Today's Hardware on Tomorrow's Games

GweeDo writes: "Anandtech has gotten their hands on a recent build of the Unreal Engine to give today's hardware (Geforce 3 ti's and upper-class Radeons) a run for the money to see how they will do on tomorrows games. The article is here and quite a good read ..."

Comments

Fast, Hard-core 3D GFX != Good game

[Masem]

I've got a very bad feeling that the gaming industry is heading towards a black hole of development. Sure, the GF3 and other graphics boards are truely amazing in terms of HW, with all the new pluggable rendering devices, hardware T&L, etc. And I'm certainly not going to complain about the graphics in a game that take advantage of such graphics.

However, you can put all the greatest graphics in the world, but if you don't add something interesting in terms of the game itself (plot, gameplay (both single and multiplayer), etc), then all you've got is a pretty looking game that no one is going to buy. And too many of today's games are just that; there hasn't been anything 'different' in the FPS arena since Half-Life, Deus Ex and No One Lives Forever, Diablo 2 in terms of RPGs, and so forth. There's only two interesting areas of games that I've seem them take great steps above their predecesors as to make them different; first is the X4/real-time strategy games such as Black & White and the recent Dune title, which are now combining good 3d engines with good gameplay (though Myth would be the first real entry in this catagory). The other is the simulation area: recent entries of games like Startopia combine the graphics and a rather detailed but playable ruleset to make a good game.

So while the hardware makers keep pushing out better cards capable of running all the graphics effects today, the game makers seem to be too tied up in taking advantage of that and not of improving the underlying game itself. I'm hoping that we hit a plateau in the graphics card ability, as once that is hit, then the game makers will turn back to the game since they can no longer optimize the pretty-ness of the game itself.

Re:Fast, Hard-core 3D GFX != Good game

[NutscrapeSucks]

I'm hoping that we hit a plateau in the graphics card ability, as once that is hit, then the game makers will turn back to the game since they can no longer optimize the pretty-ness of the game itself.

There's also the real problem that game companies are outpacing the market penetration for top end hardware. While you can do wonderful things with a GeForce 4 or whatever, it's a real question if there will be enough GF4 users to profitably sell games to.

Gone are the days when you could slap Doom or Quake on pretty much any old computer in your office and have a network shootemup. Requiring recent 3D hardware eliminates the vast majority of PCs in the real world.

Another example -- it sounds like Unreal Tournament greatly outsold Quake III. UT can be played unaccelerated (don't laugh, I know people who do this and have fun), and it played fine on the shitty i810 PC I had at an old job. QIII barely played on the hardware that was out at the time of it's release (PII, TNT2, for example). Half-Life is another game friendly to low-end boxes. Not to mention SimCity, RR Tycoon, Sims, and other big sellers that didn't require much hardware-wise.

Not to mention that the average user probably finds the whole world of 3D card to be a mess of confusing brandnames, limited retail outlets, driver woes, hardware upgrades, and so on. It starts to become real work for something that's supposed to be fun.

I guess I'm being presumptuous in telling game companies how to run their businesses. Just that from what I've heard, it sounds like the PC game sales are in the dumps, and only half the problem is unoriginal game concepts. Maybe they should consider quitting chasing the l33t gamer crowd or techies like myself that find it fun to keep up with the gear, and get back to the broader market of people who just want to blow off some steam after work.

Re:Fast, Hard-core 3D GFX != Good game

[uebernewby]

game makers will continue to put out good games. Top of my head, Black and White

Hmm... I'd say you just named the one title most of my friends (and myself) consider to be a prime example of "looks good but sucks". No one I know played it for more than a few hours, because in the end it turned out to be Populous With Extras That Take Away The Fun. I'm really glad I was smart enough to try the warez version instead of coughing up the full USD 50 for it like most of my buddies.

Re:Fast, Hard-core 3D GFX != Good game

[ergo98]

Doom blew everyone away with its revolutionary 3D engine, but the game was pointless. Fun, but pointless, and got old fast

Nice job with revisionist history there. The reality, of course, is that for the overwhelming majority of people Doom was extremely engrossing (the classic is jumping out of your seat when one of those red bull things appeared) and was directly responsible for billions of hours of slack time. Duke Nukem 3D was very similar in that it's a simple concept, but I played that game multiplayer for hours upon hours upon hours.

including everything id has done since

Now this is just dumb. While I haven't ever really gotten into Q3, Q2 gave me thousands of fun hours, especially with the mods. The Quake series, and this is something that many pundits fail to realize, is more of a sport than a RPG : You excel and because exemplary in it just like you would perfecting the perfect dive or running the 10s 100m, and it's the same sort of quest for perfection that draws people to excel. When I see complaints about the Quake series I often wonder if these people expect some RPGing to break out in the middle of the Olympics : Maybe the downhill skiers can have a pseudo hill economy. I mean otherwise they're just falling with gravity right?

The whole point of this? Don't discount a game just because it's not a genre that you prefer, and don't presume that if an element works in one game (i.e realism, or RPG factors, etc.) that therefore it should be in all games.

Bummer

[Matey-O]

Show of hands here, who actually wanted to SEE pics of the new engine in action?

What's the point of saying 'Gee these are really nifty in this demo' if we've got no visual point of reference?

A major part of a GPU benchmark is how well the display _appears_

Re:Bummer

[Score0, Overrated]

There's this page with some screenshots of the engine in development.
The latest is a few months old though, but there's pictures of landscapes and special effects.

Kyro II

[13Echo]

The even more interesting thing is just how well the Kyro II line of card (Herc 3D Prophet 4500) is standing up to the GeForce 2 line of cards. That's not bad if I say so myself. At 1024x768x32, the card that nVidia dubbed "TNT 2 class" is keeping up with the GF2 pack, and is right behind the high-end GeForce 2.

Personally,I think that the Kyro 2 is the best deal in video accelerators right now. It's got plenty of juice for current games, produces a beautiful image, and can be puchased for a price as low as $60-$70. There really is no reason to buy a GF2MX considering the performance gain that you get with a Kyro 2. And, when the chips finally get a hardware T&L unit, they will be smokin.

Now, if only they would release those Linux drivers...

I work with the engine every day..

[Anonymous Coward]

I work at one of the many companies that license Epics Unreal technology and I can tell you something of what is happening with the engine compared to the older titles like Unreal and Unreal Tournament.

There are a LOT of rendering improvements. The new renderer depends heavily on the GPU to offload the triangle rendering from the CPU. There are new primitives dubbed in the engine that are there to explicitly call for GPU support and render very, very fast.

This is why most games based on the new engine is going to have a lot more polygon detail and can use these rendering primitives to step up from blocky, repetitive levels to much more realistic environments with more depth.

Terrain is done in a similar manner, and the editor tools allows you to paint and modify the terrain in realtime preview. Multiple layers are allowed and you can control the blending in many ways.

A lot of other small improvements are in as well, such as texture compression, native skeletal animation, advanced particle systems, render anti-portals (for manual occlusion specification).

And the thing runs in very acceptable FPS :-)

(sorry about being an AC but I don't want to be pinned to the wall and shot for saying anything I shouldn't have)

D3D vs OpenGL

[BadBlood]

I also didn't see any mention of which graphics API was used. My hunch is that it's Direct 3D.

What follows is simply my opinion: I prefer the looks of OpenGL rendering on Nvidia hardware. My order of preference from a visual perspective was OpenGL, Glide, then D3D. I know Daniel Vogel (once a Loki guy - PS: Good career move) was responsible for most of the OpenGL work on UnrealTournament (i.e. using the S3TC-based textures on the 2nd CD), so my hopes are that this new engine will have OpenGL rendering.

I definitely take a performance hit going from D3D to OpenGL, but with pageflipping enabled in the drivers it's not too bad. I also am willing to do this for my perceived visual enhancements.

Ultimate 3D quality... plus Blast from the Past 3D

[dpilot]

I look at the idea of playing Capture the Flag on Quake-family games, and see the quest for ever-more-real 3D, and wonder why people just don't go for the Ultimate. Pick a decent night, go outside, and play Capture the Flag. Real Reality, the Ultimate in Virtual Reality. I remember real Capture the Flag from Boy Scout campouts, and the nights weren't always that decent, but that was part of the fun.

The 3D gaming is getting just a bit bizarre, but I'm still reasonably happy with Quake3 on my Matrox G400 - bought on the strength of 2D image quality as well as Open Source 3D support. Unfortunately the latest'n'greatest drivers seem to be headed back to closed source.

Wow. A lot of misconceptions floating around...

[TheGreatGraySkwid]

Firstly, a correction of the initial post, this is not just "a recent build of the Unreal Engine," it's a build specifically designed and packaged to stress rendering hardware to their limits. The 2 games nearing release using the Unreal Engine (Unreal Tournament 2 and Unreal 2)will be using a dramatically different codeset than this "UPT 2002" does, and those games will be better optimized for more efficient utilization of system resources than this thing is, while still using a number of cutting edge features that this thing doesn't (like custom particle engines, vertex/pixel shaders, and nifty stuff like that).

Quoting Mark Rein, who works for Epic:

The Unreal Performance Test 2002 (UPT) has nothing to do with any of the games using our engine and should not be used to draw conclusions about game performance. I'll contact Anand and make sure he knows to highlight this. The benchmark is designed specifically to push the latest and greatest graphics cards as hard as we can. UPT is about the future, not the present. All of this will become clearer over the next few months. We will also be adding more features and content to UPT2002 to push things even harder. "Influence, Educate and Improve" will be our motto for the Unreal Performance Test. A lot of cool things are planned. In popular game development lingo what you're seeing now are some preliminary results from what equates to an early alpha version. Stay tuned! Afterthought: I guess I shouldn't say "as hard as we can" because I'm sure we could push things even harder if we weren't so busy making our game. The Unreal Engine is no longer CPU bound so if you want to make a game that pushes the absolute upper limits of xxx [i.e. insert the name of some imaginary future card that has massively higher fillrate and poly throughput and might be announced the week after next] you certainly could but then, of course, nothing earlier than that could run it. This test is very much about making something to test today's high end cards and the cards of the future. Putting the lower-end cards in the test wasn't really fair because the content wasn't designed to support those levels of performance. Certainly the games using the engine this year wouldn't want to be so aggressive with the content and detail settings. To borrow a phrase from Spinal Tap, the UPT detail settings are set at "11". Unreal2 and Unreal Tournament 2, for example, certainly won't be set at "11" because we want EVERYONE to be able to enjoy them. I suppose Unreal Championship could be dialed up to "11" because it's on Xbox, maybe even "11.5". One more thing (this will be corrected in the article shortly): Dan Vogel said the flyby had "as many as 100,000 triangles." - to clarify that it should say as many as "100,000 triangles in view." There are certainly a LOT more than 100,000 triangles in the demo.

This is all being discussed extensively in Infogrames' Unreal 2 forum.

Oh, and one more thing: Unreal 2 will be D3D only, and I wouldn't be surprised if UT2 is the same (although I don't follow it as closely). You may commence your moaning and bitching.