Slashdot Mirror

pushing polygons is wrong. It's a short term solution to a long term problem. Keeping to a poly budget is grunt work that should be done by the tool not by the artist. Programmers who think bitrot is bad should have a look at the "asset rot" of 3d models. The deprecation of graphics assets is so fast that it's a stretch to use the word "asset" to refer to them at all. A character model from a AAA title from six months ago has some value, to B quality games, but mostly none of them are reused.

For 3d models to earn their name as "assets" they need to be created with infinite resolution. This is not hard. Constructive Solid Geometry is a well understood technique for modeling and is typically used in CAD applications. An object described even at what would be considered a course level of detail of typical CSG modeling is orders of magnitude higher resolution than the typical game model. What's more, they can be incrementally improved, whereas the b-rep that is typical for a game model today is a one shot affair.

Today, the vast majority of CSG models are created for raytracing architectures which, although they give stunning results, are too slow for realtime applications such as games. Even the attempts to create realtime raytracing systems are aimed at rendering b-reps because of the opportunity to perform an acceleration stage which greatly reduces the scene complexity. B-reps are a win for realtime applications, but just because your application requires a b-rep doesn't mean artists need to get their hands dirty pushing polygons.

Two solutions exist which can render CSG models in realtime.

The image-based rendering algorithms with implementations (such as OpenCSG) which take advantage of z-buffer and stencil-buffer hardware in popular GPU cards. The image-based systems are more compatible than raytracing with the current 3d graphics rendering state of the art, but still require intricacies to integrate that have prevented them from appearing on the market.

The other option is the automatic generation of a b-rep from a CSG model. This has the advantage that it requires no change to the 3d graphics rendering in games and the algorithm can be parametrically tuned to produce b-reps that are fast for different applications. For example, the same CSG model can be used to generate b-reps of different Level Of Detail for when the object is close up vs far away, or for pre-rendered applications such as cut scenes or trailers.

It just happens to be really hard.

Maybe I'm just not picky, but I haven't really been terribly upset with the Cube's graphics.

Nor am I; although it could stand higher-resolution output. Otherwise, it's pretty much fine.

The Xbox 360 I would say, from what I've seen, is barely hitting that metric, except of course it can do HD, which really only matters for 5-10% of gamers anyway.

Pretty much. What I've seen has been pretty much crap compared to the current generation. Slight improvements at best; no high-poly models, particle effects, etc.

Sure it's early, and I'm sure by the time we're almost done with the 360 it'll far surpassed anything I can differentiate (would anyone have possibly imagined a game that looks as good as Shadow of the Collosus could have come out of the PS2 when it launched in 2000?)

People still complain that the FF8 demo was faked, yet we have games like Jak 3, God of War, Shadow of the Colossus, and others that far surpass it. Even the Cube has had various graphically awesome works (despite the involvement of some bafmodads). Unfortunately, the XBOX never really got that much spectacularly better than its first generation; maybe it never had the time to mature, or maybe there was just nothing else there. In the same way, maybe the 360 will... but maybe it won't, too. Especially if it'll only last another 4 years.

but we're hitting the photorealistic barrier pretty hard as it is.

OK, this I have to quibble with. Even various PS3 trailers, regardless of your opinion on their veracity---while amazingly better than what we're used to---are still not photorealistic. I point you at, for instance, the Yafray gallery, which has some fairly amazing pieces. I don't believe the PS3 even claims realtime radiosity and raytracing, which we'll want to see before we truly claim photorealism. And there's a ways to go beyond that.

(This is one reason the Cell, despite not being the PS3's main GPU, is exciting; radiosity and raytracing can be done in amazing parallel. The more CPUs you throw at it, the faster you can do it, down to the pixel. Wouldn't it be fun to build such an engine on a box with 4-8 cells? All those independent SPEs... tasty.)

However, hype is everything, and having an "underpowered" console isn't going to help Nintendo regain their hardcore group. It seems like they've given up on them anyway. Which is fine, they aren't terribly good gamers anyway.

I think we have different definitions of "hardcore" here. "Hardcore" gamers are not the sort whose first game was Halo in the frathouse. Hardcore gamers are the sort who still get out their NES, and have the imagination to play tabletop games. These are the sort Nintendo is still catering to.

Of course, if all they make is party games, and don't start putting out a serious library, they'll never get anywhere. This is Nintendo's main problem right now. Not how many pixels they can push; how many games they can make.

You're right...there is no comparison. I haven't tried DAZ|Studio myself because of the fsck'd up download process, but judging from a lot of posts here it's something akin to Poser and the Sims ;-) ... because you are supposed to buy expansion packs contataining material that's impossible to make yourself in this "3D package".

Blender is the total 3D package. It's got nearly all state-of-the art modeling and (paired with Yafray rendering features, it's not the baddest 3D software out there but it's free as in (beer|speech) and with DAZ|Studio there is no comparison. Blender is very useful and practical, but like Photoshop it takes some skills and understanding of principles to create decent output. I worked my way up from the likes of Bryce and Infini-D and after getting proficient at Cinema4D, Blender wasn't that much of a challenge. But I know it is complicated and totally non-standard. Unfortunately that is partially true for any pro 3D package.

DAZ seems to be a toy, much like Bryce. Toys are nice, easy and always fun. Probably a necessary introduction to grasp basic 3D. I welcome this addition to free 3D products but it really does not alter in any way Blender's position as top free 3D package.

Perhaps if they opened their code they could begin to compete with the free and open-source offerings.

Compare DAZ to Blender/Yafray

Blender was free, closed-source software for some years. No doubt DAZ will also make the decision to emancipate themselves in order to grow in time with their users.

where's the Maya/3DS/LW/Softimage alternative? It doesn't exist (dont be a bone head and suggest Blender here, its like comaring a 79' VW to a Ferrai).

Maya is ported. So is a little thing called Houdini. But you probably never heard of Charlie and the Chocolate Factory, Final Fantasy X, or X-Men...

Also, to call Blender a '79 VW compared to any of the above is ignorant. No, it's not Houdini, but it will easily take on any of the modellers you mentioned, and there are some pretty nifty renderers available, too.

What Linux really lacks is the equivalent of Digital Performer. And no, rosegarden or any of the others don't even come close. :-(

Screw all that crappy, hard to use stuff.

Wings 3D and Yafray are what you want.

why? what does it offer that's better than the excellent http://yafray.org/?

Blender, as far as I know, is just a design tool, it doesn't actually do any rendering, it just allows you to develop models in a range of formats that you can then throw at a renderer.

No, it has a scanline renderer built in and an optional ray tracer backend called yafray.

I think the most reasonable solution for global illumination is photon mapping. The algorithms are elegant, produce very good output, and (unlike radiosity) are not horribly resource intensive.

Studios are not going to just open source and give out all their code. Even if they did, they'd be huge and confusing to the open source world and nobody would know how to use it. We'd have another Netscape/Gecko/Mozilla thing.

If you want to have open source 3d tools (which there are already), you've got to work from the other end. Creating your own. Taking on the studios at their own game. Growing up between their toes.

If you're a graphics nerd, don't sit around pining like this, start using/hacking on blender and yafray. They are already seriously good and getting better by the day. If they don't meet your requirements yet, start using them and they soon will with all the extra attention. Besides, half the "really cool" stuff done/needed by 'professional' 3d artists are implemented in custom scripted things. Blender's fully python scriptable. Has been for a long time.

Gee. Things screwd up there:

povray export script

yafray

I could not agree more.

Applications like Yafray completely blow away POV-ray. They are faster, and more importantly they look better. POV-ray renderings still look like something from the 90's. Dispite updates, it still looks the same as it did way back when. Back then it was decent, but by todays standards POV-ray looks crappy.

POV-ray is like OpenBSD. That is, yeah it's kinda cool, has some solid code, is based on some "perfect" goal, but development is extremely slow-paced, it runs slow, and uses way outdated concepts. They will never catch up. Like socialism, the idea of a "perfect" system just doesn't work in practical, real-life, terms.

Oren-Nayar, Blinn, Phong, and Toon shaders were added to Blender in version 2.28. There was a massive UI overhaul for version 2.30 (read the release notes at blender3d.org) and work is still continuing on this front. There is also work going on to integrate yafray (a global illumination renderer, under fierce and rapid development) seamlessly into Blender, too.

Cheers

Ton's had the raytracer written for some time now, but it never got incorporated into Blender. The preview is the first to incorporate the code.

You could already do shadows and reflections in Blender, but they were simulated with shadowmaps and reflection maps, the same way that Pixar's Renderman renderer had done it.

• Worthless trivia: Renderman only recently acquired raytracing - for the few times that it was actually used, they used the BMRT (Blue Moon Rendering Toolkit), a raytracer developed by Larry Gritz. Larry quit Pixar and formed ExLuna, which marketed another Renderman compliant renderer. Pixar sued ExLuna for IP infringement (the exact details are hazy, since they came to an out of court agreement), ExLuna was bought out and all the renderers (including BMRT) disappeared. Soon thereafter, Pixar's Renderman added raytracing support. Still, full raytracing is used in Renderman quite sparingly.

The Yafray (Yet Another Free Raytracer) is a stand-alone full raytracer with a lot of features that has nice integration (thanks to Python scripting) in Blender. Future versions of Blender promise to integrate it more tightly, and seems more likely that's where a 'full raytrace' option for Blender will come from.

If you haven't seen it yet, you should check out yafray. Truly a 2nd generation free 3d renderer.

RenderMan didn't have a raytracer until fairly recently. For movies such as A Bug's Life where a raytracer really was needed, Pixar ended up using Larry Gritz' BMRT (Blue Moon Rendering Tools), a RenderMan compliant raytracer.

Ironically enough, Pixar ended up suing ExLuna for infringment of intellectual property, which ended up in ExLuna's products (including BMRT) being pulled from the market.

So how did Pixar get by all those years without a raytracer? Well, there are a lot of shader tricks that can be done, as well as clever use of lighting. Radiosity, reflections... all these effects can be simulated.

All this is to say that the real limit of tools isn't so much the tools themselves, but what people do with the tools.