Just out of curiousity, does anyone here have an idea behind Sony's reasoning to make a 8 pixel pipeline (with texturing) rasterizer on a console which has a target resolution of only 640x448? Why would not a 2 or 4 pipeline GPU with more flexible functionality make more sense? At the very least proper mipmapping would have largely mitigated the problems PS2 has with shimmering.
Brief inquiry on PS2 GS capabilities.
- Thread starter JesusClownFox
- Start date
- Status
akira888 said:
I've always thought Sony wanted games for PS2 to look as different as possible from the competition, at the same time pushing the most amount of geometry for a long time after launch. That might be one of the reasons behind many decisions they made in designing the PS2, although this is all MY own opinion.
By focusing on geometry and not on textures, PS2 games had the potential to look DIFFERENTLY from the competitors, giving PS2 a sort of "character" that was diverse from PC and other consoles that relied on textures rather than on geometric detail.
Add to that the fact that they couldn't spend £1000 on each machine, therefore having to compromise "somewhere" (see: amount of memory etc") and you might get close to a "reason" of Sony's decision...
Someone here might have different ideas though.
While I'm sure alot of ppl may disagree with me here, IMO they Made a mistake and misread the direction relatime graphics was developing at the time. in other words a cock up in which (fortunately for them) had it's impact mitigated by the sheer degree of penetration the PS2 ended up taking.
notAFanB said:
I guess that's more like it, they believed (and i think they still do) real time graphics is 3D, and that means having loads of polygons, even if that goes at the expense of everything else. Textures and more importantly bump mapping is a way of faking 3D.
Sadly the hardware wasn't at a level where it could substitute the 2 simply because it doesn't push enough polygons yet, but i think this is the right direction. Displacement Mapping and real geometric detail with micropolygons is where we have to go, and Sony is trying to push it in that direction.
Until then we have to make do with textures, or, better, shaders. Although shaders will always be there in some form...
Of course PS2 wasnt anywhere near that level yet, and it's "easier" to get a WOW effect with a system with 64MB of RAM to allow higher resolution textures and a GPU that supports per-pixel lighting than having a system that does all that with micropolygons...
But i guess that was Sony's "idea"... We'll see what happens next time, and the next one and the next one...
Chap:
Not truth, just your opinion. While you may be able to point to the comments of a few devs that support your opinion, you won't find unanimous support for that opinion among all devs, which would be a bare minimum required for it to approach truth.
If that's what you believe, then that fits scenario 1 that I listed above: the majority of gamers don't care about visuals that much. It falls far behind a number of other features in terms of importance. The GS therefore doesn't suck because it's output is far from a dealbreaker for the vast majority of gamers.
See above. The GS is, at bare minimum, not a dealbreaker. And as several devs have pointed out, it isn't exactly the weak link in the PS2. We are talking about video games after all, so if the GS truly were "sucky 3d" then I would expect it's impact on the PS2 success to be very negative. But nothing could be further from the truth - performance analysis of PS2 games shows that the GS isn't even being taxed close to its limit in many cases and yet tens of millions of gamers are already satisfied with the output that the GS offers onscreen. If the part in question is more than suitable to the task required of it, how can you define that as "sucky"?
Meanwhile, you have yet to successfully refute any of my points. So your opinion on what my "needs" are is pretty irrelevant until you do.
Not truth, just your opinion. While you may be able to point to the comments of a few devs that support your opinion, you won't find unanimous support for that opinion among all devs, which would be a bare minimum required for it to approach truth.
If that's what you believe, then that fits scenario 1 that I listed above: the majority of gamers don't care about visuals that much. It falls far behind a number of other features in terms of importance. The GS therefore doesn't suck because it's output is far from a dealbreaker for the vast majority of gamers.
Because we're not talking about 3d hardware that's only meant for display of gee-whiz tech demos in a lab environment, we're talking about 3d hardware that is meant for the consumer electronics sector and its success and suitability to task can be measured in very practical terms as a result.
Am I being accused of not seeing the trees for the forest here?
See above. The GS is, at bare minimum, not a dealbreaker. And as several devs have pointed out, it isn't exactly the weak link in the PS2. We are talking about video games after all, so if the GS truly were "sucky 3d" then I would expect it's impact on the PS2 success to be very negative. But nothing could be further from the truth - performance analysis of PS2 games shows that the GS isn't even being taxed close to its limit in many cases and yet tens of millions of gamers are already satisfied with the output that the GS offers onscreen. If the part in question is more than suitable to the task required of it, how can you define that as "sucky"?Meanwhile, you have yet to successfully refute any of my points. So your opinion on what my "needs" are is pretty irrelevant until you do.
The theory that Sony wanted their console to rely on polygons and that 'PC-like' systems rely on textures seems to be an after-the-fact rationalization.
The obvious truth: the goal of designers and engineers is to do their best to take into account all possibilities and provide the most optimal balance for the time. They don't intentionally try for sub-par. They also don't design for theory (big paper specs), but rather for application (sustainable in-game performance).
The dubious theory in question also draws rather arbitrary definitions for its standards. If one machine can sustain marginally less geometry than another while performing nicer texturing, where exactly does that make the first machine "focused on texturing" for its 3D solution? Maybe they got geometry throughput to the point where they thought better results would be yeilded from using the remaining design budget (die space, R&D time) to enhance texturing rather than to wring more diminishing returns from extra geometry. There's no cut-off line where the balance for creating 3D suddenly becomes "made from textures" rather than "made from polygons".
Even the actuality of things doesn't seem to support the theory. For sustainable geometry, the DC chipset family was very much ahead of the curve for its size and release date too, and it went on to handle more complex T&L conditions than PS2 when it eventually received dedicated T&L in 2000.
The obvious truth: the goal of designers and engineers is to do their best to take into account all possibilities and provide the most optimal balance for the time. They don't intentionally try for sub-par. They also don't design for theory (big paper specs), but rather for application (sustainable in-game performance).
The dubious theory in question also draws rather arbitrary definitions for its standards. If one machine can sustain marginally less geometry than another while performing nicer texturing, where exactly does that make the first machine "focused on texturing" for its 3D solution? Maybe they got geometry throughput to the point where they thought better results would be yeilded from using the remaining design budget (die space, R&D time) to enhance texturing rather than to wring more diminishing returns from extra geometry. There's no cut-off line where the balance for creating 3D suddenly becomes "made from textures" rather than "made from polygons".
Even the actuality of things doesn't seem to support the theory. For sustainable geometry, the DC chipset family was very much ahead of the curve for its size and release date too, and it went on to handle more complex T&L conditions than PS2 when it eventually received dedicated T&L in 2000.
I thought it was more balance, developers can forgo polygons for better textures, isn't that what FF12 reportedly be doing ? Its just early on developers goes for polygons, beyond the point where aliasing takes over and textures becomes dull.
Actually, they use the remaining space of GS, to embedd the memory. Embedded memory will save cost in the long run. As well as providing large bandwidth for effects.
In the end, its a more balance solutions than what the Dreamcast offered.
IMHO, the PS2 was designed with the following goals:
A). economics: spend as little as possible, keep IP inhouse, break-even as fast as possible, collect revenue for future investments
B). performance: create a design, that lasts for at least 5 years under consideration of point A). and possible competition.
Those goals are competing, keeping in mind a user-base in the likelihood of 100 mio units and while looking at their current success and market-share .... I would say Sony did a LOT right.
A). economics: spend as little as possible, keep IP inhouse, break-even as fast as possible, collect revenue for future investments
B). performance: create a design, that lasts for at least 5 years under consideration of point A). and possible competition.
Those goals are competing, keeping in mind a user-base in the likelihood of 100 mio units and while looking at their current success and market-share .... I would say Sony did a LOT right.
chapban. said:
Well don't use my name, because I agree with ERP. It's a basic chip, but it's also insanely fast at what it does do and remains so to this day. Like most developers, I spend a fair % of my time swearing at the hardware. But I'd do that no matter what you put in front of me. So I'm not it's biggest fan, but I'm certainly not "unfond" of it.
I've hated it for not letting me do the things other chips do, but I've loved it again when I've discovered a neat new way to just throw fillrate at a problem which just plain wouldn't work on a PC, XBox, or Cube.
Slapping high-res textures and bump-mapping on everthing in sight is all very well and good but it doesn't automatically make things look nice. ICO didn't need any of that to be one of the most beautiful games I've played.
You could do that on any platform you wanted, but if you have a lazy way out, you tend to take it. So in a way, the sparse feature set of the GS forces developers to take the time to do things differently to compensate. If they can't be arsed - then sure, their games will look better on other platforms - but in return I can't really be bothered playing them.
In short, it has good and bad points, but it sorts the men from the boys, and I kind of like that
So please don't use me in attempt to prove your addled view of the world again...
Kaching,
don you see? there are graphiX and there are graphiX. Just because poorer display didnt turn people off, ain meaning it is poorer display. By your account, we know what happen with PS1/N64 and DC. Can you not say PS1/N64 3D suck compare to DC? YET people are not jumping ship, YET we KNOW PS1/N64 3D is bad vs DC. How about N64 vs PS1? N64 with better bilinear filtering look more smooth than PS1 ever, so much playing N64 games on emulator now is better than PS1. How about SNES vs Genny, same thing too. YES, normal people are less eyeful about pretty tweety 3D, but it still doesnt make something fundamentally yucky, un-yucky just because people arent rejecting it, ESPECIALLY in console wars.
IN FACT, it be widely acknowledge that Xbox graphics wowzers even by the norm, UNLESS we speak of those crazy fanbys around every consoles. You be surprise how many crazed fanbys are present in norm gamers. MayB not totally fanbys, just uninformed in cases. Now did that successfully make things clear? Not that i didnt b4.
NOW b4 someone get cranky over again, i ain saying GS is as bad as PS1 vs DC=Xbox, just another ex.
don you see? there are graphiX and there are graphiX. Just because poorer display didnt turn people off, ain meaning it is poorer display. By your account, we know what happen with PS1/N64 and DC. Can you not say PS1/N64 3D suck compare to DC? YET people are not jumping ship, YET we KNOW PS1/N64 3D is bad vs DC. How about N64 vs PS1? N64 with better bilinear filtering look more smooth than PS1 ever, so much playing N64 games on emulator now is better than PS1. How about SNES vs Genny, same thing too. YES, normal people are less eyeful about pretty tweety 3D, but it still doesnt make something fundamentally yucky, un-yucky just because people arent rejecting it, ESPECIALLY in console wars.
IN FACT, it be widely acknowledge that Xbox graphics wowzers even by the norm, UNLESS we speak of those crazy fanbys around every consoles. You be surprise how many crazed fanbys are present in norm gamers. MayB not totally fanbys, just uninformed in cases. Now did that successfully make things clear? Not that i didnt b4.
NOW b4 someone get cranky over again, i ain saying GS is as bad as PS1 vs DC=Xbox, just another ex.
Chap, stop ruining threads. Stop making this thread that's about a discussion of certian graphics techniques in general about how the Xbox graphics are the best. If you continue to do it and do it in another thread unrelated to the Xbox's graphics in any specific way you will be banned. This is your last warning.
marconelly!
Veteran
You are talking about Naomi 2? But doesn't it also have two GPUs + dedicated T&L processor? I don't think ELAN was the only thing added but another Pover VR as well.
Even then, I think N2 is specced at 10Mpps, admittedly with a large number of light sources (6). I don't know if PS2 would still fare better than it with fewer light sources, though. I'm not sure how much does polygon throughput on N2 increase with less light sources, but there are games on PS2 that peak above 10M.
Hey Chap,
Console's are always based on compromises...
Look at PS1 - 66Mpixel fill rate for flat shaded poly's compared well with graphics cards at it's launch - and geometry throughput was pretty reasonable for it's time.
N64 came out later, and sported much better feature set, as well as more powerfull cpu, but the PS1 kept up - it often didn't look as pretty, but the framerates were there, and as a whole it offered a better gaming experience..
DC , PS2, GC and Xbox are all effectively the same generation, so they all compete together in certain ways. DC 2nd generation games compared quite well with PS2 1st gen games - ( I'm not just thinking about 1st games from individual companies, but 2nd generation in terms of support /optimisation tips from the console manufacturers ) PS2 games have compared pretty well with GC and Xbox games in general - of course the Xbox version's will have the graphical advantages, it is a much more feature filled gpu - but not without some limitations of it's own..
About the nicest things graphically from the Xbox / GC ( to a certain extent ) and the DC was the support for VGA/HDTV which allows visual quality that isn't often present through normal TV - This is great for the high end comparisions, but does skew things when most of the mass market isn't using it..
My favorite thing about consoles is how devs rise to the challenge - The HW doesnt change, but programmers tend to see effects and then implement them, raising the visual quality...
Console's are always based on compromises...
Look at PS1 - 66Mpixel fill rate for flat shaded poly's compared well with graphics cards at it's launch - and geometry throughput was pretty reasonable for it's time.
N64 came out later, and sported much better feature set, as well as more powerfull cpu, but the PS1 kept up - it often didn't look as pretty, but the framerates were there, and as a whole it offered a better gaming experience..
DC , PS2, GC and Xbox are all effectively the same generation, so they all compete together in certain ways. DC 2nd generation games compared quite well with PS2 1st gen games - ( I'm not just thinking about 1st games from individual companies, but 2nd generation in terms of support /optimisation tips from the console manufacturers ) PS2 games have compared pretty well with GC and Xbox games in general - of course the Xbox version's will have the graphical advantages, it is a much more feature filled gpu - but not without some limitations of it's own..
About the nicest things graphically from the Xbox / GC ( to a certain extent ) and the DC was the support for VGA/HDTV which allows visual quality that isn't often present through normal TV - This is great for the high end comparisions, but does skew things when most of the mass market isn't using it..
My favorite thing about consoles is how devs rise to the challenge - The HW doesnt change, but programmers tend to see effects and then implement them, raising the visual quality...
I agree with ERP and MrWibble.
GS lacks a lot of features, but it's a fact it's a blazing fast rasterizer.
And It could have been a lot better (quality and speed wise..) with small bug fixes, like better LOD calculation and a smarter rasterizing order that take in account memory pages breaks..
cio,
Marco
GS lacks a lot of features, but it's a fact it's a blazing fast rasterizer.
And It could have been a lot better (quality and speed wise..) with small bug fixes, like better LOD calculation and a smarter rasterizing order that take in account memory pages breaks..
cio,
Marco
- Status