First Killzone screenshot/details? So says USAToday..

Personally i dont think texture quality matters as much in FPS. You're pretty much always in constant motion for the most part and the action is more often then not fast paced so you dont really have time to stop and stare at the ground or wall like you would in Oblivion which also had low res textures in many places. Half Life 2 didnt exactly have superb textures either but the art and game play carried that fine. Granted its a few years old now but my point is you didnt notice it and i'd bet most wouldnt if they played it today either.

Also keep in mind the memory limitations on all consoles. You allocate your resources as best you can.
 
I think that Killzone 2 may be one of the first FPS games to significantly sacrifice screenshot quality for ingame vividness. Thus, they put the textures where the matter when you play, not when you analyze screens.
 
Ive sighned up to the playstation forums and sent some PMs to a couple of the Killzone devs asking them if they would'nt mind sighning up here and talking to us :D
 
Personally i dont think texture quality matters as much in FPS. You're pretty much always in constant motion for the most part and the action is more often then not fast paced so you dont really have time to stop and stare at the ground or wall like you would in Oblivion which also had low res textures in many places. Half Life 2 didnt exactly have superb textures either but the art and game play carried that fine. Granted its a few years old now but my point is you didnt notice it and i'd bet most wouldnt if they played it today either.

Also keep in mind the memory limitations on all consoles. You allocate your resources as best you can.

Agreed.

I bet most of the people complaining are the same people who have no problem playing counterstrike and Half Life one to this day..

Personally I think the textures are fine since they are easy to ignore and they don't take away from the overall visual impact of the game..
 
I completely disagree.

Why ? don´t you think that in the future the best way to get CGI truly levels is using polygons and real raytracing methods ?
Isn´t Intel in that boat, at least if what is romoured about Larrabee is near the truth ?
 
What do you understand by using pollies?And think hard before answering, don't rush things.
 
sorry to be off-topic here but where is London_Boy? he use to be in this kind of thread but there is nearly 1000 posts and i dont see him.
 
Why ? don´t you think that in the future the best way to get CGI truly levels is using polygons and real raytracing methods ?
Isn´t Intel in that boat, at least if what is romoured about Larrabee is near the truth ?

What exactly makes you think that offline CG doesn't use shaders?
 
What exactly makes you think that offline CG doesn't use shaders?

I don´t think so!. It is only a discussion derived from a appreciation i made about heavy usage of shaders to make things look more real. I prefer the look of characters built of many polygons rather than built from less polys and more normal mapping. And I would also build the bricks of a wall, imperfections and other things based on polygons rather than with shaders, although in this generation this is impossible. But that´s just my opinion, Fran seems to think other way.
 
I don´t think so!. It is only a discussion derived from a appreciation i made about heavy usage of shaders to make things look more real. I prefer the look of characters built of many polygons rather than built from less polys and more normal mapping. And I would also build the bricks of a wall, imperfections and other things based on polygons rather than with shaders, although in this generation this is impossible. But that´s just my opinion, Fran seems to think other way.

No he doesn't..

It's obvious that greater polygonal detail will increase the visual impact (but only in the shaders ability to accurately "describe" the objects appearance from the information given) but this same increase can be obtained by providing that information in other forms (more and higher resolution maps for example)..

In the purest sense, vertices don't have the capacity to describe complex, life-like materials since their purpose is to provide structural information above anything else..

just adding more vertices will only get you so far if your still using SM 2.0 (capability equivalent) shaders..

It's the right balance of both that can ultimately increase fidelity into the realm of true photo-realism..

But then again.. Vertices themselves provide nothing more than information which can also be described using other means (voxels, virtual atoms, sophisticated boolean geometry derrivative..?) and thus, could one day be put to the wayside in favour of a new data model..

Shaders however probably won't go away for a very long time (unless raster displays do..)
 
Funny thing is, that quote does not address the low resolution of the textures, just the fact that they have variety and decided against upping the reflection levels.

Nobody is saying it doesn't have variety or that the lighting/material selection (lack of reflection) is bad. People would just prefer higher res textures. I would, but I realize there are limitations with limited ram available.

I'd like to see the textures improved but if it shipped with the same res textures, I'd still enjoy the game and still be impressed.

Point is it doesnt matter that much if they are low resolution. They are good in other ways
 
No he doesn't..

It's obvious that greater polygonal detail will increase the visual impact (but only in the shaders ability to accurately "describe" the objects appearance from the information given) but this same increase can be obtained by providing that information in other forms (more and higher resolution maps for example)..

In the purest sense, vertices don't have the capacity to describe complex, life-like materials since their purpose is to provide structural information above anything else..

just adding more vertices will only get you so far if your still using SM 2.0 (capability equivalent) shaders..

It's the right balance of both that can ultimately increase fidelity into the realm of true photo-realism..

But then again.. Vertices themselves provide nothing more than information which can also be described using other means (voxels, virtual atoms, sophisticated boolean geometry derrivative..?) and thus, could one day be put to the wayside in favour of a new data model..

Shaders however probably won't go away for a very long time (unless raster displays do..)

So is that we need raytracing as well as many vertices. I thought that raytracing algorithms could recreate all kind of surfaces ( wood, metal... ). Wouldn´t that take the place of pixel shaders, or at leas of most of them ?
Maybe i should have added to my prior sentence: "future is geometry AND RAYTRACING". But yes, that is very far away from now.
On the other hand KZ2 engine seems to approach more to this side of tech by having many vertices and a advanced lighting engine to show them, far from even a ray caster but good anyway. It is clear that power is limited and this kind of approachment has its own and huge limits nowadays, that is why people will always see mistakes.
 
Last edited by a moderator:
So is that we need raytracing as well as many vertices. I thought that raytracing algorithms could recreate all kind of surfaces ( wood, metal... ). Wouldn´t that take the place of pixel shaders, or at leas of most of them ?
Maybe i should have added to my prior sentence: "future is geometry AND RAYTRACING". But yes, that is very far away from now.

Raytracing has memory implications..

VAST memory implications..

Do a search in the Tech forum for more info..

In the meantime lets get back to KillZone!

:D
 
So is that we need raytracing as well as many vertices. I thought that raytracing algorithms could recreate all kind of surfaces ( wood, metal... ). Wouldn´t that take the place of pixel shaders, or at leas of most of them ?
Those raytraced materials basically are pixel-shader algorithms. For each ray, you perform caluculations and texture operations to define the lighting of that ray, which is what pixel shaders do for each surface.
Maybe i should have added to my prior sentence: "future is geometry AND RAYTRACING". But yes, that is very far away from now.
Geometry is just too expensive. Even using offlive CG, you'll use normal maps for small details rather than model in the miniscule. You can use displacement mapping which works much better up close, and in the context of that KZ floor you'd use modelling, but normal maps aren't going anywhere soon, if ever (counting similar techs like parallax mapping)!
 
Taking a diffrent slant not really discussed (due to I admint not being seen in any form yet) what do you guys think of multi-player implications of footage we have seen.

Here are some of the things from the footage that may have gameplay impact on MP.

Cover system both firing around walls, and blind firing in First person.
Vaulting walls (to get at defenders?).
Tracer fire making enemy players more easily targeted (i.e. trace back the path of bullets to see where the enemy is).
Very dark environments for potential sneeky game play
Distructable environments (may be some cover could be distroyed to get at enemy defenders).
Ability to sprint (from cover to cover)
Close quater interior MP may be quite confusing (realistic?) due to the amount of strobe lighting from guns/grens going off!

Just thinking of potential MP impact on the above. There is no evidance that any of the FP footage will make it into MP. However its nice to speculate!!!
 
I don´t think so!. It is only a discussion derived from a appreciation i made about heavy usage of shaders to make things look more real. I prefer the look of characters built of many polygons rather than built from less polys and more normal mapping. And I would also build the bricks of a wall, imperfections and other things based on polygons rather than with shaders, although in this generation this is impossible. But that´s just my opinion, Fran seems to think other way.

Modelling the high frequency details of a surface with polygons is an inefficient way of driving current graphics hardware based on rendering quads. What you call "heavy use of shaders" is nothing more than a more or less lossy compression method: you want to obtain the same quality of a high poly count model that would be extremely inefficient today and consume a lot of memory, by using clever per-fragment techniques (POM, Relief mapping and so on) which also reduce the memory footprint or your data set.

With "lossy" I intend a visual difference that can't be noticed by an average user.

The alternative is rendering lots and lots of 1-fragment polygons that waste an enourmous amount of shader power that could be used to add more effects and increase the quality of the scene, to obtain a questionable increase of visual quality that will be even less noticeable when more shader power is available and more clever per-fragment techniques can be implemeted in the future.

That's my point and why I completely disagree with your statement: in the future I see more work being moved from polygons to fragments. More or less the same number of polygons in the average scene and more and more complex and interesting fragment shaders.

Unless we see completely different GPU architectures, which is something I don't really see happening in the near future (4/5 years).
In long-term? Well, we could speculate for pages and pages and probably be all wrong :)
 
Modelling the high frequency details of a surface with polygons is an inefficient way of driving current graphics hardware based on rendering quads. What you call "heavy use of shaders" is nothing more than a more or less lossy compression method: you want to obtain the same quality of a high poly count model that would be extremely inefficient today and consume a lot of memory, by using clever per-fragment techniques (POM, Relief mapping and so on) which also reduce the memory footprint or your data set.

With "lossy" I intend a visual difference that can't be noticed by an average user.

The alternative is rendering lots and lots of 1-fragment polygons that waste an enourmous amount of shader power that could be used to add more effects and increase the quality of the scene, to obtain a questionable increase of visual quality that will be even less noticeable when more shader power is available and more clever per-fragment techniques can be implemeted in the future.

That's my point and why I completely disagree with your statement: in the future I see more work being moved from polygons to fragments. More or less the same number of polygons in the average scene and more and more complex and interesting fragment shaders.

Unless we see completely different GPU architectures, which is something I don't really see happening in the near future (4/5 years).
In long-term? Well, we could speculate for pages and pages and probably be all wrong

Out of intrest do you think that the Cell with its 750 thousand triangle rendering @ 60fps (from edge tools) per SPE. Freeing up the RSX to do more shading?

There have been arguments on this forum that generally a huge increase in phsyics processing would have to be made for noticable diffrance to be seen in game. Given the sheer power of the Cell eventually would it be possible that coding will be efficient enough that game code (AI, Phyiscs, Sound) might be restricted to the PPE and 2 SPE's), allowing 4 SPE's to contrabute very heavily towards complex polygons?
 
Point is it doesnt matter that much if they are low resolution. They are good in other ways

Read it again. It has nothing to with the textures. Its the lightning engine, Guerilla (spelling?) hasn't put on some kind of special coating on the textures to make the light look more realistic, they have a very good lightning engine, the textures themselves are normal ****ty ones.
 
Back
Top