Support NeoGAF

[Xaeroxcore666]

I'm pretty sure no PS2 game has 14-poly characters - but I figure you meant something else Anyway this goes back to what are we actually measuring though (and why the genre matters - if it does)
If you're just looking for skinned&animated characters set on a polygonal stage - then plenty of PS2 games exceed those numbers (eg. a certain horror game where the final boss exceeded 100k polys by itself, with volumetric shadows and all, or Jak games where realtime cutscene models were in 10-20k range with 4-5 characters on screen at once, alongside using actual AA on them as well), but best in class examples indeed, likely aren't fighters.
Although if you have 4 7k characters on screen does that make it somewhat less impressive than 2 14k? (I can only comment from GPU execution complexity perspective - but I know everyone has their own metrics for what they count and what not).

The light source thing is one of those 'headcanon' topics I referred to above. I've seen plenty of debates in 20+ years across every console gen about 'lighting this/that' and it virtually never gets qualified with numbers. For instance - on PS2 the 'low-end' number of influencing lights per object/triangle (I was gonna say average, but really it was incredibly rare to go below this) was 4. This is in stark contrast to everything that came before it - where indeed, you'd have games with single or even no dynamic light sources for what was being displayed. But no discussion starts from 'lighting on PS2 is 4x more complex on this game' - the common argument usually goes like 'oh they probably have better lighting >because< it's PS2, but textures, colors blahblah'.
Now - it's been decades since I've seen Naomi 2 specs. While I recall total throughput with lots of lights being mentioned, 'spec sheet' does not equal real-world use, as - basically every console ever released demonstrates. 'But but hw acceleration' doesn't really say anything without that context. If VF4 is the measuring stick - what are factual stats being compared - to discuss 'what'(if anything) was 'comparable' we need goal-post to begin with.


I agree DC has a large gap - but so did others, and if there's one console that stands out with having the widest gap from bottom to top - it was XBox.
The best they got was things like Riddick or HalfLife 2 port, or some Gen 6 ports in literal HD (SC2 in 720P, Dragon's Lair 1080p).
The worst was literally 'PS1/N64 pro/high-res' - yes even that console received shovelware from that category.
PS2 had incredible lows/highs as well - but I think XBox gap actually has it beat.

Hahahaha i meant 14K, my bad . But seriously is there at least any PS2 fighter which looks better than Naomi 2 VF 4? I mean, folks on Beyond 3D have shown that polycount wise there was nothing close per char on PS2 fighting elite (Tekkens, SC 3). Also it would be worth to see how deep are the cutbacks on Initial D and King of Route 66 Naomi 2 vs PS2 ports. Sadly there is nothing similar to Jak games on Naomi 2 to make the comparison, but if we can "extrapolate"the comparison of most complex titles on each plattform, disregarding genre of type of game, we can for example put Shenmue 2 vs any model 3 game and see what DC did better, what did worst and if the gap is that big as some folks in here states.

(eg. a certain horror game where the final boss exceeded 100k polys by itself, with volumetric shadows and all)

any clue?

And who said VF4 uses the power of the PS2? I mean, all Sega games on the PS2 are way way below what a first party could do, I doubt they use at least 60% of what the PS2 is capable of. Yes, no Naomi 2 game is ported 1:1 to the ps2 but from a hardware perspective the ps2 could outperform the arcade Naomi 2 if that is in the producer's interest. Naomi 2 is a 120k per frame poly machine at 60fps, the PS2 can do 150k at 60fps. Perhaps VF4 was the best a recovering Sega could do in terms of deadlines.

The NAOMI (New Arcade Operation Machine Idea) is also Japanese for beauty above all else.
CPU : 2 x Hitachi SH-4 32-bit RISC CPU (200 MHz 360 MIPS / 1.4 GFLOPS)
Graphic Engine : 2 x PowerVR 2 (PVR2DC-CLX2) GPU's - (under the fans)
Geometry Processor : Custom Videologic T+L chip "Elan" (100mhz) - (Under Heatsink)
Sound Engine : ARM7 Yamaha AICA 45 MHZ (with internal 32-bit RISC CPU, 64 channel ADPCM)
Main Memory : 32 MByte 100Mhz SDRAM
Graphic Memory : 32 MByte
Model Data Memory : 32MByte
Sound Memory : 8 MByte
Media : ROM Board / GD-Rom
Simultaneous Number of Colors : Approx. 16,770,000 (24bits)
Polygons : 10 Million polys/sec with 6 light sources
Rendering Speed : 2000 Mpixels/sec (unrealistic max, assumes overdraw of 10x which nothing uses)

Additional Features : Bump Mapping, Multiple Fog Modes, 8-bit Alpha Blending (256 levels of transparency), Mip Mapping (polygon-texture auto switch), Tri-Linear Filtering, Super Sampling for Full Scene Anti-Aliasing, Environment Mapping, and Specular Effect.
Compatibility : Fully backwards compatible with all Naomi and GD-Rom gam

''2x pvr2 overclocked'' 10M pps
''Polygons : 10 Million polys/sec with 6 light sources''
''2x 1,4gflops'' < 6,2Gflops


How many polygons does ps2 make with 6 light sources ?

So VF 4 did not use the power of PS2, but does Tekken 4 or 5, or SC3, or any PS2 elite fighter looks 40% better than VF 4 Evolution? It would be worth to verify this "How many polygons does ps2 make with 6 light sources ?" and of course to see under the hood if really PS2 is above Naomi 2 level contrary to popular belief, even stated by experts....

 

[Alexios]

The epic additional content in the PS2 releases of VF4/Evo is something that didn't happen often at all for their arcade ports and shows it was a pretty big (and with the two releases continued) effort for Sega with one of their still famous IP and not some throwaway release they didn't put resources to make it good in. But, maybe Sega sucked at making games and ports. I guess the Dreamcast never had a chance to be utilized properly/fully with such an awful first party turned shit third party, too bad Sony never made a game for it (just to prove it sucked and was in fact fully utilized anyway!).

Also let's ignore Tecmo, earlier proposed as easily able to make a late Model 3 game to top DOA2 on DC and Virtua Fighter 4 on Naomi 2 seemingly couldn't really top VF4 on the ever stronger PS2 (after the initial release fiasco with the properly updated and finished game). Never mind Namco. I guess now we can just say all fighting games simply suck instead because systems are only really utilized in other genres, never mind everyone going on about how fighting and racing games was the easy way to show off a machine and that's why DC has some good shit there and less elsewhere.

It's not that we're always arguing with whatever gotcha of the moment only to say essentially the opposite in the next sentence that suits our new argument and subject better, no, it's all logical, scientific even, trust the method people, that's how epic vidya gaem discussions are made. Polygons? Here's many! What, you got more? But polygons don't matter, let's talk effects! What, resolution? Less with fullscreen effects is more! What, you got better resolution and other stuff than this other thing that doesn't have screen effects either? Polygons matter the most, are you listening? What's that, between these two things you got more polygons and lights and effects and pretty much everything visible seems to be better on that side? Well, what about those 3 particles in that one spot, what a fail to not have anything like this awesome beast has! What a fail it's not my personal preference!!!

Gotta love this dude going all, I don't know anything about this hardware and its games but let's just start with the assumption you're basically all wrong in any of your praise for it and we can work from there by just propping up whatever it was you compared it with to quantify its qualities. Let's go!

How awesome we end up with a PS2 vs Naomi 2 string in a freaking Dreamcast thread just because I questioned someone praising Model 3 as better than Naomi 2 (while I was just talking DC Model 3 ports/comparisons) so now of course we must again prove PS2 is the king of everything forever!

Some of you can always go and make a PS2 thread so you can pour all this sweet, precious, unstoppable love that ends up leaking out every chance it gets somewhere it belongs, er, somewhere every lover can see it instead of miss it because it's in the topic about a trash system they don't care for.

 

[dano1]

Power had nothing to do with the Dreamcast's lifespan. They ran out of money, period.

PS2 killed it

 

[Fafalada]

How many polygons does ps2 make with 6 light sources ?

It's been awhile since I've looked this up - but more importantly - like with all programmable shader solutions, the answer is 'it depends' - on type and composition of lights.
IIRC - single VU throughput worked out roughly as follows.
Precomputed/no lights -> 40M pp/s
4 directional lights -> 18M pp/s
8 -> 12Mpp/s.
6 point(or spot)-lights ~6Mpp/s

I could write an essay here on (lack of)practicality of using 6 influencing point-lights 'per polygon' on entire scene basis - but suffice to say it woud require some pretty exotic setup that games of that era just didn't do(it really only starts to make sense if you're doing virtual point-lights to simulate GI, and vertex density has to go above gen-6 or switch to pixel-light sources entirely, to really benefit going from 3 influencing lights to 6). So usually numbers would be some combination of there-of. Flipside is that very few games of that era were actually vertex-limited to begin with - average 3rd party game on PS2/GC/XBox ran around 3mpp/s - and it wasn't because geometry processing was running slow.
Now - it 'was' technically possible to double those numbers (two VUs doing geometry) but this was quite uncommon setup (and largely unnecessary because - rasterization throughput of 10-20M rendered polygons was on the very high end for that generation of consoles - even with XBox GPU).

Hahahaha i meant 14K, my bad . But seriously is there at least any PS2 fighter which looks better than Naomi 2 VF 4?

I play so few fighting games I'd never have known tbh. The moments I was impressed by fighting games were SC1, and later the TTT reveal at TGS (which didn't quite make it into final game). The rest all kind of melded together for me into nothing standing out particularly during PS2 era, and starting with 360 onwards, the genre as a whole seems like it lost funding confidence to compete with the big 'AAA' releases, no matter the hardware/developer behind it, so nothing stands out anymore.

I mean, folks on Beyond 3D have shown that polycount wise there was nothing close per char on PS2 fighting elite (Tekkens, SC 3).

Sure - but that's also a bit arbitrary metrics. As I asked above - having multiple chars add up to the same polys (or similar) doesn't work out GPU any less. But not sure how that works out across different fighters either - like I said - no expert there. And that's also a metric for performance, not 'better/worse' looking - which is the original point. What are we even comparing.

Also it would be worth to see how deep are the cutbacks on Initial D and King of Route 66 Naomi 2 vs PS2 ports.

Those would be interesting to analyse - I imagine ram is as always - the biggest differentiator if other Arcade ports to DC/PS2 etc. are anything to go by - but indeed.

Sadly there is nothing similar to Jak games on Naomi 2 to make the comparison, but if we can "extrapolate"the comparison of most complex titles on each plattform, disregarding genre of type of game, we can for example put Shenmue 2 vs any model 3 game and see what DC did better, what did worst and if the gap is that big as some folks in here states.

I think if it's purely to measure hw-throughput (not some arbitrary art-style powered visual appeal) that can work. It's just important to keep in mind 'what' we're measuring. When I complained about CT compared to simulation heavy open-world games, that wasn't in reference to 'polygons on screen' - just everything else those games do (which for open-world titles back then was the real bottleneck - pushing polys was mostly on the backburner).

 

[Geometric-Crusher]

So VF 4 did not use the power of PS2, but does Tekken 4 or 5, or SC3, or any PS2 elite fighter looks 40% better than VF 4 Evolution?

unfortunately no fighting game pushes the ps2 even close to its limits, I mean until the beginning of the sixth gen with Tekken Tag, fighting games received prominence and big budget but as it became cheap to just make a game that works, they took their foot off the accelerator. for example DOA3 is running on a very powerful system being described as an 8.4M game according to urban legend I don't need to say how much superior to the arcade version of VF4 it is. To make matters worse, some teams at Sega have always made games PS2 grudgingly.

According to the user who extracts the models from DC/Naomi VF4, in overview, 120k poly per frame (the limit) 60fps 7.2M is the limit in real world games on Naomi 2 (all games) I'm not sure but the data I have says that VF4 operates with 5 light sources.
In the fight when it says ''round 1'' Naomi 2 pushes ~ 80k so it's 5M in-game. The PS2 version does in game 2,5 to 3,6M 3 light sources.

Games like Tekken 5 and SC3 are also far from making good use of the ps2 hardware, I mean Namco was focused on making the Tekken franchise evolve but then retreated again finding the sweet spot in Tekken 5, SC3 is a glorified SC.

The PS2 lacked that fighting game with characters with 15,000 polygons, fake bump mapping, lots of particles like in Burnout, etc.
an intermediate game between doa3 (xbox) and sc3, even low poly with fake shaders etc.

PS2 (contrary to popular belief) was the console that had the least correctly used hardware in history, only Shadow of the Colossus, MGS3 and Hitman made decent use of the console, all those 60fps games you know could have been even better at 25~30fps all them.

 

[Geometric-Crusher]

It's been awhile since I've looked this up - but more importantly - like with all programmable shader solutions, the answer is 'it depends' - on type and composition of lights.
IIRC - single VU throughput worked out roughly as follows.
Precomputed/no lights -> 40M pp/s
4 directional lights -> 18M pp/s
8 -> 12Mpp/s.
6 point(or spot)-lights ~6Mpp/s

I could write an essay here on (lack of)practicality of using 6 influencing point-lights 'per polygon' on entire scene basis - but suffice to say it woud require some pretty exotic setup that games of that era just didn't do(it really only starts to make sense if you're doing virtual point-lights to simulate GI, and vertex density has to go above gen-6 or switch to pixel-light sources entirely, to really benefit going from 3 influencing lights to 6). So usually numbers would be some combination of there-of. Flipside is that very few games of that era were actually vertex-limited to begin with - average 3rd party game on PS2/GC/XBox ran around 3mpp/s - and it wasn't because geometry processing was running slow.
Now - it 'was' technically possible to double those numbers (two VUs doing geometry) but this was quite uncommon setup (and largely unnecessary because - rasterization throughput of 10-20M rendered polygons was on the very high end for that generation of consoles - even with XBox GPU).

Thank you very much for the technical opinion. In my layman's opinion, the Naomi 2 would be more competitive hardware against the ps2 despite the ps2 having the edge. The King of Route 66 runs on Naomi 2 with 120k per frame, ps2 version has the same geometry the same, with only 26,000 tris per frame, it also has better lighting.
try to remember the 100k boss from the horror game.