New Killzone screen grabs

[London Geezer]

If the guns really have 10.000 polygons on them, then it would be a big fat waste of resources they could use for something else...
(Oh, just read about the multi-texturing part, take that back).

I think this whole polygon talk is total nonsense, i was playing around with the settings on Farcry (PC), and by changing the LOD settings, the polygon count went from 3-5 million (default LOD) to 25million (when i took it off via the console - LOD set to OFF, with the engine rendering everything is in the view at highest geometric detail). Framerate on my 5900U was pretty much identical BTW, which is great.

I must say, the 2 settings looked practically identical, and i really appreciate the fact that it is now trivial to "cut" useless polygons and details that can be used for ther parts of the engine. Therefore i think it's useless to talk about "peak" polygon performance as such.

And let me tell u, if Farcry is pushing around 5 million polygons looking like it does (amazing draw distance and vegetation, pretty high poly characters and structural details etc), then i'll be VERY VERY happy if Killzone "only" pushes around that amount, considering it is for PS2.

It all depends on the LOD engine really, if Farcry looks identical whether it pushes 5 or 25 million polygons, then this talk about "peak" and "average" polygon performance if utterly useless.

 

[Bohdy]

I'm still not sure if it Killzone even equals Metroid Prime graphically yet..

What do you people think? The two engines strike me as somewhat similar.

 

[Deepak]

I hope they show KZ in a playable form at E3.

 

[Guden Oden]

I'm still not sure if it Killzone even equals Metroid Prime graphically yet..

What do you people think? The two engines strike me as somewhat similar.

Comparing the two is unfair IMO. MP is room-based where all resources can be heaped on to form a highly detailed yet entirely enclosed area. Also, almost all resources are used by the rooms themselves with little left for enemies - as there generally are so few of them.

KZ (which by the way is a rather scary contraction) is much more open, so naturally you wouldn't think the game uses polys in quite the same way. For example, draw distance is likely much greater, even the biggest of rooms in MP aren't THAT big, and THE biggest room (the boss lair in the ice world, I believe) isn't particulary detailed...

So it's apples and oranges, really. So far we've only seen shots of really old KZ builds, and crappy scans and blocky videos of somewhat newer builds, I say let's wait to make any judgements before high-quality shots of an at least near-final build are available...

 

[Bohdy]

Well... KZ also seems to have a lot of "room" areas from the footage shown so far, and the "open" areas are not that much bigger than the larger rooms in MP. The Phendrana drifts area and some of the larger Magmoor rooms were very large and very detailed.

 

[whipper snapper]

I hope they show KZ in a playable form at E3.

I do believe it is, and the online mode, well there's a chance it'll feature at Sony's press conference. With players in LA & Amsterdam (Guerrilla's HQ) battling it out. 8)

 

[Evil_Cloud]

With players in LA & Amsterdam (Guerrilla's HQ) battling it out. 8)

Small chance.

 

[marconelly!]

What do you people think? The two engines strike me as somewhat similar.

It looks nothing like MP to me. MP mostly has much more colorful look, has those typical 'volcano', 'ice', 'snow', etc. levels like you come to expect from more oldschool games, like platformers. It's explosions and energy blasts are in all the different colors of the spectrum. UT games also follow that sort of look and feel. KZ is so grity and dirty looking, it's nothing alike...

To illustrate my point:
MP:

KZ:

All that aside, saying that KZ doesn't 'even' look as good as MP, as if MP is not a benchmark looking game on a more powerful hardware, is more than a little ill-meaning, don't you think?

 

[Lazy8s]

Typical performance is gauged through an unbiased sampling of a game and made highly representative of the set through the vast percentage of all games included in the sample. This methodology is dictated by statistcal theory and is certainly not misrepresentative or reflective of random/arbitrary circumstances compared to someone using their personal bias to choose the most complex samples as "most representative".

If the numbers "deserved" to be higher as some are implying, tests would have run across more instances of that and reflected it as being typical. The results are just as likely to be too high as they are to be too low - which is the ideal statistcal balance.

Typical figures best describe what the game's doing, but peak numbers are misrepresentative even when comparing across multiple games. You most definitely are not getting a worthwhile comparison when you learn what one game's peak numbers are against another game's peak numbers. Unlike typical performance, whose standards are established by the majority condition of the game, there are no standards for how limited a game's scenario might get to result in its peak performance. One game might sustain a lower rate but include an instance where every condition is so simplified that it can peak extremely high, while another game is actually sustaining a better rate but doesn't have any such ideal condition to peak nearly as high (think about comparing a game where objects must be kept close to the camera and highly detailed versus a game using dynamically texturing LOD where dense patches of distant geometry might just come into view and not need to be textured yet.)

Unless you know that the conditions under which the peaks were attained between two games are comparable (you can't assume that this is a given), it's not meaningful to contrast them using that like it is using typical performance.

Phil:

Please quit the polygon nonsense. This has been debated enough and at the end of the day, the games overal appeal is what makes a game, not its number of polygons. If you so greatly feel the need to pursue this argument, please start a new thread before this one gets locked.

It was a question brought up by the topic starter directly related to the footage of this game, to which was provided a realistic guideline. It's been the defensive reactions to what is quite true that is causing the disruption. I've just been trying to clarify the issue so they can understand why the sustained numbers are most applicable.

 

[Fox5]

I sure hope that people are not looking for a misrepresentation when they're inquiring about the performance! I'd guess what they're looking for is the best description of the game's rates - not a peak, some number which is about as revealing as knowing that the trough is 200k polygons/sec or the like.

And developers are most definitely not all spin doctors. Thank goodness. For years, sustained figures instead have been given for games like Daytona USA, Virtua Fighter 3 and Soul Calibur, and real-world performance specs for hardware like Model 2/3 and Naomi2.

A recall an early spec number for Dreamcast saying it could do 10 million polygons per second, it was in the official dreamcast magazine.

What about Shenmue 2? It only sustains about 3 or 4 characters on screen, but peaks at over 20 characters. I guess we can disregard the 20 characters on screen because that's a misrepresentation.

Yep, because the framerate takes a dive after 2.

 

[jvd]

are there any new shots for this game. All the shots I have seen show very bad fogging. It be like playing turok for the n64.

 

[Guden Oden]

Jvd,

You must not have played Turok for a while, because KZ is simply nowhere near that level. Comparing the two is just stupid, all they have in common is they're games with fog in them.

 

[Fox5]

Turok on gamecube.

Turok on n64.

N64

Xbox

Ps2

Xbox

Even for n64 this one looks bad.

N64 on pc.

Turok n64.

Look at those killer shadow effects!

 

[jvd]

whats your point. I'm only comparing the fog in both games. Both games had very bad fog in them .

I don't care what the other graphics in the game looks like. If i can't see 10 feet in front of me because of fog then the game does not impress me.

Under stand what I"m trying to say to you ?

 

[Fox5]

Well, at least KillZone has nicer looking fog.

 

[Jabjabs]

This post is all over the place, first it's on killzone then Ps2's polygon counts and now we are comparing Metroid prime and Turok!

On the Dreamcast polygon figure (yes, i'm a little hypocritical) the 10 million figure was a peak theoretical figure based off the SH4's floating point abilities, the 3 million figure came from a real world software test Sega did in the early days and wasn't an indication of the total power but only a representation for that point in development. The fastest recorded Dc game was Le mans 24 hour at 5 million pps.

Depite this the adverage game barely reached 1 million pps, sort of defeated the wholoe purpose.

Oh an Bohdy I will show you the 4 minute Kill zone video this week end, then you can decide on which is better, personally it would have to be KZ (ignoring frame rates).

 

[marconelly!]

The video posted earlier in this thread shows a pretty large open area with no fog whatsoever...
Fog used on that one level that was most publicized is there for atmospheric purposes, as the level's name is 'Helghast Mist'.

 

[Fox5]

This post is all over the place, first it's on killzone then Ps2's polygon counts and now we are comparing Metroid prime and Turok!

On the Dreamcast polygon figure (yes, i'm a little hypocritical) the 10 million figure was a peak theoretical figure based off the SH4's floating point abilities, the 3 million figure came from a real world software test Sega did in the early days and wasn't an indication of the total power but only a representation for that point in development. The fastest recorded Dc game was Le mans 24 hour at 5 million pps.

Depite this the adverage game barely reached 1 million pps, sort of defeated the wholoe purpose.

Oh an Bohdy I will show you the 4 minute Kill zone video this week end, then you can decide on which is better, personally it would have to be KZ (ignoring frame rates).

I can't believe that the average game barely reached 1 million pps, isn't that what the dc could do without tile based rendering? Well, if that's the case, then it looks like Dreamcast would have been better with a 3dfx based card.(assuming it didn't cost more) I think the voodoo2 approached 2 million polygons per second, and voodoo3 did well over 3.

 

[whipper snapper]

New preview scanned from OPM.

http://albums.photo.epson.com/j/AlbumIndex?u=4177782&a=31170943&f=0

 

[marconelly!]

Typical performance is gauged through an unbiased sampling of a game and made highly representative of the set through the vast percentage of all games included in the sample.

Which is fine is you look at the whole sampling as a group. Their global averaged numbers are informative as a whole, due to large number of games they sampled. However, individual example you brought up is almost irrelevant knowing the way it was sampled.

The results are just as likely to be too high as they are to be too low -

Which is exactly the problem with an individual game. We have no idea if that 7.5M number is typical for the majority of that game, or a high-peak situation, or a low-peak situation. Sure, we can just assume it's not the high-peak, as we know from some developers mouth their games are peaking at 15-18M, but it's still very incomplete information. We don't even know if that's a real sustained performance - it's measured over just several seconds. Combine that with the fact we don't even know what game we are talking about, it becomes even more irrelevant. There could easily be another game that would have scored much better have they sampled another level, etc. That is why singling out one game from their sampling simply does not work - it provides no relevant point of reference. At least when you know what game is being talked about, and what it's peak performance is, you have a clear point of reference. You have a name, and you have performance situation pinpointed - which is much more informative.

It was a question brought up by the topic starter directly related to the footage of this game, to which was provided a realistic guideline. It's been the defensive reactions to what is quite true that is causing the disruption. I've just been trying to clarify the issue so they can understand why the sustained numbers are most applicable.

If only that guideline really was relevant, and if only that was actually what he was asking about, it would all be fine. However, he was clearly talking about peak numbers, and you tried to provide sustained numbers, thus comparing apples and oranges, or trying to prove some point of yours.