PS2 vs Dreamcast Graphics

[sheath]

Since my decades old archived web pages keep being brought up I thought my view was clear. The PS2 "should" technically be capable of same-ish quality texture maps as the Dreamcast PowerVR 2. It just isn't in most games, possibly due to the move in the industry to "shaders" and the PS2 approximating them by layering relatively non-memory intensive "polygons" on top of polygons using the VUs.

I do my best to not resort to opines, so I have documented what I have found here:
http://www.gamepilgrimage.com/conten...000-benchmarks

Contrary to the site owner's opinions, mine has changed over the years. I just don't see with high def emulator glasses like the majority do. Also, it's an official statement from Sony themselves (and documented above):
http://www.gamepilgrimage.com/sites/...rHaveWeGot.pdf

See Pages 18-20, and the general recommendation for palatalized textures to reduce stalls. That is, from what I've seen, 4-bit, 16-color, palatized textures in virtually all PS2 games. It's no different in my mind than Saturn devs needing to use the VDP2 as much as possible to raise the fillrate.

I should also probably point out, again, that my original posts were against the PS2 being "10X" as powerful, or 66 times as powerful in Sony's specs. Context is almost everything.

[SegaAMD]

It is interesting that the palletizing technique is nothing more than a texture compression technique. The PS2 does not operate at 4/8 bit as if that were a negative the machine, was designed to handle this hardware compression - and people say that the ps2 does not compress. yes it compresses in this and other ways, including compressing the polygons as we will see below-.

I believe there is an understanding of what things essentially mean, what paletized means for example people read: ''4 bit paletized '' what is this ? A big nothing there are other factors, clut, filtering, application of light on the object and the like that modify the entire final look.


if 4/8 bit means ps2 has worse textures this is restricted only to theoretical specifications because in practice the games are there. The Dreamcast stores its textures in vram so that the more polygons the less size for textures, making use of Vq compression.
PS2 doesn't work like that, polygons don't occupy much size in the vram they are sent by the bus at high speed opening along with the texture, it's something analogous to ''polygon compression''. So if the Dreamcast stores 2MB polygons Vram on the PS2 this is decimated. Half Life is the best example, we know that the game is not fully finished but according to what I read, it was only missing the optimization of loadings and other minor details, nothing involving the textures, so we have: Low definition textures and worse colors on the Sega console. Why if console has supposed advantage in this field ?

yes, We didn't see any other comparisons because unfortunately the Dreamcast down too fast but I'm 100% sure that it wouldn't be able to replicate the textures of ps2 games seen in Tekken Tag, MGS2 (yes they are gray but they are high definition and purity not even the xbox replicated for not dealing with textures paletized) Red Faction 1, Time crisis 2 etc basically almost no 2001 game just look NBA 2Nights or NFL 2K2.

But the most interesting thing is not even that, but the PS2 showing more polygons per frame and more textures per frame for example Soldier of Fortune the textures are sharper on the Dreamcast but the Dreamcast has less textures and polygons per frame in this game as well as in Stunt GP, although it also has a higher sharpness. 4x4 evolution identical but slightly better in DC as progressive scan was not available before Tekken 4 in 2002, again less polygons in the frame,in theory this should not happen. but in practice as we see it occurs.

[SegaAMD]

I think this analysis PS2/DC needs to be done in the

1- soldier of fortune with higher resolution textures (DC) but less textures and polygons in the frame. Maybe it would be possible for the PS2 to do 1:1 reducing the amount of textures and polygons in the frame to the same as the Dreamcast (like Unreal T) and then increasing the resolution of the textures but this is just a guess it would take a dev to explain.
2- HL/ET where we have less textures, less resolution in textures and less polygons per frame (DC).

[sheath]

WinCE, versus various versions of Sega SDKs, versus pre/post Performance Analyzer PS2 games, and the specs of their development systems and the funding and size of the development teams are also absolutely factors on the performance of the final products. An overnight WinCE port on the Dreamcast by often times a single programmer is obviously not going to equal an entire team source porting the game to another platform. The same game being ported today with a modern computer, by a single dev, would do better if the tools and hardware knowledge aren't lost to time.

I don't even consider the PS2's palatalized color blob textures a major issue, they just stand out to me as one of the ways it mitigates that generation's limitations. Far, far more obvious to me is the severe lack of image quality in the A/V output, and the 224 lines being artificially interpolated to make a full screen image. It screws with most video capture equipment and only looks like a solid image on some televisions from the time.

Then, back in time, we had by far most internet and game store commentary claiming the PS2 was actually displaying, observably, measurably, 66 or 33, or 10, or some random number more polygons per second. Per frame wasn't even a consideration then, layers for effects weren't a consideration then, it was just a popular myth.

[gamevet]

Any game made with Windows CE on the DC was bound to suck. Why they used it for SEGA Rally 2 is mind numbing.

[sheath]

I thought I read somewhere that SR2 used WinCE specifically for it's early online functionality. I remember other companies turning to Microsoft at the time, claiming they were the industry leader in online connectivity. This actually reminds me of another developer quote on these forums. Something to the effect of gameplay no longer being play tested during development, but what would happen if the memory card, or controller, etc was unplugged took priority.

It shows to me in most of the games. At least SR2 plays well. And I think it is actually pushing WinCE to its limits. When I use the codes to raise the framerate I tend to prefer the original textures and details instead.

[gamevet]

I don’t think that SR2 had any online play. It was a very early title for the console.

[sheath]

If I recall correctly, which I very well may not be, it was just the Japanese version and it was a limited "beta" kind of thing.

[Yharnamresident]

So I only feel a bit like garbage, not complete garbage. I should be able to get some work done.

[Yharnamresident]

Any game made with Windows CE on the DC was bound to suck. Why they used it for SEGA Rally 2 is mind numbing.

Yes I don't bother looking at WinCE games whatsoever. If you see that WinCE logo it means crappy graphics.

[TrekkiesUnite118]

Since my decades old archived web pages keep being brought up I thought my view was clear.

They keep coming up because you keep coming in here repeating the same arguments on those pages.

Contrary to the site owner's opinions, mine has changed over the years. I just don't see with high def emulator glasses like the majority do. Also, it's an official statement from Sony themselves (and documented above):
http://www.gamepilgrimage.com/sites/...rHaveWeGot.pdf

See Pages 18-20, and the general recommendation for palatalized textures to reduce stalls. That is, from what I've seen, 4-bit, 16-color, palatized textures in virtually all PS2 games. It's no different in my mind than Saturn devs needing to use the VDP2 as much as possible to raise the fillrate.

Yes, but it also points out that a lot of games are around 2-5 Million Polygons Per second mark in 2003, and the highest they saw was 7.5 Million. Which again disproves your claim that most of the generation was hovering around 1 Million Polygons Per Second. Which the whole reason the previous argument broke out was because you came in here trying to claim the info posted vindicated your arguments from years ago, and those arguments happened to be that most games of that generation were hovering around 1 Million Polygons per second.

I should also probably point out, again, that my original posts were against the PS2 being "10X" as powerful, or 66 times as powerful in Sony's specs. Context is almost everything.

Which when you look at the specs and how those figures were calculated, they're not really wrong. They're simply taking fillrate in various different scenarios and dividing it by the amount of pixels a relatively small polygon takes up. For example I think one you pointed out was something like 66 Million Untextured Polygons per second. That figure isn't really wrong or inaccurate going by just pure fillrate. With Untextured pixels PS2 has a fill rate of about 2.4 Gigapixels. So if you're drawing a bunch of small polygons that only take up about 36 pixels, you can in theory get 66 Million Polygons per second. Does it really make sense for a game stand point? No not really, but it's a spec that if you understand what it means gives you a metric to compare against.

And Sony isn't the only ones that did this. Sega, Nintendo, Microsoft, etc. they all did this. Sega listed that VDP1 in Saturn could draw 500,000 Flat Shaded (untextured) polygons per second. Again this isn't really impossible given VDP1's fillrate, it's just silly is all. With a fillrate of 24,000,000 Pixels per second with untextured pixels, you could in theory draw 500,000 8x6 untextured quads with that fillrate. It's not really practical for a game stand point, but it's fine for tech flexing.

[SegaAMD]

PS2 really can do that number of theoretical polygons but that's certainly ''in the void''. Deep down what Sony did was show numbers and it showed the biggest possible did it masterfully.

However, we know that games are not about polygons, they are about textures, lighting, resolution and effects. That's why two games are emblematic even early games 7 Blades and Dynasty Warriors 2, we can imagine that these games move around 2-3 million pps and that's what's cool, because it's never been seen before in video games a crowd of people sensation, PS2 did it masterfully Matrix The Path of Neo in the scene of hundreds of agent Smith, did it in State of Emergency, Demon Chaos 65,535 enemies can be on-screen at once . the particles in its games can reach up to 18 million particles while in the dreamcast it is rare to find games with sparks it's all integrated into what Sony described as a ''polygon'' during marketing like every blade of grass.

It doesn't matter if it's millions of polygons on screen but the sense its amazing see hundreds of characters in a game, even if can be made with less polygons than people can feel, the impact is so big that whoever sees it will think there are millions and millions. In the end it was just hyperbole. IMO I think that's what Sony wanted to convey in their marketing.

[SegaAMD]

So I only feel a bit like garbage, not complete garbage. I should be able to get some work done.

I'm waiting for the measurements on the games you promised, SF, SC, Test Drive 6 etc put some images of games to break this sequence of texts.

[sheath]

They keep coming up because you keep coming in here repeating the same arguments on those pages.



Yes, but it also points out that a lot of games are around 2-5 Million Polygons Per second mark in 2003, and the highest they saw was 7.5 Million. Which again disproves your claim that most of the generation was hovering around 1 Million Polygons Per Second. Which the whole reason the previous argument broke out was because you came in here trying to claim the info posted vindicated your arguments from years ago, and those arguments happened to be that most games of that generation were hovering around 1 Million Polygons per second.



Which when you look at the specs and how those figures were calculated, they're not really wrong. They're simply taking fillrate in various different scenarios and dividing it by the amount of pixels a relatively small polygon takes up. For example I think one you pointed out was something like 66 Million Untextured Polygons per second. That figure isn't really wrong or inaccurate going by just pure fillrate. With Untextured pixels PS2 has a fill rate of about 2.4 Gigapixels. So if you're drawing a bunch of small polygons that only take up about 36 pixels, you can in theory get 66 Million Polygons per second. Does it really make sense for a game stand point? No not really, but it's a spec that if you understand what it means gives you a metric to compare against.

And Sony isn't the only ones that did this. Sega, Nintendo, Microsoft, etc. they all did this. Sega listed that VDP1 in Saturn could draw 500,000 Flat Shaded (untextured) polygons per second. Again this isn't really impossible given VDP1's fillrate, it's just silly is all. With a fillrate of 24,000,000 Pixels per second with untextured pixels, you could in theory draw 500,000 8x6 untextured quads with that fillrate. It's not really practical for a game stand point, but it's fine for tech flexing.

.
My claim or yours trekkies? You have been following Agostinho Barone's "full on retard" protocol since he decided he couldn't find a legitimate way to discredit what I wrote on Gamepilgrimage. Can't you see that? You probably can't, because you can't stop and admit a single mistake in your posts, and you cannot, ever, cite anything other than yourself. You're all over the place with your posts, but all you ever cite is yourself. This is your biggest clue.

So let's look at my "bonkers" artical from 2001, edited in 2005 one more time shall we? It includes....
http://www.gamepilgrimage.com/DCPScompare.htm
"As late as 2005, Capcom was forced to drop the polygon counts of the PS2 port of Resident Evil 4 from the Gamecube to between 900k to 1.5 million polygons per second, at 30 frames per second, while the Gamecube original ran at nearly twice this polygon count. This means that the entire generation of game consoles was relatively close to the Dreamcast's specs. This is in opposition to the idea, propagated by the media, that the PS2, Xbox and Gamecube's games were running 2-4 times the polygon counts of the Dreamcast's max specs. "

Lots of caveats there, but you can't be bothered to admit it. You and your overlord posse crassly dismissed the Usenet source I used, which stated quite clearly the PS2 would nominally display two times the polygons on screen. It's in the same article, from before I even got my History degree:
""The PS2 handles everything needed to render 3D graphics, and that's about it. The EE sends polygon data (three or four 2D coordinates, and pointers to texture, bump, light and dark data), and the GS pumps the data to the screen. It offers little in effects, and places the burden of rendering on the CPU. (...) Since the bottle-neck is with the processor, I'll take a moment here to discuss this further. The EE can render 36 million polygons with some effects on (though it still doesn't do many things the PowerVR can). This is full bore, which means the CPU is doing nothing but dumping polygons. With game physics and AI bundled into the mix, expect polygon counts to drop. More complex games will hurt more in the graphics department. Of course, the polygon count, even in the potential worst case (all bezier surfaces, 50% CPU spent on AI and physics) is still faster than the PowerVR. In fact, about twice as fast. What does this mean. Well, look at the current DC game models. For every straight edge you can see, subdivide it once (so that each edge is broken into two), and that's the detail improvement you'll see. Pretty substantial? Of course, as game developers make better use of the CPU(s), I'd expect polygon performance to increase."

You jumped up and down on this very forum in the Gen threads that this was wrong. Now you're waffling on that topic. The original point, stated again here for all to read, was that the PS2 was not, and never was going to be, 10X, or 33X, or 66X as "powerful" as the Dreamcast. Good luck finding a benchmark that objectively maps 224 interpolated lines full of sparklers and polygons on top of polygons.

I have stated since before the Internet turned into an idiotic tabloid, that the PS2 should be able to double the polygon counts of the Dreamcast with similar texture mapping capabilities.

[Yharnamresident]

So I looked at what might be Sega's version of Street Fighter II. This game was a insane commercial and critical success.

Remember this is on Naomi 2.



Yes these graphics are insane for 2001. The guy has the polygons of some entire DC/PS2 early games.



I don't know why I'm fighting the same guy but thats how it was.



Yes it has polycounts of a GC/Xbox game.



This Han lady was pretty tough but I finally managed to beat her.