John Carmack Discuss Mega Texturing

An anonymous readers writes in to say that "id Software has introduced a new technology dubbed Mega Texturing that will allow graphic engines to render large textures and terrains in a more optomized way while also making them look better. Gamer Within has Q & A with John Carmack on Mega Texturing."

mega texture commands in Doom3

It may be insignificant, but I accidently saw two relative commands in Doom3, r_showMegaTexture and r_megaTextureLevel.

Re:mega texture commands in Doom3

Mega and mega are not the same thing

Really? What's the difference?

keep the spelling the same if you mean to refer to the same thing

Camel casing is an exception to this rule. Making it so that the M at the "beginning" of the identifier was caps would break consistency with other identifiers. And unless there really is a distinction between Mega and mega, and both had meanings within the code, I think the latter is more important.

Carmack Owns My Wallet

This is exactly why Carmack owns my wallet and why ID Software does so well. Gamers rejoice!

Article Text

0 replies and the site was already moving really slow. Here's the text in case /. kills it.

Publish Date: 01 May 2006
Cain Dornan

One of the most respected and well-known game developers in the world, John Carmack hardly needs any introduction. Having mastered the skill of game programming, Carmack co-founded developer id Software, and has also worked on such classic series as Doom, Quake and Wolfenstein 3D.

In this Question & Answer with Carmack, he discusses the new MegaTexture technology, which will be used in the upcoming Enemy Territory: Quake Wars for PC. Definitely a worthy read for any programming, designing or general development enthusiast, as well as any gamer slightly interested in the development process behind games.

Q1: What is MegaTexturing technology?

Answer: MegaTexture technology is something that addresses resource limitations in one particular aspect of graphics. The core idea of it is that when you start looking at outdoor rendering and how you want to do terrain and things in general, people almost always wind up with some kind of cross-fade blended approach where you tile your textures over and blend between them and add little bits of detail here and there. A really important thing to realize about just generally tiling textures, that we're so used to accepting it in games, is that when you have one repeated pattern over a bunch of geometry, the texture tiling and repeating is really just a very, very specialized form of data compression where it's allowing you to take a smaller amount of data and have it replicated over multiple surfaces, or multiple parts of the same surface in a game since you generally don't have enough memory to be able to have the exact texture that you'd like everywhere.

The key point of that is what you really want to do is to be able to have as much texture as you want to use where you have something unique everywhere. Now normally, you just can't get away with doing that, because if you allocate a 32,000 by 32,000 texture, the graphics curve can't render directly from that. There's not enough memory in the system to do that, and even when you have normal sized textures, games are always up against the limits of the graphics card memory, and system memory, and eventually you've got hard drive or DVD memory on there, but you wind up with a lot of different swapping schemes, where you'll have a little low-res version of a texture, and then high res versions that you bring in at different times, and a lot of effort goes into trying to manage this one way or the other.

So when Splash Damage was starting on, really early with Enemy Territory: QUAKE Wars, they were looking at some of these different ways to render the outdoor scenes with different blends and things like that. And one of my early suggestions to them was that they consider looking at an approach where you just use one monumentally large texture, and that turned out to be 32,000 by 32,000. And I - rather then doing it by the conventional way that you would approach something like this (i.e. - chopping up the geometry into different pieces and mapping different textures on to there and incrementally swapping them for low res versus high res versions), just let them treat one uniform geometry mesh and have this effectively unbounded texture side on there, and use a more complicated fragment program to go ahead and pick out exactly what should be on there, just as if the graphics hardware and the system really did support such a huge texture.

In the end what this winds up getting us is the ability to create a great outdoor terrain texture that has far more complex interactions than anything that you would get with any kind of conventional rendering, where you've built it up out of pieces of lots of smaller textures on there, where they do some sophisticated things with growing grass up between bump maps. And then you can go back and do hand touch ups in a lot of different places to accent around features that are coming out of the surf

Re:Article Text

So if a mod team wants to make their own map you either need to reuse one of these behemoth textures or find an artist that can wrap their head around the technology and create one themselves.

Nah, you just need good tools. Use the game itself as a tool and let people run around the level spraying the texture with spray paint cans (or the digital equivalent). Then spit the MegaTexture out after they're done.

Genius

John Carmack is a Genius in the gaming industry. Quake 3 was by far the best game of its time. Unreal Tournament was fun, but it just wasn't Quake 3. I hope he continues to be innovative and keep the gaming industry steaming forward, and maybe create a few more games thats never been done before.

Re:Genius

That's right, why would you want to buy UT which had 7 gameplay modes out of the box and bots that weren't either retarded or cheating, not to mention the easiest mod switching system of seen to date in mutators; when I could have the "innovative" perfection of Deathmatch or Team Deathmatch?

The modding community filled the gap eventually, but that's not points for Q3, that's points for all of the dedicated people who were upset by the lack of options in that fanboy love-fest.

Sorry if I sound bitter about it, but I can recount back to the days when PC Gamer stated, more or less in its review that UT was vastly superior to Q3 in every imaginable way, and then gave it a lower score and handed the editors choice to Q3 instead. They were flooded with mail but never really could explian whose bathwater they were drinking when they either wrote a review that was too good or a score that was too low. I suspect a rolled up wad of hundreds under the table and nothing more.

So how long ...

... before Creative Labs asserts a patent over this?

Why not go procedural?

This is a very interesting technique for realistic terrain, but I wonder what advantage this approach would have over procedurally rendered textures? I very much like the idea of being able to (effectively) zoom infinitely into a texture being 'generated' as opposed to 'drawn'... and the strengths of modern consoles play to this procedural generation quite well (PowerPC chips, Cell chips). Maybe thats why Carmack isn't so interested?

Re:Why not go procedural?

John C is highly opposed to procedural textures. His thinking is that it requires that programmers act as artists. MegaTexturing allows artists to use their favorite tools to create their textures without worrying about programming or whether a certain effect is supported by a specific procedural language.

I agree with him.

On Carmack

Carmack is really good as a person who pushes teh technology.

As a game developer, though, it's just not there. anytime I hear about an id game now, I just wait until someone brings out a truly great game using the engine that Carmack has developed.

Seriously - let's review teh last few: Doom3? Enter room. Kill. Lights go out. Kill more. Repeat. Q3A? See also: UT Q2? See Q1. then the origin Doom games. Then Wolfenstein.

id software make great tech demos. Not great games. Beyond the engine, id's games do nothing that hasn't been seen in all the other clones. They get a pass on gameplay though, strictly on name.

Re:On Carmack

One game I'd love to see is Hexen III. For the time, the graphics in Hexen II were incredible and the gameplay was great - there was more focus on solving the puzzles (how the heck do I get out of this level) than there was on killing. Sure, there were plenty of monsters to kill but the, er, mazes seemed more intricate in Hexen. I don't recall whether that was just due to the different visuals or if it actually was the case. I've actually been playing the old Id games again in order (Doom, Doom II, Final Doom, Heretic, Hexen, etc) and find Heretic and Hexen more entertaining than Doom.

Variation on a theme

Uniquely texturing entire terrains sounds pretty cool, but the concept isn't entirely new - just an evolution on an already-existing idea.

I think the Myth RTS games from Bungie used very large textures for the terrain, and this moved on to Halo - terrain there is drawn using a large, low-resolution texture - the red, green and blue channels are used for the colour, while the alpha channel is used to determine which of two detail textures should be used - e.g. grass or sand. It works quite well [halomaps.org]. I think Far Cry does something similar, but more advanced still.

The former are still low-resolution, though - but the not-a-game Celestia [shatters.net] has 'virtual texture' support, for rendering silly levels of detail on planet surfaces. Like, up to 128k by 64k pixels [celestiamotherlode.net]. The textures are split into many, many files for each level of detail, which are streamed in from the hard disk when required. Works fairly well.

Combining the two approaches, though, seems very new - the 'Mega Texturing' from John Carmack is probably dramatically different from an implementation point of view, and sounds rather interesting at any rate - the description of the upgraded, non-Quake-Wars version makes it sound like it could uniquely texture a whole world beyond just terrain, so could work for simulating real cities, as opposed to smoke-and-mirrors game cities.

I'm sure he talked about this in a .plan thing some years ago - anyone want to find it?

But don't forget the game

High quality graphics are great, unless in the process the quality of the game ( story, environment, gameplay, etc) is forgotten. I would rather average graphics and great game-play, over average game-play and great graphics.

too bad

I've already patented placing any texture on any polygon.

patent?

What I want to know is whether or not ID will patent this technology, and of course as to whether it actually is patentable. I know John is a major advocate of abolishment of software patents (he famously drew a parallel between software patents and being mugged), but surely with new technology like this he's put in a position where if he doesn't patent it other people will.

The question is: can using a very large texture with fragment shaders on top be patented? I'm not qualified to answer that but if so surely John & ID are put in a difficult position.

Personally, I think software patents are a terrible thing, which could potentially leave software innovation in the hands of the few largest most litigious software companies in the world which benefits nobody (except their stockholders).

Great idea though John!

Gameplay needs innovating.

This is neat. It's been a shortcoming I've noticed in most games, where landscape textures tend to be lacking.

However, what we really need is gameplay innovation. Actually, what we really need is for developers to stop making every last first person game a damn shooter. Can't they do anything else with a first person perspective. The potential here is enormous and yet it looks like developers have a fetish with gunplay.

There have been games with potentially strong storylines that get mired down by this nonsense. There's little discovery and certainly no problem-solving. These games come down to who has more firepower and occassionally discerning some basic pattern in enemy movement.

Maybe the problem is that these developers invest so much energy in graphics that there's little room to refine the other aspects of the game. Or they just think that the consumer doesn't want to do anything other than destroy things and kill people.

Voxels

What ever happened to voxels? Remember games like Deltaforce, Comanche and Werewolf? That was back in the days when CDs were the big thing and those terrains looked beautiful, and I'm sure with a bit of research were also destructible. Think about it, terrain that will actually explode.

I've actually been thinking on this a lot myself

While I've been extremely time-limited lately, I've always wanted to get back into playing with 3d development and I've been looking a lot at sci-fi space-based games. In terms of visuals, one of the biggest problems I've had with many of the ones currently available, is that space games generally tend to go with the "large and looming" aspect. You have these really big objects, such as starports, battlecruisers, or planets. Making a texture that would nicely cover them would be huge (especially the planets). Therefore, when you get close you start seeing lots of nasty pixellation. The same has held true for many games such as FPS's etc with wall-textures. Nice-at-a-distance, crappy up-close. But it's moreso in the space games.


Of course, having nice smooth textures is nice, but the next step would be better mesh-conversion from simple to complex based on distance. I've heard of this being worked on and/or possible used, but I've not yet seen a game that would use it. How about a space-game where you come up on a planet. You get close, textures go pixelly, and then you "crash." What if you could get in nice and close, and then not only do the textures increase in detail realistically, but you so does the mesh. So at this point, you can actually move right in to the 'ol planet and find a nice dock to land on as the round-ball-from-a-distance becomes more realistic with mountains and various human settlements/bases.

Alternately, my plan would be to use MextTextures until you get close, and then maybe use a cutscene where you break atmosphere before ending up nearer the surface (otherwise mapping the geography of a whole planet would be insanely time-consuming for a large space game).

Giga Texturing

is what I'm waiting for. Mega Texturing is so last magnitude.

don't see

I don't see this as anything revolutionary or even big for that matter. The idea that you use a ground texture has been around for a very very long time... recently there has been many different techinques to handle this such as texture blending and many other ways. The idea that you can use such huge image files over a large area is nothing new... it just has never been done with a single image before which doesn't drop off qulity over distances like most engines do. This, in my opinion is just one of the next step logical steps now that game developers have alot more power and space to work with.

Spelling error or made-up word?

Does "optomization" mean "optimization" or is it some slang combination of "opto-" and "optimization" I'm not familiar with? (Since the article is about improving game visuals, "opto-" would almost make sense...)

Truly, this is an age of wonders.

At last.. now I can look forward to Quake 5 offering a previous unparallled highly detailed panorma of three hundred shades of brown.

editors?

..more optomized
 
Optimized? Do you get paid?

Why people don't take Slashdot seriously

will allow graphic engines to render large textures and terrains in a more optomized way while also making them look better.

Perhaps the submitter and Taco should "optomize" their spell checker software.

ClipMapping (or ClipTexture)

Here is an interesting article I saw in Doom3World.org: http://www.doom3world.org/phpbb2/viewtopic.php?t=1 0673 [doom3world.org] Check the fourth post. Here are some interesting links about ClipMapping/ClipTexture: http://www.cs.virginia.edu/~gfx/Courses/2002/BigDa ta/papers/Texturing/Clipmap.pdf [virginia.edu] http://www.sgi.com/products/software/performer/pre sentations/clipmap_intro.pdf [sgi.com] http://techpubs.sgi.com/library/tpl/cgi-bin/getdoc .cgi?coll=0650&db=bks&srch=&fname=/SGI_Developer/P erf_PG/sgi_html/ch15.html [sgi.com]

Next Gen Consoles

The best info I find about how the next Gen consoles are going to stack up, I find from devs.

Read how the devs that are working on them, and what they are liking, I find provides better incite than PR prelesses, or Pony shows for the press.

Press releases are garbage, reading inbetween the lines of Carmack's or Sweeney's comments provides far more information about where consoles are going.

Most interesting answer:

"Q14: Do you think that the MegaTexture technology will be accessible to mod teams? I'm making the connection there in terms of thinking of some of the smaller teams out there.

Answer: It doesn't help them. In general, all the technology progress has been essentially reducing the ability of a mod team to do something significant and competitive. We've certainly seen this over the last 10 years, where, in the early days of somebody messing with DOOM or QUAKE, you could take essentially a pure concept idea, put it in, and see how the game play evolved there. But doing a mod now, if you're making new models, new animation, you essentially need to be a game studio doing something for free to do something that's going to be the significant equivalent. And almost nobody even considers doing a total conversion anymore. Anything like this that allows more media effort to be spent, probably does not help the mods."

A pretty honest and insightful answer, if you ask me. This is a feature that'll allow the big boys to make ever-more realistic environments, but it'll mean indie developers and mod developers will have to work that much harder to make anything comparable.

But HOW does it work?

Carmack says:
And one of my early suggestions to them was that they consider looking at an approach where you just use one monumentally large texture, and that turned out to be 32,000 by 32,000. And I - rather then doing it by the conventional way that you would approach something like this (i.e. - chopping up the geometry into different pieces and mapping different textures on to there and incrementally swapping them for low res versus high res versions), just let them treat one uniform geometry mesh and have this effectively unbounded texture side on there, and use a more complicated fragment program to go ahead and pick out exactly what should be on there, just as if the graphics hardware and the system really did support such a huge texture.

What does it mean? Unless I missed something, the closest approach to describe how MegaTexturing works is "a more complicated fragment program to go ahead and pick out exactly what should be on there". So? Carmack talks about how awsome the technique is but he won't tell us how it works in reality. Of course, he has no obligation to tell the world his trade secrets, but the article itself seems mostly just to be there to hype this technology.

Video card support for mega texturing

Something not mentioned is what video cards support the mega texturing technique. The new NVIDIA 7900 hardware (a card aimed specifically at gaming computers [widowpc.com] enthusiasts) supposedly supports it. However, the only real way to test that is to get a copy of the new Doom.

Good enough? No way.

How can anyone say graphics today are good enough? They're damned good compared to what we used to have (check Half-Life against Half-Life 2), but we're nowhere near the realism required for a truly immersive experience.

Sure, gameplay is the most important part of the game, but presentation, including graphics, sound, and general design, play a vital role in creating atmosphere. You could be playing the most amazingly fun game in the universe, but if it's running at 320x240@4bpp/43Hz interlaced with nothing but flat colours and no textures or sprites anywhere, and sound consisting of a single beep at the start of the game, you're looking at a pretty bland experience.

Conversely, you could be playing the most realistic game in the universe, and it might be about as fun as dragging a sack of bricks up a hill. That's why it's important to let the people who are in charge of these separate tasks do what they have to do. The programmers and designers can focus on gameplay, the graphics designers and artists can focus on the graphics, and the sound engineers and composers can focus on the sound. When it all comes together, you can have your cake and eat it, too, as long as each does an equally good job.

Realistic graphics can only add to an already-great gameplay experience. However, high-resolution does not equal realistic graphics. The issue with a lot of the next-gen consoles is that the high resolution absolutely has to be married with massive textures and aggressive filtering in order to create the same effect as their standard-resolution ancestors/competitor. You can have your games running at 1920x1080, but if you can't push the big textures through, you're not going anywhere but to mudsville, population your monitor. The higher the resolution, the more the need for higher resolution textures to create believeable graphics. If the textures are blurry, then you've successfully shot yourself in the foot as far as presentation goes.

Like I said, great gameplay is paramount. Graphics and sound are secondary - However, bad graphics and sound in a modern video game take away from the presentation of the game. If you're playing a first person shooter, and you've got a pistol that, for no real reason, looks like a twig, and the sound it makes is some guy saying "Bang", it'll be funny at first, yes. But it'll also get quite annoying very quickly and you'd end up hating the game for it.

Or at least, I would.

is it me or ..

games based on Valve's game Engine look graphically more polished and have more realistic effects (fog, light diffusing through fog, moving water ..etc). I have personally played quake2, ut2004, CounterStrike, CounterStrike-Source (or CS2). Haven't played doom3 or quake3.

is it a matter of DirectX vs. OpenGL?

no troll, just a gamer trying to get my head around the engine technology.

Megatextures video

This hi-res movie [gametrailers.com] of Quake Wars went up recently, and honestly it doesn't look that great. The ground textures looks flat and splotchy. I understand that the "MegaTexture" tech may be about making games run faster, not necessarily a graphical imporvement, but, for example, I've seen screenshots and videos of S.T.A.L.K.E.R. from a couple years ago that looked much more photorealistic (the environment, not the character models which were below average looking) and when I played a leaked version it ran really fast and had huge draw distances and enviroment to explore.

Interesting, but a step in the wrong direction?

His fundamental argument -- texture tiling is just a form of data compression -- is at the heart of why I think this is not the right direction to go.

Fundamentally, texturing objects is a way of saying "this object is made of something that looks like this". Using textures as mere "appearances" is a shortcut for representing geometry and physical properties. E.g. if you represent a door in a wall made of bricks and mortar as a rectangle with a picture of bricks and a door on it you are taking a shortcut. If you're going to build highly complex environments, giving artists a 4GB bitmap to paint isn't going to suddenly fix your problems. (Carmack doesn't say that it will either.)

In the end, a balanced use of geometry and textures will get you far more than going hog wild with one or the other.

I also wonder if this is an appropriate way to use artists' time. If you imagine a typical World of Warcraft "zone" suddenly your megatextures aren't going to come close to cutting it anyway. If your artists end up filling megatextures with repeating fill patterns ... what have you accomplished?

How Does This Work?

I've seen bits on mega texture for awhile but I have yet to be able to divine how the hell it is suppose to work.

My best guess is that one starts with a tiled texture like you would in any other game but that some engine allows artists to add modifications to the texture in different areas. Thus you take up less memory than actually having a full texture of that size but each area has it's own unique touchups.

Is this really what it does? I'm getting really frustrated at these stupid little gaming articles that never really explain the tech.

Wheres the Beef

Its nice that finally someone has used this kind of technique in a game, but unless Carmack gives more detail, things like this has been done before. ex. SGI's clipmapping [virginia.edu] technique

For more general 3D check out imageSynth

Luxology and Allegorithmic have this really cool concept. They are taking images and using them as the seeds to generate procedural texture shaders.

I can't explain it well enough, so take a look [allegorithmic.com].

I don't know how useful this is in the game realm where things have to be realtime, but it could definitely be very useful for film and video work. I am eager to try it out.

Charles Shackleford: Slashdot Editor

"Left hand, right hand, it doesn't matter. I'm amphibious."*

* The basketball "Shack" as opposed to "Shaq."

Flight sims

Flight sims would benefit from this, since you're often several thousand feet int he air looking down at dozens of square kilometers of texture at any one time. Low resolution or repeating textures soon become very apparent.

How long till we can incorporate something like this into FlightGear? Or is it bogged down with IP issues?

Google Earth

Let me get this straight, what he is advocating as MegaTexturing seems to be the use of arbitrary-sized continuous textures which the game engine manages in texture memory on the fly, instead of the normal practice of using layers of fixed-sized tiled texture maps. Isn't this what Google Earth does already?

Mega-Textured Advertisements

Given John Carmack's recent statement of support for in-game adverts, it will not surprise me to see a gorgeous, mipmapped, mega-textured plug for Coke® or a Hummer® in id's next title.

Talk about missing the point

Do some elementary calculations kiddies.

    32,000 x 32,000 is 1024 million pixels. (A gigapixel?)

    Presuming 16 bits per pixel, that's a couple of gigabytes. That's not going to be sitting in texture memory. Think about it.

    The way I read this is pretty simple. Presume you have a texture that large to work with. Not necessarily just one but potentially many. Now go to town on the art.

    Unless I'm completely misreading his intent, I suspect a decent portion of the magic is probably in the pre-processing. From what I can gather he's proposing that instead of getting your artists to do the compression by constructing textures and appying them to geometry, he's proposing the artists go insane and then get the pre-processor to chop it up into optimum texture slices.

    Which, when you think about it, is kind of a "duh" moment. Of course code is going to be better at compression than human beings are.

    "John Carmack. He's a pretty famous game programmer, and together with John Romero he made FPS games popular."

    This is kind of a myth. Romero pretty much had nothing to do with it.

    ID's games have been winners because of one simple thing. Technology. In game design terms they've always been example of what NOT to do. Quake was a sweet piece of rendering but other than that it was one boring as hell game. I have never completed it because I cannot do the same thing over and over again in different environments and find it entertaining.

    "What about multiplayer?" I hear you cry. Simple. That's another win for tech. Quake multiplayer pretty much lucked out in the gameplay stakes. Nothing out there like it at the time and frankly Team Fortress was a much bigger step in gameplay terms than simple deathmatch ever was.

    I have never seen anything which demonstrates that Romero even remotely understands the fundamentals of good gameplay. The guy is a level designer and gameplay is something which goes far beyond simple architecture.

    ID creates the engines. And then Raven Software (or Valve) creates good games out of them. Quake was boring, Half_life was brilliant. Quake 2 was marginally better, Soldier of Fortune was vastly superior. Doom 3 was brilliant technologically speaking but the gameplay was a gross disappointment. (Dark thing in corner, BOO! Dark thing in corner, BOO!). Quake 4 was significantly better (although not up to Raven's usual standards).

    ID does good tech. ID does good art (environments and creatures), it even does good level design. Id just doesn't do good gameplay and never has. On a fundamental, they just don't get it. They think good gameplay is a single player version of deathmatch. They think it can be dumped down to walking up to stuff and going "activate". They've got a few toys in Doom 3, but there is precious little in the way of actual game DESIGN.

    I keep hoping their next game will finally demonstrate that they've got it, but they're stuck in DeathMatch nirvana in which multiplayer is king and single player is just the player shooting at bots with varying scripts.

What is actually going on ?

Ok, so we've had the usual "games are not what they used to be" discussions, and we've had the hype. So any ideas on what the nitty-gritty is ?
Sound to me a bit like the old "fractal compression" technique that was all the rage several years ago. Here you would presumably have one (or more) extra levels on indirection in your texture, so that, at a lower res, you could say "get these pixels for over there, using this transform". This matches quite well the hints about how tiling is just a special form of compression. The trick is (as it was for fractial compression) how to optimally generate the transformations.

Geometry level of detail

Level of detail wise, the terrain does not render with any sophisticated geometry morphing situation. That's one of those things that for years I think most of the research that's gone into has been wasted. Geometry level of detail on terrain...there have been thousands of papers written about it, and I honestly don't think it's all that important. The way the hardware works, you're so much better off setting down a static mesh that's all in vertex and index buffers, and just letting the hardware plow through it, rather than going through and having the CPU attempt to do some really clever cross blended interpolation of vertices.


Ahh, the sweet sweet sound of infinite memory! Geometry detail algorithms ARE useful and relevant when you are dealing with tough memory constraints, especially on embedded platforms. Sure, it's nice to be able to just leave all the models in memory and just glDrawElements over and over, but this is only an option for platforms with a lot of memory or those that can quickly load the data over and over.

Texture Compression

Granted, I didnt read through TFA thoroughly, but here's my thoughts on that:

The urge for bigger (and thus more detailed) textures is nothing new. I remember when S3 introduced S3 Texture Compression (S3TC) and they had some tweaked maps for Unreal out - it was a blast.
Then M$ licensed it for the upcoming DirectX and dubbed it DirectX Texture Compression (DXTC).

Devs claimed that DXTC was incredibly slow due to the framework, so it essentially only worked on S3 cards with special S3 graphic libs. (MeTaL) I dont have to talk about the S3 market penetration, do I ? ;)

The maximum texture size advertised was 2048x2048, then. Given a rather limited bandwidth of AGPx4, which i s roughly 1GB / sec. (talking AGP 4x pumped).

A Savage4 had less than 1GB/sec bandwidth. And in said Unreal levels were barely playable.

Nowadays graphic cards utilize as much as 40GB/sec and above.

So, lets assume the following: S3TC compressed textures sized 2048x2048 were compressed at a ratio of 1:8
A texture would take up around 512 KB.

Now, those mega textures would sport 9x the size. Even compressed 1:8 one texture would take up toughly 135 MB.

While the improvement in bandwidth is ca. factor ~50, the difference in actual size is factor ~300.

If it wasnt really viable then, how (or, more interesting: why) would it be feasible now ?

Optomized?

Kann wi optomize artical speeling to?

Re:Over the hill

John Carmack. He's a pretty famous game programmer, and together with John Romero he made FPS games popular.

Re:Ah, but

Are you complaining because ID games are retarded? hehehehe...

If games had to have a "story" and complicated environments it may mean that they couldn't pump out 80 games a quarter... /me trying to find that damn red skull key.

Tom

Re:Ah, but

The graphics are good enough already. Latency is too high for proper internet play - at least, I don't like doing really well at BF2 for ages then getting killed at the last minute because 0.1 secs of lag gets my squashed by my own side's tank...
Get a real internet connection.
- Mr. 4ms ping, never leaving campustown