Gran Turismo 4 pushes PS2 to its limits

[Deepak]

"We have taken the PlayStation 2 to its limit and feel there is no more room for advancement of the series,"

And he's right, better psyhics = more processing power needed which the PS3 will provide.

Calm down guys.....it is pretty ambigous statement but the statement doesnt imply that they are using PS2 to 100%....btw I dont think anyone would complain if all PS2 games look like GT4 level....!!!!

 

[V3]

I forgot, I just saved it in text file, jumble with all other interview, for reading later on.

Here is full thing

The god-like creator of the Gran Turismo series gives a frank and unmissable interview on his online masterpiece-in-the-making, GT4

18:52 It still shocks us how amazing Gran Turismo 3 looks alongside more recent next-gen racers, its astonishing, near-photo realistic sheen as impressive as ever. But beneath the series' façade of shimmering beauty lies a game of unprecedented precision, with each GT title offering an uncanny recreation of what it's actually like to race some of the world's greatest automobiles.
But the fourth instalment of the globally respected series is set to enter uncharted territory, making the onerous - though much longed for - leap into online. But although by all accounts GT4 will be the most accurate videogame racer in history, but that does not automatically equate to a life-changing online experience.

Developer Polyphony Digital, headed by the masterful Kazunori Yamauchi, was on hand at E3 last week to present the game - in playable form - to the world for the very first time. And after a thrilling demonstration, we were lucky enough to grab some time Kazunori, alongside select members of the press, to delve deeper into PS2's biggest game.

Open and frank throughout, Kazunori discussed the heated car damage issue, explained the myriad problems facing his online dreams and his goals to build a mighty GT community. Without further ado, here's the full transcript:

Will there be any car damage in Gran Turismo 4? Surely the more realistic the game gets the more a lack of damage sticks out like a sore thumb, undermining this realism?

Kazunori: The simple answer to your question is: no. We are not planning to include any visual car damage. However, we are considering something, although we are not 100 percent sure... For instance, you saw the Grand Canyon course with the car running close to the ridges of the canyon?

If the driver messes up and misses the course, we might show the car jumping off the cliff, but we won't show it falling into pieces at the bottom. We're still considering that and we don't know how far we can go with it.

Instead of visual damage, we are considering some kind of a penalty system in which players will be penalised for trying to run corners along a wall, as they did in Gran Turismo 3 for instance. Also, pushing against opponents' cars, again seen in previous GT titles, will be penalised. We're currently trying to find the best solution.

Do you foresee a time when you will be able to include car damage in a GT game?

Kazunori: One of the reasons it would be difficult to include damage is that, because Gran Turismo is a real driving simulator, we would have to consider damage to be real. I have dome experiments, and a very high percentage - maybe 80 percent - of crashes in Gran Tursimo will cause the car to fall into pieces maybe two inches big.

Current specifications of the hardware will not allow us to represent this fully. If the specs are higher, then maybe, but then again, there are also other issues with manufacturers, licence issues and so forth, which keep us from doing this. There are many hurdles we must jump over.

It's not all manufacturers, but there is a trend generally speaking that manufacturers have become a little bit more lenient towards the direction we would like to head, in terms of allowing for damage on cars in games.

You do see that in other games - there are ways - it's just that with the range of cars we have, it's more difficult.

Will GT4 be the last PS2 game you make?

Kazunori: Maybe! [laughs]

Can you highlight what got you into cars and making car games in the first place?

Kazunori: There's no specific incident in my life that triggered the passion. I feel it's an instinct a lot of boys have - it's just natural for boys to like cars and that's where it stems from.

And are there any other racing games which have inspired you during your career?

Kazunori: One of them is Winning Run by Namco and also Microprose Grand Prix.

Can you tell us a little more about the online side of things?

Kazunori: With online, the reason why you haven't heard much is that there isn't much we can talk about, unfortunately. But my main area of interest is not actually the racing element of online, for instance, where people will compete against each other online. My main interest is to find a setting where players can come to share knowledge of cars - just car talk. Car talk is endless: you can talk about tyres, you can talk about oil, car parts, modification... "I did this! Was it good?" "Yeah, it's alright, but I tried this!"

This kind of communication is what I value the most and would like to implement somehow in the online segment.

What about a kind of online garage where you can buy, sell and trade motors?

Kazunori: Technically it's possible, but we're currently in the middle of examining all of our options.

Will you offer voice communication via the USB headset?

Kazunori: During racing? No, because Gran Turismo 4 uses the specs of the system to the max. Unless you would be happy settling for a 30fps frame rate? But I'm sure no-one wants that.

How about beforehand? When I played before, I picked an interesting looking car, but couldn't see what the other player did, who just picked the fastest of them all creating a real mismatch.

Kazunori: What you've brought to our attention is absolutely right, but that is only because we were very limited in time preparing for E3. For the finished game, there will be a system in place so that players know what each other has chosen, otherwise it wouldn't be fair.

I think most of you will have experienced the demo downstairs and noticed in some of the online competitions that you might see an opponent car in front of you moving a little bit awkwardly. This is a technology problem due to latency and cannot be eliminated 100 percent.

But the team is very, very smart and knows what it's doing. They know the problems, but they also know the solutions. It's just that due to the network, there are hurdles they cannot overcome. So you will always see that, and the further you are away from someone, the more "off" that will be. If you're closer it's not that bad.

There are so many areas we need to study if we're to say we're comfortable with where we're going. The biggest problem for me is that the images that players see will no longer be the standard of Gran Turismo, and that's the greatest concern for me we have in terms of taking the game online.

I don't think that problem will ever be solved. No matter how hard the team studies the programming side of the game, there will always be a network-related problem in which the quality of the gameplay will be decreased when seen in relation to the potential of the rest of the game.

Being the perfectionist that I am, this will always be a problem in my mind, and I will have to continue examining it.

Maybe if this were another game - no specific names - it might not be such a significant point. But because it's Gran Turismo and because of the quality we've had, once we take the online step, we're afraid it won't be the Gran Turismo that we all know. That's not the way we would like to move forward.

Considering its accuracy, have you ever thought about commercial uses for Gran Tursimo in the sense of teaching people to drive particular cars, or will it always be a game for entertainment purposes?

Kazunori: We have been approached by a number of car manufacturers and we are looking into some of those possibilities, but we haven't actually taken the step into taking action in terms of creating some kind of training simulator.

We're not as interested as the manufacturers at this point.

Johnny Minkley

 

[maskrider]

Thank you, V3.

 

[PC-Engine]

You just cited how PD said that GT3 only used 25%, but you found that a ludicrous statement to stand-by, and now PD says GT4 uses 100%, and all of a sudden you are dead certain they are right? By the first premise, why should you believe anything that PD says thereafter? ...because you WANT to believe it. As to whether it is really true or not, who freakin' knows?! Who freakin' cares? (Some care, obviously, because they fully intend to use it as "substantiated" ammo for some future console war debate. As if that is the sole thing they live for...)

It's not about ME. THEY know who they are.

I've always suspected that GT3 used way more than 25%. The statement from PD pretty much confirms this. Look at the neglible improvement from GT3 -> GT4. You think GT4 has improved 75%??

 

[Fafalada]

I've always suspected that GT3 used way more than 25%. The statement from PD pretty much confirms this. Look at the neglible improvement from GT3 -> GT4. You think GT4 has improved 75%??

So - PD told false info about GT3.
And proof for it is that : "PD told false info about GT3".

Well, how could anyone argue with that reasoning... 8)

 

[PC-Engine]

I've always suspected that GT3 used way more than 25%. The statement from PD pretty much confirms this. Look at the neglible improvement from GT3 -> GT4. You think GT4 has improved 75%??

So - PD told false info about GT3.
And proof for it is that : "PD told false info about GT3".

Well, how could anyone argue with that reasoning... 8)

How about this reasoning. YOU think GT4 is not maxing out PS2?

 

[London Geezer]

I've always suspected that GT3 used way more than 25%. The statement from PD pretty much confirms this. Look at the neglible improvement from GT3 -> GT4. You think GT4 has improved 75%??

So - PD told false info about GT3.
And proof for it is that : "PD told false info about GT3".

Well, how could anyone argue with that reasoning... 8)

How about this reasoning. YOU think GT4 is not maxing out PS2?

well, i tend to trust someone who actually developed a racing title for PS2, rather than u PCEngine.... the fact that i trust ANYONE more than i trust u is another thing j/k

 

[rabidrabbit]

Did PD (Kazunori Yamauchi) previously tell that GT3 only uses 25% of PS2?
If they did, and if they now say GT4 uses almost 100% of PS2, there certainly is a fault in logic somewhere.

I find it hard to believe that Mr. Yamauchi really is that much involved in programming these games, so his words should not be taken too literally.

I still think that he meant they are using 100% of what they can get out of the system, with current tools and resources.

 

[London Geezer]

my personal view on this is:

Mr whoever only talks with Sony's marketing heads behind him, and isnt actually particularly involved in PS2 programming.
what happens...
GT3 is released. PS2 needs an ego boost, and what better than The Man who developed the most gorgeous game ever saying that it only uses 25% of PS2's power.
then GT4 comes out (almost)... PS3 talk is taking over everyone's mind, the hype is already starting, PS3 needs an Ego boost, and what better than the man who developed the most gorgeous racing game ever saying that PS2 is done, and the next big step will be PS3...
all this 25%, 75%, 100% is just numbers to infuse a bit of hype into the fans... i dont think anyone could ever say what percentage of performance a game is pushing...
but maybe it's just me....

 

[rabidrabbit]

... i dont think anyone could ever say what percentage of performance a game is pushing...
but maybe it's just me....

Performance Analyser could...

 

[London Geezer]

... i dont think anyone could ever say what percentage of performance a game is pushing...
but maybe it's just me....

Performance Analyser could...

i don't think the PA (NOT to be confused with Prince Albert sorry couldnt resist) will give u a final percentage of what GT3 is pushing and stuff like that... i never used one of course, but i thought it only highlighted the flow of data in the system, so that the developer sees where the bottlenecks are and stuff like that...
FAF or Archie will know more about it

 

[Nick Laslett]

Kazunori Yamauchi made a very similar statement when GT2 was released about maxing out the power of the PSOne.

The Performance Analyser showed that GT2 was using about 80% of the PSOne's resources. I'm sure he has reached a similar figure with GT4.

All of the GT titles have been developed hand in hand with the Performance Analyser. They are using the PA constantly, so it would not be a surprise that they have reached a similar performance figure on the PS2.

When the game is released it will be much easier to assess the performance improvement. At present there is very little to go on.

If you divide GT into three components: Graphics, Physics, AI. With GT3 they achieved pretty good graphics, pretty good physics and poor AI.

The extra performance in GT4 could be prog scan graphics, excellent physics and good AI. This would account for the improvement over GT3.

 

[PC-Engine]

Anyone who thinks Axel Impact 2 can push the PS2 further than GT4, please raise their hand.

Falalada, I've already got your vote.

 

[maskrider]

my personal view on this is:

Mr whoever only talks with Sony's marketing heads behind him, and isnt actually particularly involved in PS2 programming.
what happens...
GT3 is released. PS2 needs an ego boost, and what better than The Man who developed the most gorgeous game ever saying that it only uses 25% of PS2's power.
then GT4 comes out (almost)... PS3 talk is taking over everyone's mind, the hype is already starting, PS3 needs an Ego boost, and what better than the man who developed the most gorgeous racing game ever saying that PS2 is done, and the next big step will be PS3...
all this 25%, 75%, 100% is just numbers to infuse a bit of hype into the fans... i dont think anyone could ever say what percentage of performance a game is pushing...
but maybe it's just me....

Whatever, I also remember somewhere about the 25% with GT3.

I think the truth is spong.com might have mis-quoted E3 Famitsu as the issue does not contain what they quoted from what I've seen.

But more like the other quote from V3 (if really come from the horse mouth).

 

[Fafalada]

How about this reasoning. YOU think GT4 is not maxing out PS2?

I think that the notion of talking % of system use to begin with doesn't make much sense. Especially since you could have same utilization % with vastly different visual results just by changing approach (for better or worse).
Regardless though, in current state GT4 isn't maxing anything out - no code this early is even close to being really optimized.

And besides, they'll need to work on IQ a lot before they'd convince me of anything of the sorts...

memory the limit, bezier cars and bezier road need for 90% of PS2"s power, but PD not use bezier, why?

IMO, there's a bunch of issues with HOS that you can't just skip over without planning to use it from the start. GT4 is built on top of GT3 assets, scrapping all of that would not only be costly but it would take extra time as well.
It's not a decision you could make lightly, especially when there's no gurantee what improvements you'll get as result, if any.

 

[marconelly!]

It is my understanding that they improved the graphics not as much as they improved the physics model and AI - that's basically what everyone with some playtime was talking about. Physics and handling are suposedly "best evah".

 

[maskrider]

The quote in the article from spong.com is definitely not in the latest E3 Weekly Famitsu that I have posted (with the help of a fellow nettizen who is also a Japanese translator).

 

[mark_f1man]

What does the performance analyser measure?

Does it just measure all of the bandwidth useage for the various buses in the PS2 vs. the theoretical maximums?

Can it analyse the processing units (core, VU0, VU1, GS etc) to detect "idle time", nops or waiting loops?

In any case could inefficient code (ie. resending geometry & textures unnecessarily) produce a high percentage number? If this is the case would comparing numbers between games be pointless and the PA mainly used for analysing your own code (which you would know well).

 

[Josiah]

everything stated has been valid, but does anyone really think we'll see any major visual improvent beyond what we're seeing in GT4? I doubt "in game" graphics are going to improve all that much more.

 

[archie4oz]

What does the performance analyser measure?

Basically it captures state info over time in various segments of the EE and the GS.

Does it just measure all of the bandwidth useage for the various buses in the PS2 vs. the theoretical maximums?

No it doesn't grab any data off of the buses.

Can it analyse the processing units (core, VU0, VU1, GS etc) to detect "idle time", nops or waiting loops?

Yeah, things like dual-issue on the CPU, DMA accesses, cache misses and bus interrupts, how busy the GIF paths are, VU usage, XGKICK stalls, polys entering DDA, pixels being output, DDA void, non-poly data being output, pixel stalls, GS idle, etc...

In any case could inefficient code (ie. resending geometry & textures unnecessarily) produce a high percentage number

Well there really isn't a 'percentage' it's basically a bunch of graphs showing states over time so while it is quite easy to have code generate some silly amount of usage, chances are it'd reflect elsewhere that you're not really doing anything useful and be regarded as a stall (e.g. feeding a ton of polys to the GS that are outputting zero area polys will reflect that your stalling in DDA)