ps2 and ngc technical details

[Farid]

The ELAN chip is also fixed function and only capable of 10 million polys/sec raw. Now why was the lighting and poly numbers downgraded in the PS2 version of VF4 since it can do 20+ million polys/sec in-game while NAOMI 2 can only do 10 million in-game???

1)Ram.
2)Sega-AM2 never mastered the PS2 hardware.
3)Other priorities for the ressources ( with the PS2 architecture you often have to chose between polys or textures).
4)The polycount downgrade of VF4 evo is not that obivious, at least to me, of course.

 

[PC-Engine]

The ELAN chip is also fixed function and only capable of 10 million polys/sec raw. Now why was the lighting and poly numbers downgraded in the PS2 version of VF4 since it can do 20+ million polys/sec in-game while NAOMI 2 can only do 10 million in-game???

1)Ram.
2)Sega-AM2 never mastered the PS2 hardware.
3)Other priorities for the ressources ( with the PS2 architecture you often have to chose between polys or textures).
4)The polycount downgrade of VF4 evo is not that obivious, at least to me, of course.

How do you explain the downgraded lighting then???

I'll tell you why, because on PS2 more lights = less polys

Lighting has nothing to do with the amount of RAM available to PS2.

Oh and regarding point number 2...umm doesn't hold water. It's like saying Namco hasn't mastered PS2 either...sorry weak argument.

 

[Fox5]

The ELAN chip is also fixed function and only capable of 10 million polys/sec raw. Now why was the lighting and poly numbers downgraded in the PS2 version of VF4 since it can do 20+ million polys/sec in-game while NAOMI 2 can only do 10 million in-game???

1)Ram.
2)Sega-AM2 never mastered the PS2 hardware.
3)Other priorities for the ressources ( with the PS2 architecture you often have to chose between polys or textures).
4)The polycount downgrade of VF4 evo is not that obivious, at least to me, of course.

How do you explain the downgraded lighting then???

I'll tell you why, because on PS2 more lights = less polys

Lighting has nothing to do with the amount of RAM available to PS2.

Oh and regarding point number 2...umm doesn't hold water. It's like saying Namco hasn't mastered PS2 either...sorry weak argument.

Number 2 could be true, wasn't VF4 am2's first(and perhaps only?) game on ps2? AM2 has typically focused more on the arcade versions of its games than the home ports, and virtua fighter 4 isn't the best looking game on ps2. Sega just wanted a quick buck on ps2...though Yu Suzuki takes a lot of pride in his skill, I'm sure he'd say vf4 pushed ps2 about as far as it could go....he did say that he could easily do DOA3 on ps2 too.

 

[PC-Engine]

VF4 Evo is one of the best fighters if not THE best on PS2. The only other fighters to come close is DOA2, SC2, and Tekken 4. You can obviously tell that Evo is probably as far as the PS2 can go in terms of fighting games.

Doesn't really matter if VF4 is the ONLY game to come out of AM2 if it's on par with games like Tekken which is made by Namco which has made dozens of games on PS2. If Tekken is the best Namco can do on PS2 with dozens of games under its belt then VF4 is obviously pushing PS2 very hard. Going by that logic PD's GT series isn't pushing the PS2 either nor are they familiar with programming it since they've only made one game series for it.

 

[Farid]

How do you explain the downgraded lighting then???

I'll tell you why, because on PS2 more lights = less polys

Elan can add 6 lights in hardware...
Did Virtua Fighter used 6 lights sources per scene, and if it did, could one argue that there's a huge difference between 6 lights sources and , let's say 2 lights sources on a fighter game that doesn't use special lighting system?

Lighting has nothing to do with the amount of RAM available to PS2.

I know PC-Engine
That wasn't my point, i gave you "reasons" for "why there's a "potential" polycount downgrade on a Naomi2 >PS2 port".
I wasn't talking especially about the lights.

Oh and regarding point number 2...umm doesn't hold water. It's like saying Namco hasn't mastered PS2 either...sorry weak argument.

As i said this wasn't meant to be a "point" but a "possible explanation", this said, it could make a very good point .
Sega-AM2 never did something incredible on the ps2, VF4 was their first game on the machine (as Fox5 explained it better btw).
Btw why the Namco example?

And bringing a Sega game in a PS2 technical discussion is like bringing a knife to a gunfight.
Only Visual Concept did something "good" on PS2, and maybe the last PS2 of Sega-AM2 (Macross).
I've played Nightshade (Kunoichi) from overworks lately, it a plain sub-Dreamcast game (there's only one cool effect, the per pixel shadowing when the airplaine pass above you...). The game is aliased as hell , has circles shadows, low polycount, bad textures...

 

[London Geezer]

EE can do 66 million polys/sec with no lights.
ELAN can do 10 million polys/sec with no lights.
EE can do probably 5 million polys/sec (MAX) with 6 complex lights.ELAN can do 10 million polys/sec with 6 complex lights.

That's a lot. I don't know about the ELAN numbers, but i'm fairly certain EE would have a fair amount of trouble pushing 5 million (textured?) polys with 6 lights. And what is a "complex" light anyway? Care to explain?

Anyway, about VF4. Ps2 has roughly about 3 times less RAM than Naomi2. The main reason why VF4 PS2 didn't look as good as the original was the downgrade in texture detail, which is only expectable, since Naomi2 also has texture compression available to it, on top of around 3 times (or is it 4?) the usable RAM. Naomi2 can also push more polygons than PS2 under certain conditions (more than 2 texture layers, more lights etc) so OF COURSE the PS2 version would look downgraded.

I do not think that there are AT ANY TIME 6 lights in VF4 on at the same time. The only stage i can think of is the "Top of the building" one, with 2 spotlights (the helicopters). Add one for ambient lighting (IF even that one was there) and u have 3 lights AT MOST at the same time.

But of course, and as usual, the above is open to correction, should anythign i have said be incorrect.

 

[Simon F]

EE can do 66 million polys/sec with no lights.
ELAN can do 10 million polys/sec with no lights.
EE can do probably 5 million polys/sec (MAX) with 6 complex lights.ELAN can do 10 million polys/sec with 6 complex lights.

That's a lot. I don't know about the ELAN numbers, but i'm fairly certain EE would have a fair amount of trouble pushing 5 million (textured?) polys with 6 lights. And what is a "complex" light anyway? Care to explain?

"Complex light" == A spot light (i.e. a point source light with a principal lighting direction) with distance attenuation and angular attenuation, with diffuse + specular lighting calculations and with choices of several different rules for behaviour with various normal orientations.

IIRC, it also includes the necessary calcs for local light interactions with bump mapping.

OH, BTW, I think you also get the ambient light thrown in for free

 

[London Geezer]

EE can do 66 million polys/sec with no lights.
ELAN can do 10 million polys/sec with no lights.
EE can do probably 5 million polys/sec (MAX) with 6 complex lights.ELAN can do 10 million polys/sec with 6 complex lights.

That's a lot. I don't know about the ELAN numbers, but i'm fairly certain EE would have a fair amount of trouble pushing 5 million (textured?) polys with 6 lights. And what is a "complex" light anyway? Care to explain?

"Complex light" == A spot light (i.e. a point source light with a principal lighting direction) with distance attenuation and angular attenuation, with diffuse + specular lighting calculations and with choices of several different rules for behaviour with various normal orientations.

OH, BTW, I think you also get the ambient light thrown in for free

Thanks, so there u go, u just gave me a confirmation that at NO time has VF4 more than 2 lights on at the same time.

And with your explanation, even more confirmation that the EE would have trouble handling millions of polys with 6 "complex" lights..

 

[PC-Engine]

Did Virtua Fighter used 6 lights sources per scene, and if it did, could one argue that there's a huge difference between 6 lights sources and , let's say 2 lights sources on a fighter game that doesn't use special lighting system?

It's irrelevent whether or not VF4 on NAOMI 2 used 6 lights. What's relevent is that the lighting in the PS2 version is downgraded from the NAOMI version regardless how many lights were used in the NAOMI 2 version. Since lighting has nothing to do with the amount of total RAM available, it means EE just cannot handle the lighting that was displayed in the NAOMI 2 version. NAOMI 2 version probably used per pixel lighting. Also the reason why I brought up ELAN was to show the Flipper can add lights just like ELAN without being affected like EE. Since the argument was about in-game polygon performance, the number of lights as well as the quality of the lights have a major affect on EE while not affecting Flipper as much.

Sega-AM2 never did something incredible on the ps2, VF4 was their first game on the machine (as Fox5 explained it better btw).
Btw why the Namco example?

Umm...it seems you're not getting the point. Namco was brought up because they've mastered programming PS2 yet their best fighting games are only equal to VF4 by AM2 who supposedly according to your theory has not yet mastered PS2 programming since they only have one game under their belt. If AM2 only has ONE game under thier belt but it equals the BEST on PS2 from seasoned PS2 programmers then what does that say about AM2 and VF4??? We're talking fighting games here are we not??? Or would you rather compare SH4 to VF4 to make your argument seem less flawed???

That's a lot. I don't know about the ELAN numbers, but i'm fairly certain EE would have a fair amount of trouble pushing 5 million (textured?) polys with 6 lights. And what is a "complex" light anyway? Care to explain?

Per pixel spot lights.

I do not think that there are AT ANY TIME 6 lights in VF4 on at the same time. The only stage i can think of is the "Top of the building" one, with 2 spotlights (the helicopters). Add one for ambient lighting (IF even that one was there) and u have 3 lights AT MOST at the same time.

Like I explained above, the lighting was downgraded in the PS2 version for a reason and it's not RAM regardless of how many lights were used. This was just to point out the flaw in Qroaches argument that Flipper is somehow weaker at pushing polys in-game than EE since it has fixed function TnL. It's boils down to how it's designed not whether or not it's fixed function as demostrated by ELAN. Both Flipper and ELAN was designed to have lighting added without a major performance to polygon performance unlike EE. Looking at RAW poly numbers which was what Qroach was using to gauge in-game poly numbers is flawed unless you're talking about games like REZ or Battle Zone. Even with said games they aren't even pushing the hardware limits with regards to polygons. Anyone here thinks REZ is pushing 66 million polys?

 

[Simon F]

Per pixel spot lights.

Elan worked per-vertex but you could use the bump mapping to do per-pixel lighting.

Just adding my tuppeny's worth to the earlier comments: One of the reasons Naomi2 has so much memory is that it ganged two CLX2s (i.e. DC|N1 rendering chips) in parallel. These were never designed to share texture/framebuffer memory with another chip and so that had to be duplicated. It's not a cost issue with a arcade system and, besides, it was only 100Mhz SDRAM which would have been dirt cheap.

 

[PC-Engine]

I'd also like to add that VF4 on N2 probably didn't even used texture compression like the other N1 games.

 

[London Geezer]

I'd also like to add that VF4 on N2 probably didn't even used texture compression like the other N1 games.

Even then u're missing the point. The main reason VF4 looked vastly better than the PS2 version was IQ and textures.
IQ - no arguing there, we've been through this countless times.
Textures - same, having around 3 times the memory is a big help. Also, the fact that you "believe" VF4 didn't use compressed textures doesn't mean much. And even then, it would be 3 times as many untextured textures than PS2.

As for the lighting, the only noticeable difference was that characters had much better looking specular lighting on them, whereas PS2 ones had a weird environment map or some wierd effect on them, not real specular maps.

 

[Simon F]

As for the lighting, the only noticeable difference was that characters had much better looking specular lighting on them, whereas PS2 ones had a weird environment map or some wierd effect on them, not real specular maps.

That means that on N2 they probably diced the models up into lots of polygons and let the vertex lighting take care of it. It's a lot easier than mucking about with specular texture maps.

 

[London Geezer]

As for the lighting, the only noticeable difference was that characters had much better looking specular lighting on them, whereas PS2 ones had a weird environment map or some wierd effect on them, not real specular maps.

That means that on N2 they probably diced the models up into lots of polygons and let the vertex lighting take care of it. It's a lot easier than mucking about with specular texture maps.

Why would they do that?! Doesn't N2 support per-pixel lighting? They could have just stuck a specmap on them and let the hardware take care of it...

I mean, i would understand if they did that on PS2, makes perfect sense and has been done before, but on N2?

 

[Farid]

It's irrelevent whether or not VF4 on NAOMI 2 used 6 lights. What's relevent is that the lighting in the PS2 version is downgraded from the NAOMI version regardless how many lights were used in the NAOMI 2 version. ... NAOMI 2 version probably used per pixel lighting.

awnser:

Elan worked per-vertex

So there's no complex PP lights, nor a huge number of them in Virtua Fighters 4, so how could it be "downgraded"?
And the point of the discussion was the polycount of the machines only.

Since the argument was about in-game polygon performance, the number of lights as well as the quality of the lights have a major affect on EE while not affecting Flipper as much.

Yes... I don't think somebody said the contrary.
But talking about "in-game", doesn't mean talking about all the possibilities ever imagined. Of course NGC could give more polygons when a certain numbers of lights are added to the scene... But is there a lot of games that uses more than 2 lights in any scene?
Qroach took the classical setup, One light, one texture... voila!

The classical setup that Nintendo themselves took on their spec sheet.

If AM2 only has ONE game under thier belt but it equals the BEST on PS2 from seasoned PS2 programmers then what does that say about AM2 and VF4??? We're talking fighting games here are we not??? Or would you rather compare SH4 to VF4 to make your argument seem less flawed???

We were talking about the polycount perf of the PS2 and GC, therefore any games actually could be compared to another as long as they use the same setup (cell shading vs cell shading, etc...).
Comparing SH4 to VF4 could be possible if SH4 ran at 60fps, also the latter could compare unfavorably because it has bigger environment, Stencil shadows and more AI to deal with.
So any game that use the same techniques and setup that was used on VF4 could be in someway compared to it. IMHO.
BTW VF4 is not the best looking fighter of the machine. Hey, even Tekken Tag (usa version of course) looked better if you ask me. (the gameplay is another story , VF4 owns all).
SC2's graphics are above VF4's too.

Even with said games they aren't even pushing the hardware limits with regards to polygons. Anyone here thinks REZ is pushing 66 million polys?

It's a DC game, and the 66Mpps fig is only if all ressources are allocated to polygons rendering.
Yeah, i know you were jocking.
But the example was badly chosen.

 

[Jov]

....he did say that he could easily do DOA3 on ps2 too.

Now that I'll like to see. I am not saying its not do-able on the PS2, I am just stating nothing comes to mind which is on par.

Come to think of it SC2 would be the closest to DOA3.

 

[Simon F]

As for the lighting, the only noticeable difference was that characters had much better looking specular lighting on them, whereas PS2 ones had a weird environment map or some wierd effect on them, not real specular maps.

That means that on N2 they probably diced the models up into lots of polygons and let the vertex lighting take care of it. It's a lot easier than mucking about with specular texture maps.

Why would they do that?! Doesn't N2 support per-pixel lighting? They could have just stuck a specmap on them and let the hardware take care of it...

I mean, i would understand if they did that on PS2, makes perfect sense and has been done before, but on N2?

It supports per-pixel lighting in the form of Dot product/Normal maps but that was a rare thing in those days. I think DC was the first piece of HW (at least consumer HW) that actually could do such a thing.

Anyway, you don't really need it all that often unless the surfaces you are modelling are rough. Vertex lighting generally is sufficient (and light maps are really only needed if the polys are huge compared to the rate of change of the lighting).

 

[PC-Engine]

So there's no complex PP lights, nor a huge number of them in Virtua Fighters 4, so how could it be "downgraded"?

Well to a blind man it wouldn't look downgraded

Qroach took the classical setup, One light, one texture... voila!

He did??? Regardless, the classical setup is only valid for old architectures where single texturing and a single infinite light source is common ie PS1 games. Going forward I don't think Xbox 2, N5, and PS3 will be doing single texturing and single lights.

The classical setup that Nintendo themselves took on their spec sheet.

And on their spec sheet it says 6-12 million polys with AI, physics, etc.

We were talking about the polycount perf of the PS2 and GC, therefore any games actually could be compared to another as long as they use the same setup (cell shading vs cell shading, etc...).
Comparing SH4 to VF4 could be possible if SH4 ran at 60fps, also the latter could compare unfavorably because it has bigger environment, Stencil shadows and more AI to deal with.
So any game that use the same techniques and setup that was used on VF4 could be in someway compared to it. IMHO.
BTW VF4 is not the best looking fighter of the machine. Hey, even Tekken Tag (usa version of course) looked better if you ask me. (the gameplay is another story , VF4 owns all).
SC2's graphics are above VF4's too.

Umm...ok...yeah...if you say so.

But the example was badly chosen.

Actually the example is perfect for illustrating the absurdness of using RAW polygon numbers as a means of comparison for 99% of the games of this generation.

 

[London Geezer]

So there's no complex PP lights, nor a huge number of them in Virtua Fighters 4, so how could it be "downgraded"?

Well to a blind man it wouldn't look downgraded

And you keep ignoring the simple fact that the main reason the game looked downgraded was because of the IQ and the Textures. And there are obvious reasons for that. Keep on your crusade.

 

[PC-Engine]

So there's no complex PP lights, nor a huge number of them in Virtua Fighters 4, so how could it be "downgraded"?

Well to a blind man it wouldn't look downgraded

And you keep ignoring the simple fact that the main reason the game looked downgraded was because of the IQ and the Textures. And there are obvious reasons for that. Keep on your crusade.

You obviously have no answer for the obvious downgrade in lighting. Have you seen the arcade version of VF4??? Textures have ZERO relevence on lighting unless you have an excuse there too

BTW why does there need to be a crusade? Can to explain? Must be another one of you conspiracy theories that don't hold water. It's not a religion man. It's not a knock at Kill Bill either

I see you've tried to convert Phil into thinking Matrix Reloaded and Revolutions were sh*ty and failed. Just get over it and stop being a drama queen around every discussion.