Results 1 to 9 of 9

Thread: A 3D model Library

  1. #1
    Join Date
    Aug 2010
    Posts
    58

    Lightbulb A 3D model Library

    I was figuring out if there is any sort of 3D model library available on the web, and I haven't found something really exciting.
    So I posted an idea on Ubuntu Brainstorm suggesting this.
    http://brainstorm.ubuntu.com/idea/26623/
    I wanna know what do you think about it.
    Since one of the most expensive parts of a game development is developing it's graphical environment, I think that it should be a good idea to reduce the burden of the development.

  2. #2
    Join Date
    Nov 2009
    Posts
    91

    Default

    OpenGameArt.org already exists and seems to be what you're suggesting: http://opengameart.org/

  3. #3
    Join Date
    Nov 2007
    Posts
    1,024

    Default

    The problem with such approaches is that having a shitload of models doesn't actually help much.

    There's a number of technical issues that makes using random models difficult:

    (1) Model format
    (2) Material format
    (3) Effects engine requirements
    (4) Rigging specifics
    (5) Animation requirements
    (6) Skinning requirements
    (7) Interaction specifics

    Basically, say you upload a model using skeletal animation dependent on dual-quaternion interpolation for skinning using Cg for effects and material properties with attachment points labeled "left_hand", "head", etc.

    Somebody else uploads a model using key/frame animation using simple skinning, HLSL and a custom framework's support package for effects, and attachment points like "HAND_LEFT_1" and "UPPER_HEAD."

    Even if you convert to the same core format, and rewrite all the effects/materials to use GLSL and COLLADA, you still end up with two models that require multiple implementations of the same feature in your engine to use them. Or you need a highly skilled modeler to convert the models, which is an awful lot of work.

    Second, there's the stylistic points. Look at any two games and you'll see huge stylistic differences in character design, animation choices, colors, textures, etc.

    Making a simple change like switching between clean clothes and disheveled clothes makes a single character look very different thematically, and is really really important to any good, cohesive, and engaging game design.

    If you want art, you basically need to have a single set of technical choices made that boil down to picking a single game engine and saying "all models must load, render, and animate perfectly in this engine." You then need an art director to lay down a large number of stylistic choices and guide and direct artists to follow those guidelines, starting at the concept art stage and basically overseeing everything through until the final model is complete.

    It's certainly possible to do this for a free repository, but it's going to raise the bar for entry, resulting in less overall art. The alternative is having a ton of models, only a small percentage of which are actually usable, and most of those then looking very odd when put together into the same game.

    The best you can really achieve with such a library is some easy to acquire place-holder art that lets devs get an engine built and tested early on before seeking artists. Which is pretty important, because if you try to get artists to make art before you have a workign engine, then you can't guide the artists in terms of those technical decisions I listed above. Plus, volunteer artists need motivation five times as much as paid artists need, and that pretty much means they need to be able to load their art up into the game and see it and play it and enjoy it and be able to go "wow cool look how much better I made this look and feel!"

    If you do push such a library, you really really need to organize it appropriately, and get some artsy people to make some documentation. Pick a few sample engines with well-documented formats, and label any art uploaded with the engine it was intended to work with. Also, make some documented "art themes" that include character design guidelines such as color guidelines, texturing guidelines, body shape guidelines, personality guidelines, and other general thematic elements. For example, the site may have a theme called "Toon Land" described as "a child-friendly, colorful, playful world filled with friendly people, animals, and objects." Items uploaded with that tag would be expected to follow the details guidelines, be cel-shaded, and otherwise fairly Disney-ish. Another theme might be "World of Tomorrow" which is described as "a high-tech but familiar environment in the near future, filled with a light-hearted sense of adventure." You would expect models in that tag to be realistic but never raunchy or ghastly. Yet another theme might be "Camelot" with a bright and colorful but realistic Arthurian aesthetic. I'm sure themes for post-apocalyptic future, space opera, WWII, and horror games would also be popular.

    There's also still going to be a lot of technical hurdles to get well documented, like the attachment points (the information in the models that makes it possible to put items in characters' hands and such) as well as the animations (which are easily the hardest part of modeling, takes by far the most effort to both develop and test, and requires a working engine to test with; a model without animations is barely better than not having a model at all).

    The point is, the library MUST be organized in such a way to let a game developer find relevant art which fits together, and the documentation and guidelines MUST be present to help skilled artists create works that do mesh together.

    Otherwise, you have a very expensive clipart library, and I can guarantee you've never seen an actually good-looking and highly professional poster, flier, or presentation that had a smattering of clipart from any library.

    If you look at something like the community art for games like NWN or Fallout, they have a huge advantage over a general library. Namely, there are both a very specific engine to develop for and test against as well as a huge library that provides a "de facto" style and theme standard that an artist can easily "test" his art against for cohesion.

    Starting from scratch is a LOT harder. No major commercial game I'm aware of has ever done it without revising the art very heavily halfway through the process, as a lot of tweaks end up being necessary to all the art even when all the artists start off working on the same project.

    I don't mean any of this to discourage such a project. Quite the opposite, I'd freaking LOVE a high quality free game art library for my hobby and personal projects, and it would improve the work of a lot of student and indie game developers quite a bit. I'm just trying to point out that such an endeavor is going to be both difficult and require a freaking hell of a lot more work out of the organizer than just slapping together a website with an upload form and a search page in order to be even remotely useful to a real developer.

  4. #4
    Join Date
    Feb 2008
    Location
    Linuxland
    Posts
    4,724

    Default

    elanthis, you really should post these essays somewhere else as well. Even that post would be good material for the wikis of gpdev or freegamedev.

  5. #5
    Join Date
    Aug 2010
    Posts
    58

    Default

    As elanthis said, there will be problems with style and standards.
    However these problems can be solved by sorting the models in categories and by creating guidelines like tango icons has done.
    http://tango.freedesktop.org/Tango_I...eme_Guidelines
    This helps people direct their creation.

  6. #6
    Join Date
    Nov 2007
    Posts
    1,024

    Default

    FYI, I actually found a game model library full of free stuff last week while researching some foliage rendering techniques. ... I forgot the URL and can't find it on Google right now, but it's out there, though. It wasn't particularly high quality stuff, but a decent percentage of it was good enough for basic hobbyist needs.

    Quote Originally Posted by curaga View Post
    elanthis, you really should post these essays somewhere else as well. Even that post would be good material for the wikis of gpdev or freegamedev.
    I've actually been working on an article for the last 6 months for hobbyist/indie game developers, the challenges they face, the limitations they need to accept, and the advantages they have. With input from a few very knowledgeable and very experienced industry veterans who work with a lot of students and hobbyists. The art issue is one of the biggest sections, because art is obviously a very large portion of a modern game and unfortunately artists don't see open source the same way programmers usually do. It goes over most of what I mentioned above, and then a good deal more.

    It'll probably be a good while before it's ready though. I want it to be of the highest quality, and I still need to do some "interviews" with some good game art producers (thankfully I know more than a few) and some successful indie devs who've worked on story-oriented games (the hardest type of game to do as a hobbyist).

    Quote Originally Posted by fernandoc1
    However these problems can be solved by sorting the models in categories and by creating guidelines like tango icons has done.
    http://tango.freedesktop.org/Tango_I...eme_Guidelines
    This helps people direct their creation.
    It's a LOT harder than what Tango did. Realistically, you're going to get a lot more mileage out of examples and a dedicated art director. In the real world, you don't hire an artist, hand him a book of rules, and then wait for him to pump out AAA quality content. Likewise, for programmers you don't just hire someone, give them a dev manual, and wait for them to write Unreal 4.

    Rock solid high quality style documentation is totally necessary. It's still only 5% of what you need for such a project to be truly useful and successful, though.

  7. #7
    Join Date
    Feb 2008
    Location
    Linuxland
    Posts
    4,724

    Default

    Hopefully it will be freely readable online? Phoronix guest article?

  8. #8
    Join Date
    Aug 2010
    Posts
    58

    Default

    Elanthis,
    I know that people not always comply with rules of creating content. But as far as I know, when you are working on a great project and that are a lots of fans, they will comply with all your rules. Be sure of it.
    I'm thinking in a game project using WebGL and I'm sure that as soon as I release something functional, many people will come to make their contributions, because it will be really cool.

  9. #9
    Join Date
    Nov 2007
    Posts
    1,024

    Default

    Quote Originally Posted by curaga View Post
    Hopefully it will be freely readable online? Phoronix guest article?
    Yeah, it'll probably just be on my blog, though I did hint preivously I might give it to LWN (in which case it would be free one week after publishing). It's going to be a while.

    I know that people not always comply with rules of creating content. But as far as I know, when you are working on a great project and that are a lots of fans, they will comply with all your rules. Be sure of it.
    I think you misundertand. It's not that I think people will just ignore the rules. The problem is that you can't document artistic style and theme in particularly rigid way. You need two-way communication to guide artists (and developers, really), community project or not.

    Being a community project doesn't mean there's even a single drop more enthusiasm or love for the project than a proprietary project, at least in the game world. That's what makes this industry so great: it's nothing but people who truly love what they do. It has to be, for the 60+ hours/week you're expected to put in. The quality of life for game developers is relatively low for the pay grade, and it's hard work on top of that. But you get to make games. And it is just ****ing awesome.

    That said, people do break the guidelines and rules all the time in projects like these, often just because they don't realize the rules really are there for a reason. Hell, Miguel de Icaza just published an article last week about how badly most people contribute to open source projects.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •