Stories
Slash Boxes
Comments

News for nerds, stuff that matters

Slashdot Log In

Log In

Create Account  |  Retrieve Password

Computer Scientists Grow a Better Virtual Tree

Posted by Soulskill on Fri Jan 11, 2008 06:02 AM
from the branching-out-from-virtual-design dept.
Reservoir Hill points us to a story about a group of computer scientists who are taking steps to bring the creation of 3-D worlds to the casual user. As a proof-of-concept, Vladlen Koltun and the Stanford Virtual Worlds Group, using data collected by botanists, have developed software to create virtual 3-D trees with roughly 100 different tree attributes, all of which are highly variable. Quoting: "The inability of casual computer users to build 3-D objects - you practically have to be a sculptor, Koltun says - is an anchor holding back the promise of virtual worlds. Koltun's software, Dryad (a tree nymph in Greek mythology,) lets users move through the 100-attribute tree space in a fashion similar to navigating city streets on Google Maps. As in real life, not all trees are equally desirable. Since no single user is capable of mapping out the best parts of the enormous tree space, this mapping of desirability is done collaboratively, leading to continuous refinement of the software."
+ -
story

Related Stories

This discussion has been archived. No new comments can be posted.
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
 Full
 Abbreviated
 Hidden
More
Loading... please wait.
  • Dryad (Score:5, Funny)

    by Anonymous Coward on Friday January 11 2008, @06:14AM (#21997206)
    Is it me or does building a better tree *nymph* seem like a more worthwhile project?
  • Environment (Score:2, Funny)

    by Anonymous Coward
    The Virtual Greenpeace will be pleased
  • by $RANDOMLUSER (804576) on Friday January 11 2008, @06:34AM (#21997278)
    Windows and Mac only. From the FAQ:

    When will a Linux version be available?
    As soon as we finish compiling it. Given how many things we have to statically link, it may be a while.
    (emphasis mine). And then this gem

    On Windows and Linux, trees.dat can be found in the same directory as the Dryad executable.
    leads me to believe they're not real clear on how Linux works.
    Oh well - looks like fun, though.
    • As soon as we finish compiling it
      This [xkcd.com] is all that needs to be said.
      • he he. Sometimes I wish I was a programmer... (only sometimes)
          • One of the nice things about working in the database and getting those crap "data conversion" tasks is that most of the time they run for hours.......I'm "testing" my conversion, time to read the Internet.....I'll check it tomorrow, fix the bugs I notice and repeat the cycle. :D

            Layne
    • Re: (Score:3, Informative)

      It doesn't run on Windows either. Tried two different machines, and on both the program kept crashing before I would even see a tree.
    • And they are apparently not clear on how Vista works either.
    • Sounds like they're not real clear on how windows works anymore either. They don't have "application data" (uh.. .dirs?) directories in every user's "documents and settings" (home, really) just for giggles.

      Now they DO handle "documents and settings" stupidly (in XP. I'll try vista when I have a computer that can run it). What do you mean it's not advisable to mount a separate /home drive? Brilliant idea putting the mount-point information in the documents and settings tree. That way, segregating user d
    • Hmmmm, IDVinc has been doing trees for years. They are used in many games. Also do other simulations. Might be useful for researchers to research the market before setting out to reinvent the wheel.

      Tim
      • Not to be a troll, but obviosuly you're using the wrong flavor of linux. Dependency hell was solved a long time ago. Repeat after me "apt-get install".
  • It uses a tree search algorithm to find it's data files?
  • I'm underwhelmed. (Score:3, Interesting)

    by jericho4.0 (565125) on Friday January 11 2008, @06:46AM (#21997316)
    Generating visually complex plant like shapes has been a mainstay of 3D modeling software and demos for a while now. These guys might have worked with botanists and expressed hundreds of attributes, but I don't see how this will translate to better tools for 3D modeling in general.
    • I think that the idea is to not have a standard library of ready made tree blocks to use over and over. What I do believe that they're getting at is "growing" trees based on the attributes that are put in. The more attributes that are in the system the more variety of trees can be randomly generated. They are also collecting data on the aesthetic appeal of the different combinations of attributes so that they can maybe make a system that will group pleasing ones together to further refine the process. That way you get trees in games that are random, "grown", and no weird combinations of say deciduous leaves on conifer type trees. It's still pretty new software and does't look all that great, but like any project give it some time and I think you may be surprised by the results.
    • The point of this is you no longer have to be a 3D wiz or artist to make decent 3D objects - you just plug in their attributes and they generate themselves.

      So say you are a great programmer but a totally lousy artist - now you can actually make that cool 3D game you have been envisioning by yourself.
    • It might help with realism. Sure, games have good-looking trees already (Oblivion comes to mind). But like any discipline, if you get expert help, it does increase realism.

      For instance, trees that look plausible to not only gamers but botanists, given their environment, will almost certainly seem more "real" even to non-experts. The human mind is good at picking up on patterns, even subconsciously -- and weeping-willow trees in arid areas (for example) will probably look somehow out-of-place to many peop
      • Because I see this as not an end result, but as a beginning. I've often thought that the entirety of all human faces could be described by a similar, discrete set of variables. Know the variables, know the values for a person's face, and voila, you can recreate a particular person's face from a simple database.

        This is already 'ancient' tech :)

        Have a look at http://www.facegen.com/ [facegen.com]

        And therea re a dozen others too,

        LetterRip

  • Already existed. (Score:5, Informative)

    by Aladrin (926209) on Friday January 11 2008, @06:51AM (#21997330)
    http://ngplant.sourceforge.net/ [sourceforge.net]

    Plant generators have existed for a while. There was a proprietary one (that I forget the name of now) that was very good, and there's the above open source one as well. (Which I haven't actually used.)

    As for making it easier for users to create virtual worlds... This is just one small aspect of a world, and doesn't even fully support that, from what I can see.

    "Dryad trees are truly 3-D; they can be spun around or viewed from any angle. They also can be downloaded in the OBJ format and loaded into any major modeling program."

    So it only creates a static OBJ. There's no animation, no information on how it flexes... You can't make this tree sway in the wind without the same tedious work that's always been necessary.

    Saying this helps create virtual worlds is like a crayon manufacturer saying it helps create art... Sure, as long as you only want non-professional art. (And yes, just like crayons, you -can- make professional art with this if you have a ton of talent and are willing to put in the time.)
    • Re:Already existed. (Score:5, Informative)

      by szap (201293) on Friday January 11 2008, @07:05AM (#21997402)

      There was a proprietary one (that I forget the name of now) that was very good
      Speed Tree? [speedtree.com] It can generates trees real time [speedtree.com] based on predefined definition. With proper 3D lighting, it looks way better than Dryad. Mature product as well, being used in lots of games.
      • It generates real-time trees? You mean you get to plant the seed and watch it grow, day by day, week by week? Sure it'd be slow and take a while to complete, but can you imagine realism factor in games!
        • Actually that's EXACTLY what I would want in an ideal persistent world. I would like the leaves to slowly change color during fall too based on the weather cycles in game (it takes warm days and cool nights to produce the best colors). Only when we that level of detail will PW's begin to feel real. It's always bothered me that I complete a quest as part of a big storyline and nothing changes in the world.
      • Re: (Score:3, Informative)

        Or the one that's been around for ages; xfrog [xfrog.org]
      • Actually, no... Though I knew about SpeedTree, I was talking about a different one.

        A Google search finds a ton of them I didn't know about, but not the one I'm thinking of.
  • by maroberts (15852) on Friday January 11 2008, @06:51AM (#21997340) Homepage Journal
    ..till computer scientists forget trees and grow better Bush.
  • It's not quite Speedtree yet, but given time it could be. I can't help but imagine it's going to have a less expensive license. heh.
  • Similar stuff (Score:3, Informative)

    by Anonymous Coward on Friday January 11 2008, @06:59AM (#21997372)
  • Blender Treemaker (Score:5, Informative)

    by LetterRip (30937) on Friday January 11 2008, @07:03AM (#21997394)
    As noted aobove, tools for creating trees are pretty common, ie Blender now has a fairly powerful treemaker that is being used for the Peach Open Movie,

    http://peach.blender.org/index.php/trees/ [blender.org]

    The author does mention that Dryad is 'easy to use', but there are a fair number of easy to use tree making tools already so not sure how 'revolutionary' it is.

    LetterRip

  • it might be a better tree from a botanical point of view sure,

    but are any users from these games going to look at them and notice? hell, I bet they don't even look at the trees in the first place. my ten cents tells me that this software is only going to be valuable if you can store a complete tree in those few hundred botanical variables and then recreate it on demand in SW, this compressing the amount of data needed for the game storage.

    bottom line: its not an improvement if nobody except a botanis
    • Re: (Score:3, Interesting)

      I for one noticed the excellent tree models in Feralas in WoW, that area is just downright gorgeous. If they were more realistic they would probably be even more aesthetically pleasing. The human brain is pretty good at picking up on unnatural things so making trees as realistic as possible will make the game look better even to those who never look for the trees.
  • by PinkyDead (862370) on Friday January 11 2008, @07:28AM (#21997514) Journal
    Slartibartfast - The virtual fjord designer.
  • by rodney dill (631059) on Friday January 11 2008, @07:35AM (#21997548) Journal
    a poem as lovely as a parameterized tree....
  • Sketchup (Score:3, Informative)

    by rastoboy29 (807168) * on Friday January 11 2008, @07:41AM (#21997564) Homepage
    Sketchup (now Google Sketchup) is actually a very intuitive 3d modeling tool--probably the most intuitive by far, IMHO, but it actually suffers from a lack of depth.  I'm waiting for Google to work their magic on it...still waiting, actually.
  • All the trees in the gallery look strangely similar. Are they all conifers so far? It's still downloading the trees from their server (and no doubt slashdotted), but I was hoping to use it to build deciduous-type trees (virtual bonsai). Oh, well. It's still cool.

    Thanks, guys. As Ms. Fawcett would say, "I'm into trees." (Just not quite in the same way she is...)
  • that this website gets promoted heavily and posted to Slashdot just as the submission deadline for the annual SIGGRAPH conference is approaching.

    No WAY this could have anything to do with promoting the visibility of a project that might be described further in a paper that gets submitted there.

    Just my 0.2E-32

    A.
  • The inability of casual computer users to build 3-D objects - you practically have to be a sculptor, Koltun says - is an anchor holding back the promise of virtual worlds.
    I can't help but notice that the real world seems to have survived despite the average Joe's inability to quickly create custom trees. Virtual worlds are different how?
    • In the real world, trees know how to create themselves.

      Whether the lack of realistic trees in virtual worlds is a particularly bad "anchor" is probably more debatable
      • Re: (Score:3, Insightful)

        In the real world, trees know how to create themselves.

        I wasn't saying that virtual worlds don't need trees. I was saying that the creation of trees isn't democratic and easily accessible to all in the real world, so why is it imperative that it be so in the virtual world? Is it a reqiuirement of virtual worlds that they magically bring everyone to the same level of artistic talent?

        Frankly, who wants to live in a world generated entirely by a bunch of users dragging a bunch of sliders as far left as they'll go?

    • What do you mean "joe" can't create custom trees?
    • Presumably the point is that any random person could learn to sketch something, but they can't make a three-dimensional model.
    • You mean all that sex I have been having with trees was for naught?
  • Vue 6 does an amazing job of creating random trees. The only downside is they aren't really suitable for gaming. You can export most of them and I use them in other 3D apps but they tend to be pretty poly intensive and even slow down the renders in 3D apps. You've got some resolution control but even at the lowest levels they are a bit heavy for creating a forest for games. Still an excellent app for plants. I use it all the time for plates and backdrops. It works well for skyboxes and even a forested, or m
  • By Aristid Lindenmayer and Przemyslaw Prusinkiewicz. Absolutley stunning book. Several (many) of the renderings (the palms in particular) are verging on realistic. It's out of print now and you definitely can't have my copy. I won't give it up!

    It turns out it's available here http://algorithmicbotany.org/papers/#abop [algorithmicbotany.org] on the interweb for free.

    Sadly Lindenmayer died the year before the book was published and the book itself is dedicated to him. It's one of those rare science books that makes a good coffee table book too.

  • by Tronster (25566) on Friday January 11 2008, @09:21AM (#21998456) Homepage
    Has anyone here downloaded Dryad (the software) and got it to work?

    I tried it a few weeks ago when I first heard of (believe it was 1.0 not the January 1.1 release). On two beefy computers the program would start then proceeded to crash after 20 seconds of interacting with it. I never saw a tree. I wasn't worried about specs as the one computer has 3 gigs of RAM and an 8800.

    I'm all for releasing public alphas or betas, but was surprised at how brittle it seemed considering the lack of warning or documentation.
  • From the title, I was expecting some kind of either general, or special case improvement on search tree [wikipedia.org] "geometry." Which I think we can agree would have been far more exciting than the, still interesting, CG improvement.
    • Re: (Score:3, Insightful)

      "If you want an infinite number of possible trees then create an implementation for a universal Turing machine."

      Such an implementation has already been created, we computer scientists call it "the computer". Now, no matter how powerful we make our "computer", no matter how (or if) we implement floating point it makes no difference to the number of possible trees a computer can generate.

      Not sure if the post was a troll or an attempt at humour - but the insighful mod makes me sad.