Rage

[Jayel]

I've always hated stencil shadows... looks too sharp and flat, and very expensive to make them softer. Plus I suspect today's games shy away from it because its cpu usage increases with geometric complexity (doing edge tracing and all).

CryEngine3 reportedly has real-time global illumination, which means

this quality of lighting in real time.

Whether they'll be able to deliver on that promise remains to be seen.

[toblix]

Obviously it will be an approximation (or rather, more of an approximation). There's no trick that lets you do that stuff in real-time with anywhere near the same level of precision, but, as with all CG, the point isn't that it's 100% realistic, but that it looks 100% realistic. I've seen some pretty good real-time GI, so I have no doubt they've got something like that in their engine.

[Garple]

More like Fallout 3 with cars and circuses.

So, like Twisted Metal with a wasteland?

[DanJW]

Obviously it will be an approximation (or rather, more of an approximation). There's no trick that lets you do that stuff in real-time with anywhere near the same level of precision, but, as with all CG, the point isn't that it's 100% realistic, but that it looks 100% realistic. I've seen some pretty good real-time GI, so I have no doubt they've got something like that in their engine.

Biological computers and quantum computers may be able to handle that kind of thing real time, but it will be a decade at least before gamers get their hands on that stuff.

[Erkki]

Biological computers and quantum computers may be able to handle that kind of thing real time, but it will be a decade at least before gamers get their hands on that stuff.

A decade is very optimistic, I think.

[MrHoatzin]

Think ten years back. We barely even had shaders and materials in realtime then, let alone objects casting shadows on themselves (whatever technical name for that stuff is).

[Erkki]

Think ten years back. We barely even had shaders and materials in realtime then, let alone objects casting shadows on themselves (whatever technical name for that stuff is).

Yeah, but we're talking quantum/bilogical computers. They are in their infancy and there's years and years of research to be done there probably before they become mainstream.

But sure, that kind of lighting might be doable in real time in 10 years.

[elmuerte]

Think ten years back. We barely even had shaders and materials in realtime then, let alone objects casting shadows on themselves (whatever technical name for that stuff is).

1999:

- shaders: well.. not really shaders because they didn't really exist (it's sort of a hardware hack), but a lot of the effects (underwater deformation, reflections, heath deformation, ...) where present

- materials: this was an important feature of the UnrealEngine back then

- self shadowing: yep... took quite a while before real-time self shadowing made it into the engines.

But just looking back only a few years, it was 3 years ago when consumer grade stream processors popped up. Stream processors will have a major influence in realtime 3D rendering (when done right).

[MrHoatzin]

Oh, I didn't register the quantum/biological computers bit. We've been hearing about holographic data storage for about ten years now and nothing is really happening on that front quite yet either, so yeah, ten years is a pretty improbable time frame. Twenty years may be a more realistic mark. For what it's worth, the very first Pixar movie can be run in realtime on fairly modest hardware nowadays.

On the other hand, not much really changed in the basic architecture of computers in the last 20 years (a braver man would argue that nothing really changed in 30 years, but that is not my fight)—quantum/biological computing would be a far more revolutionary a shift than anything that happened in the last 20-30 years...

[DanJW]

A decade is very optimistic, I think.

It is a bit optimistic yeah... because I was being a bit optimistic You sure taught me.

I mean, I did qualify with an "at least". Dunno why I bother TBH grumblegrumblemoan

[Erkki]

It is a bit optimistic yeah... because I was being a bit optimistic You sure taught me.

I mean, I did qualify with an "at least". Dunno why I bother TBH grumblegrumblemoan

Sorry

I may have missed the "at least" part.

[DanJW]

Hehe no worries I was playing up to the inevitable nitpicking.

Anyway; when those technologies do become available, I like to think that id will be there making use of them first.

They will probably call the game Blast or Crash or Grrrr! or something...

[Erkki]

My bet's on Flash.

[DanJW]

I associate that word too much with household cleaning products and Freddie Mercury (not at the same time).

[Jayel]

I have a great respect for John Carmack (and all id does), but his days as a pioneering game programmer seem over. I just get the feeling that he's got nothing to prove anymore, so he's just content sitting wayside watching other people innovate.

BTW, cryengine 3's real-time global illumination technique at:

http://www.incrysis.com/index.php?option=com_content&task=view&id=812&Itemid=1

It feels like a series of crude techniques strung together with a duct tape, and yet the end effect is very convincing. I would've never believed it was attainable even in its current limited form. If my boss told me to create a real-time GI solution yesterday I would've laughed in his face and never even try. I really envy these people's "can do" attitude.

[MrHoatzin]

I have a great respect for John Carmack (and all id does), but his days as a pioneering game programmer seem over. I just get the feeling that he's got nothing to prove anymore, so he's just content sitting wayside watching other people innovate.

Feh, I dunno. On one hand, he did do the fancy infinite texture buffering wizardry in the new id tech, on the other, in the interviews I recall him saying something to the effect of, Well, I was shocked no one had done this thing yet, so I figured I'd take a stab at it.

[SiN]

FULL DISCLOSURE: I'm a complete, and utter id/Carmack fanboy.

He is *SO* still at the top of his game!

id had a bad run with Doom3. They misjudged the market in two very important ways: (1) the engine was PC-exclusive when it should have been multi-platform, (2) it was a corridor engine when the industry was going open world.

However, consider this: the Doom3 engine started development in 2000, and was more-or-less complete by 2002. In 2000, those decisions weren't all that outrageous. By 2002, the writing was on the wall: Carmack *knew* he'd gone wrong, but his hands were tied until Doom3 shipped, which was of course in 2004. We know this by looking at id tech 5 ... id tech 5 is the complete opposite of doom3: open world and multiplatform.

Further, I genuinely think id's engines are just built better. There's an elegance to them. Unreal is just a bunch of graphic & shader tricks thrown at the wall. Crytech is impressive ... if you've got the horsepower. IMO, that is not a great engine.

Carmack has always balanced the right amount of features with the right level of performance. Every one of his engines has a theme, and the features work around said theme ... Q1 was true 3D, Q2 was opengl support, Q3A was speed and curved surfaces, Doom3 was unified lighting, Rage is open world and megatexturing. There's a certain elegance to id tech that I can't help but admire.

Alright, I think that's enough fanboy ranting.

Cheers,

Mo

[Jayel]

Yeah I used to be total id/Carmack fanboy too - still is in some ways.

I think there were more problems with Doom 3 engine than you mentioned (although I don't know about your point 1 - doom 3, prey, quake 4, etc all got ported to xbox, and even ps3 for quake wars):

1. the impact of unified lighting wasn't that impressive, because other games beat it to the punch (DE: Invisible War for example)

2. developers weren't convinced that Doom 3 engine was capable of anything other than shiny-metalic look, because id did not showcase anything but.

3. Quake 4 on xbox360 (a launch title), had crushing framerate issues. I wouldn't be surprised if it scared off a lot of developers browsing for a new engine.

4. Unreal engines were marketed better and... simply better. In addition to having the right buzzword features, they're easy to use and came with very nice development tools that id couldn't match. It was (and still is) a major selling point. I mean look at this list!

Mega textures in tech5 sure seems impressive, but I don't know how it differs from what Carmack already did in Quake Wars.

[SiN]

Yeah I used to be total id/Carmack fanboy too - still is in some ways.

I think all us programmers are, to an extent.

I think there were more problems with Doom 3 engine than you mentioned (although I don't know about your point 1 - doom 3, prey, quake 4, etc all got ported to xbox, and even ps3 for quake wars):

3. Quake 4 on xbox360 (a launch title), had crushing framerate issues. I wouldn't be surprised if it scared off a lot of developers browsing for a new engine.

Yeah, but not really. The ports weren't done in-house and they were generally quite poor. Like you mentioned, framerate issues in Q4. Quake Wars I believe also had issues/compromises. Doom3 for xbox was stripped down significantly. Etc. Contrast this with Rage: 60fps on all platforms.

1. the impact of unified lighting wasn't that impressive, because other games beat it to the punch (DE: Invisible War for example)

Again, this was tied into the elongated dev cycle. Carmack's work was done in 2002, Invisible War came out 2003. If id had the same policy as Epic (the engine is always a WIP, release first test later) things would be different. id's policy is to ship the flagship title before licensing the engine.

(Honestly, I'm not sure id have the right policy any more ... it was admirable when games took max 2 yrs to make, it's a bit of a wasted opportunity these days. Still, there are some very valid points for going by id's policy, mostly having to do with Doing The Right Thing)

2. developers weren't convinced that Doom 3 engine was capable of anything other than shiny-metalic look, because id did not showcase anything but.

That's a very good point. Doom3 more than any other id engine was designed specifically for that *one* look. Definitely hurt them.

4. Unreal engines were marketed better and... simply better. In addition to having the right buzzword features, they're easy to use and came with very nice development tools that id couldn't match. It was (and still is) a major selling point. I mean look at this list!

Another good point ... ish. Obviously, I don't think Unreal engine is better.

Ease of use from an engineering POV, I'm guessing id's stuff is better. I've had the pleasure of briefly working with the Q3A open source code at work ... Carmack creates beautifully structured code. No doubt cleaner/more elegant than Unreal's offerings. But of course, this is speculation.

Dev tools, yeah, totally agree. I didn't mention that, but it's a very good point. But again, check out the previews of the Rage toolkit ... very intuitive stuff, especially the painting/decal stuff.

Mega textures in tech5 sure seems impressive, but I don't know how it differs from what Carmack already did in Quake Wars.

It's Quake Wars turned upto 11. QW only used megatextures for the terrain, and even then it was kinda limited. QW was a proof of concept. This is the real deal.

Cheers,

Mo

[SiN]

Sorry for the double post, but ...

Feh, I dunno. On one hand, he did do the fancy infinite texture buffering wizardry in the new id tech, on the other, in the interviews I recall him saying something to the effect of, Well, I was shocked no one had done this thing yet, so I figured I'd take a stab at it.

And that's exactly what makes Carmack so awesome. The other guys just aren't doing any of this.

The other engine developers are all pretty much doing the same thing: seeing what the video card vendors come out with, and upgrading their base engines accordingly. You can see this with Unreal Engine ... it gets prettier every iteration, but fundamentally, it's the same damn engine!

Carmack is going beyond that. He's thinking outside the specific hardware implementations and getting creative. I know people don't associate engine programming with creativity, but the best developers are *incredibly* creative ... just in a different way. There's real innovation in coming up with a different way to stream textures versus "oh, let's implement normal-bump-displacement mapping" or whatever.

Cheers,

Mo

[elmuerte]

To me it looks like Epic (, CryTech, ...) and id have different goals. Epic wants to build a game engine to license to others, as well as create some games. For id it looks more like they want to create some cool looking tech demos and build games around it. And it that process they are not looking much at what the "competitors" are doing. Id focusses a lot of a few technological features of their engine, and leave the rest where it is.

SiN your elegance argument is moot. Sure, the quake 3 engine has some elegant parts, just like the UnrealEngine had/has. Overall the elegance factor in the UnrealEngine is becoming less, but that's because the engine has been expanding rapidly since 1998, and thus less elegant solutions appear, and more proper (or less proper) solutions take a higher share in the codebase.

But it's quite difficult to compare engines on "elegance" when you don't know much about it's internals. I've seen the guts of all generations of the UnrealEngine, but I haven't seen anything from id since quake 3 was open source'ed, and I haven't seen anything at all from any of the other engines out there.

And that's exactly what makes Carmack so awesome. The other guys just aren't doing any of this.

That's not true, or at least. Tim Sweeney is doing the same stuff. And the other lead engine devs might do to, but I don't follow them much. Just check out the slides from Tim's presentation linked to in this article. These things get way to little attention in the game industry press. If there are no screenshots, movies or ign.com-grade quotes then this info is mostly ignored.

ps, props to the CryTec guys for stabbing at epic with their real time global illumination demo (i.e. using a very similar scenery as Epic did). Also, TryTech's RT-GI demo contained an interesting rendering bug. There was no GI in that red ikea grid cabinet in the NY appartment.

[Garple]

My bet's on Flash.

It will be called "Pants!" and it's coming Q4 of 2010

[toblix]

Here's how you do real-time GI: http://www.crytek.com/fileadmin/user_upload/inside/presentations/2009/Light_Propagation_Volumes.pdf. Apparently you just do a bunch of maths really fast.

[LOPcagney]

I don't see the comparisons between Rage and Fallout 3. The theme is similar but the flavor is entirely different. I see two unrelated games that just happen to have been released in relatively quick succession.

That's maybe why I'm looking forward to it so much. Whereas Fallout's drab uniformity turned me off, the sort of crazy carnival aesthetic of Rage looks much more interesting.

[toblix]

I don't see the comparisons between Rage and Fallout 3.

I guess because they're both great looking post apocalyptic wasteland shooters people will imagine some connection beyond that. I always think of the Post Apocalyptic Wasteland setting as some tired old game cliché, but when I try to think of all the games set in such a world I fail to come up with more than the Fallout games. Oh, and also that game Quarantine.