution, I take it?...That IS pretty small and pretty cool...

  • X-rays? Nah, try radio. Venus was mapped with a radar (the link [google.com] :)
  • I also know of a company, InSpeck [inspeck.com], which makes 3D scanners for objects and humans but without using lasers, just white light.
  • While the intracasies of US copyright and intellectual property laws may be bad enough, as far as I can tell, this stuff would be covered under Italian law. While the actual legalese clause cited above is most likely a symptom of overactive paranoia, it is possible the project does have the rights it claims. Presumably, they might have recieved rights to this specific digital form of the pieces from their respective owners under Italian law (the museums involved, Michelangelo's relatives, etc.), or perhaps Italy has laws in place covering just such an eventuality.
    While the point that this is silly is well taken, I don't quite think this is worthy of knee-jerk copyright paranoia or, say, a Jon Katz writeup, until more information comes to light.

    And 'mid this tumult Kubla heard from far
    Ancestral voices prophesying war!
  • I've always wanted something that I could just walk around with, or mount on my car, that would just digitize everything I saw. Give it a GPS and link it to other units owned by other people, so eventually every piece of an entire city might be "Seen" by one of these critters, and it'd all be mapped on a central server farm. If all the devices can recognize corners and distance and stuff, and automatically break down the geometric primitives that make up our world, and assign the appropriate textures to them, we'd have...

    Automatic Quake maps for every place you've ever been!

    Not to mention, instant surveillance for the spooks. Forget taking a few pictures and trying to by-hand extrapolate where the secret passages in a building are. Just wander through it with a mapping briefcase, and in the resulting data look for unusually thick walls and stuff.
  • DUDE, IF MY WIFE EVER FINDS OUT I HAVE LOOKED THERE [ef3.com], SHE WILL KICK MY ASS.

  • It's been done to the good ol' Earth a number of times (including doing it from space). There are prospecting companies that use exactly this kind of detailed information in their search for natural resources (oil, minerals etc). Geologists use the same approach to search for ancient craters - even below water using sonar.
    Since it's just radar and sonar all over again, most of the underlying technology has been around since WWII - it's just been refined to a point few probably imagined possible or worthwhile.

  • Not quite there [digital.com], but promising.
  • Yeah, it looks like I was probably right in the "new story" aspect being the resolution. For example, the Inspeck device has a resolution of 1.1mm. The Stanford people's research at 0.05mm blows that away...but still, I don't really know what's new about it, mostly because I'm having lots of trouble getting through to the linked document. Looks like graphics.stanford.edu wasn't ready to be slashdotted. :(
  • Nice pictures, too bad it requires enabling the potentialy evil java to open them up.

    Why do more and more people insist on using java for a simple link to a picture? To prevent a wget by a happy surfer? To play games? Me thinks a patch to wget for snarfing addresses out of javascript is in order so one could assemble a viewable page for one's own viewing with a simple browser.
  • Who here sees the story of Tron today as an oppressed Microsoft universe that is eventually freed by hurling that Linux install CD into the MCP?
  • you know it's comming soon. Just remember--reindeer flotilla!
  • It seems that these guys are attempting to use their meshing and modelling routines to find inconsistencies and texturing details in the actual stone

    This type of work (attempting to detect very small scale irregularities in materials) is far different than modelling an item and creating a 3 dimensional picture of that object, or creating a flashy new quake 3 skin .

    Kudos to these guys for using (and creating) some really complex algorithms. Being in the field of 3D visualizations , I can appreciate the amount of time, energy and brainpower which go into a project such as this.

    *penguin_nipple stands and applaudes*

  • To the article submitter who referred to "Those boys at Stanford", have you ever thought that maybe, just maybe, some of those Stanford researchers might not be boys?

    Now the project link [stanford.edu] appears to be slashdotted, so I can't actually confirm that the project doesn't feature all "boys", but I suspect that this is not the case. And even if it is, I think it would be better not to emphasize this in light of some of the recent Slashdot articles about gender issues [slashdot.org].

    Maybe I'm overreacting, but I'd like to see the day when more than 5% of my Computer Science class is female, and I don't think girls hearing about "Those boys at Stanford" will help that.

  • What is the gravity there on Venus compared to Earth?
  • I've worked with just such machines, only on a smaller scale, and I have one fundamental complaint, the resulting object becomes imported as a mesh.

    Oh, fine you say, thats the same as 3D studio uses, thats swell, right? wrong.

    During the past year I have painstakingly reverse-engineering the childrens toy the bumble ball. This was done all in Mechanical Desktop 4 and Inventor 2 [autodesk.com].

    Because of the details involved with the interior mechanics, I, along with my partner, when we turned to 3d studio to produce an animation, we found that mesh is nooo substitute for extrutions and constraints. Since the entire bumble ball is round, and all its features as well, the triangle constructions were autrotious. When we made a 3d model out of it using Stereo Lithography (SLA) it wasn't at all as nice as we had expected.

    And then I looked over to all the other groups who were also reverse engineering things, and saw they're troubled 3d studio projects, and i simply refused to use it. I was happily rewarded when I was given the oportunity to use inventor 2, which, in accordance with autodesks file formats, is built on extrutions and constraints. Mmmmm...

    Later into the year, when we were nearing completion, i saw what some students had '3d scanned', and it was a mess. What happens is that the scanner doesnt pick up a chamfer or a c' sink hole or an array, the object imports as a mesh. Mesh's might be swell to look at, but they're worthless to work with.

    Ok, i'm done ranting. Oh, and by the way, I'm a sophmore in high school. :)

    /nutt

    ---
    And please dont comment on my spelling..

  • Take the USA for example. The Constitution says that authors' exclusive rights must be "for limited times"; it doesn't say how limited. Limited to 999,999 years? Still limited.

    The Walt Disney Company has been taking advantage of this loophole for years. Every time the copyright on early Mickey Mouse cartoons gets close to expiring, Disney just buys a 20-year retroactive extension.

  • I can't find it but am i the only one that remembers this being posted WAAAAAAAY back? Months and months and months ago... They were still doing the work, but.. same story...
  • There are two basic approaches to what I call "3-D immersive imagery": photographic (with zooms as jump-cuts or with scanning/panning), and virtual reality. As everyone who's ever worked with it knows, the problem with VR is building the model -- in order to move around, for example, Philadelphia, some poor sap has to run around modeling the whole thing. Projections can't be screened until a huge database exists. This technique sounds as if it can help build (automatically) these databases. (Remember when the protagonist in Cryptonomicron ran around making measurements of... was it Manilla?)
  • I just want to say that GL Tron [gltron.org] is really cool. The lightcycles in it are better than the original movies!
  • If something is in the public domain, derivative works are equally public.

    That's what public domain means.
  • This is Marc LeVoy's thing. He came to Stanford with a background in reducing data from medical volumetric scanners, either X-ray (CT scanners) or nuclear magnetic resonance (MRI scanners). He's big on the problem of reducing vast amounts of scanner data into some useful form.
  • What's the purpose of GPL, then?
  • Your point that meshes are worthless to work with is nonsense.
    You have problems with 3D studio? Perhaps they have nothing to do with meshes. 3D Studio is not the best program for mesh-modeling anyway.

    mesh is nooo substitute for extrutions and constraints.

    You can do extrusions with meshes, they are one operation of many that you can do on meshes.

    i saw what some students had '3d scanned', and it was a mess

    They didnt tweak it manually? Then of course its a mess. There is no 100% automatic way to 3D-scan things. The stanford article mentions the places that are hard to get to. This has got nothing to do with meshes. BTW, we (Hammes software) model everything manually and dont scan things, thats faster for us, since we dont need high resolution.

    Oh, I am not the only one liking polygons:

    Just last weekend I was at LinuxTag, the largest european Linux exhibition. There was someone from blender giving demonstrations. All the demonstrations invloved polys. He himself said that for most of the jobs he prefered them to other modeling methods.

    In the newest edition of the German magazine "3D Live" there is a long, excellent article how to model a high-quality seagull for use in a cinema film in softimage, including of course exact texture coordinates. The author uses a polygon mesh.
  • That movie was sooo cool. I need to buy it on a high resolution video disk, maybe even a DVD if I have to.
  • Very sharp.

    Derivative works is an ambiguous definition and is open for abuse. GPL leverages copyright law to keep people honest.
  • Of course meshes aren't ideal for manufacturing complex curved surfaces.

    Polygons are just that - polygons. Theyre a bad approximation of any object, just like pixels won't support arbitary scaling up of an object with no quality loss. They are simply the easiest form of geometry to render.

    Perhaps you should take a look at products like GeoMagic Wrap, which can semi-automatically create spline models from point-clouds of triangulated meshes. You will, of course, pay big bucks for this (though if youve already shelled out for MD and 3DS Max, you can obviously afford it.

    A set of discrete points, joined up with straight lines, unless the points are closer together than the size of the smallest possible feature on your model, (probably in the order of a few molecules wide) or the smallest detail your cutting head can make, will never give you an accurate 'copy' of a physical object. The amount of data generated is formidable anyway.

    Even in Hollywood, you generally can't just scan in a model, throw some bones in it and call it jar jar binks. Polygons don't make for good animation fodder, mainly because of the sheer amount of data to manipulate.

    I have seen some interesting papers on creating subdivision surfaces to fit point-cloud or dense-mesh data, but the results weren't all that impressive to me.

    You would be best to use your mesh scans for reference, and create models probably using NURBS for manufacturability.

  • You are right it is not new. I saw a demo about 6 years ago by the CRC (Canadian Research Council) where they had developed a 3-D laser scanner using a white light laser they could produce a full color single pass scan a VERY high resolution. They showed a scan of someone's thumbprint and it was like more detailed and more realistic than a photo - it blew me away back then.
  • Congratulations, you've just learned the first real lesson CAD engineers have to face all the time: The *way* a shape is expressed determines what it's useful for.

    This is really nothing new, in fact a lot of the old B-Rep (Boundary Representation, defining the containing surface of a shape) vs. CSG (Constructive Solid Geometry, defining how a shape is built up from geometric primitives) battle centered around the fact that the models have different strengths.

    In fact, this argument predates 3-D modelling - if you've ever dealt with early "stupid" CAD translators (AutoCAD's being a prime example), you know that there's a *huge* difference between defining a 3-D surface with IGES entity type 102 (IIRC) - the "copious data" type, defining a gazillion points along a path, and a conic equation for that path. The latter is more accurate, infinitely scalable, and takes up a lot less space, unfortunately, the former is easier. (Try building an NC program to cut a surface from a gazillion points and you'll soon realize two things: 1) It's a great way to take a very expensive high speed machine tool and make it *very* slow, and 2) you will have to jump through all kinds of izarre hoops to handle streaming the data, since the poor controller does NOT have enought memory for this kind of abuse.

    Of course data formats matter here.
  • www.3dsystems.com [3dsystems.com]

    They make solid-object printers... they have a couple of competitors, but I can't remember their names off-hand.

Slashdot Top Deals

I've noticed several design suggestions in your code.

Close