Support NeoGAF

[SmokyDave]

Say you capture 60 seconds of 60fps footage (60x60), that equals 3600 frames of footage. If 70% = torn, that means 2520 frames will exhibit tearing.

Then those GT5 numbers are insane.

My guess is that the tearing is often located in the top part of the screen where the bright sky usually is, so it will be less noticeable. Sometimes games even hide the tearing in the overscan area that a lot of shitty TVs cut off, such as Prince of Persia 2008 or bizarrely the cutscenes in Vanquish (360) - but this trick is useless for people with a decent screen which doesn't have overscan problems.

It has to be well hidden. Screen tearing is one of the few things that irritates me and I'm not having a problem with GT5. Sure, I see it here & there (especially piling into a corner with 4 cars ahead, all tyres smoking), but nothing like Moto GP '08 (where the track split through every corner, every time) which those percentages suggest ought to be the case.

Could it be that the amount of tearing is nigh-on insignificant in some frames? As in, a tear that results in a shift of only a few pixels?

 

[Rad-]

http://blog.livedoor.jp/ps360/archives/51645053.html

new analysis of framerate cockpit vs bumper cam

Special Stage Route 7
通常：Avg:43.400fps Min-Max:35.5-53.5fps Tear:70.222%
車内：Avg:37.483fps Min-Max:31.0-45.5fps Tear:60.500%

Chamonix
通常:Avg:52.733fps Min-Max:43.5-58.0fps Tear:84.306%
車内:Avg:47.900fps Min-Max:40.0-54.5fps Tear:78.250%

London
通常:Avg:57.110fps Min-Max:48.5-60.0fps Tear:53.002%
車内:Avg:48.400fps Min-Max:39.5-57.0fps Tear:78.385%

Those tearing percentages can't be right. That would be ridiculous.

 

[Mastperf]

I do hope that PD will be able to patch in MLAA to help improve performance. Time will tell. Since GT6 is historically a ways away, who knows.

MLAA isn't free., The SPE's are already shitting blood here so they're not gonna have anything to spare. The aliasing would be be worse with MLAA than with 4x MSAA due to the subpixel aliasing in the game.
Polyphony isn't ignorant. They didn't forget triple-buffering or MLAA.

 

[Diablohead]

That's the sad part, a good number of cars sounded good but you've got to admit some of them are recycled from GT5P. But this.... hollly shiiit, The rotary engine and the exhaust notes. yuuumm

skip to 5:00
http://www.youtube.com/watch?v=OSz2h1b71v0&hd=1

5:41 to 6:10

 

[sajj316]

Those tearing percentages can't be right. That would be ridiculous.

Yup, been saying that since the first round from ps360frame came out. These CAN'T be right!

 

[Brashnir]

Jeeeeeeeeeeeeeeeeeesus. Triple-buffering, PD, triple-buffering. I know a torn frame is still a dropped frame, but at least make it less noticeable. They are already targeting 60fps so it's not like input lag will increase to an unacceptable level.

Triple buffering isn't free. It sometimes seems that way on PC where you often have 2-10x as much video RAM as you need, but on consoles, developers usually have better things to spend RAM on than 2 extra frame buffers.

 

[I NEED SCISSORS]

It has to be well hidden. Screen tearing is one of the few things that irritates me and I'm not having a problem with GT5. Sure, I see it here & there (especially piling into a corner with 4 cars ahead, all tyres smoking), but nothing like Moto GP '08 (where the track split through every corner, every time) which those percentages suggest ought to be the case.

Could it be that the amount of tearing is nigh-on insignificant in some frames? As in, a tear that results in a shift of only a few pixels?

Well a torn frame is a torn frame - no two ways around that, it either tears or it doesn't.

But it's whether you notice it or not which becomes the issue. In your Moto GP example, if the tearing was on the track, ie the part where you attention is going to be focused on all the time, then yeah it is going to be noticeable. In GT5, it tends to be towards the top of the screen where your attention isn't focused as much, so you won't notice it as much (and like I say, the bright sky helps mask it a bit).

Yet another point to consider is that in Moto GP the camera is probably moving a lot more than it does in GT5 - a moving camera will always make tearing more apparent. In GT5, the camera is very tightly locked to the car in all views but the cockpit - there is little movement except if you collide with someone. IMO, this could explain why tearing in GT5 is more noticeable in the cockpit where you head is shaking around.

It's a similar situation to Vanquish on 360.

 

[Domstercool]

Does anyone know how it the tearing is on 1080i? Because I hardly see any at all :/

 

[Brashnir]

Yet another point to consider is that in Moto GP the camera is probably moving a lot more than it does in GT5 - a moving camera will always make tearing more apparent. In GT5, the camera is very tightly locked to the car in all views but the cockpit - there is little movement except if you collide with someone. IMO, this could explain why tearing in GT5 is more noticeable in the cockpit where you head is shaking around.

This is true. Tearing is mostly noticable when something is moving across the screen up or down or (most prominently) side to side. In a racing game where you're generally moving forward into the screen, the most noticable place to spot tearing would be toward the bottom of the screen, where the road is showing the most apparent movement. This is mitigated by the fact that that area of the screen tends to often be a fairly homogenous road texture, plus your eyes are rarely focused there as you're looking further down the track.

A motorcycle game generally has a lot more camera rotation and tilting, making page tears more visible.

 

[See You Next Wednesday]

did you even click the link? there is nothing to read at that link.

And weren't gamereactor also the site that made the FM2/FM3 image comparisons that some people happily used to point that FM3 didn't look better than FM2?

Laguna Seca looks way better in Forza:

 

[-viper-]

wow - the tyres really do make a HUGE impact on the handling of the car.

My R33 GT-R handled like shit with Sport Hard tyres. Bought some Racing Softs and it now handles like a dream.

At least now I know the Lambo LP-640 and the Audi R8 will surely handle better with better tyres.

edit: uh.. wrong threa.d

 

[jett]

Those tearing numbers cannot be accurate, it's retardedly high.

 

[cgcg]

Well a torn frame is a torn frame - no two ways around that, it either tears or it doesn't.

But it's whether you notice it or not which becomes the issue. In your Moto GP example, if the tearing was on the track, ie the part where you attention is going to be focused on all the time, then yeah it is going to be noticeable. In GT5, it tends to be towards the top of the screen where your attention isn't focused as much, so you won't notice it as much (and like I say, the bright sky helps mask it a bit).

Yet another point to consider is that in Moto GP the camera is probably moving a lot more than it does in GT5 - a moving camera will always make tearing more apparent. In GT5, the camera is very tightly locked to the car in all views but the cockpit - there is little movement except if you collide with someone. IMO, this could explain why tearing in GT5 is more noticeable in the cockpit where you head is shaking around.

It's a similar situation to Vanquish on 360.

Actually no. Don't know what technique PD is using here but often the tears don't travel across the whole screen. Example, there might be tears on the windshield wiper but it doesn't affect the rear view mirror which is on the same plane as the screentear of the windshield wiper. A lot of the times you see screen tear *spots* in GT5. You can also see these *spots* easily if you can behind a car that's kicking up a lot of dust/rain.

 

[See You Next Wednesday]

Agreed, rewind should be kept out. When people say "Why should I lose a race just because I fucked up one single corner?" they're just not good enough, otherwise they wouldn't fuck up.

It's the worst thing to happen to racers in the last 5 years.

That's as stupid as people who say Traction Control Management in Sim racers is ruining them.

 

[I NEED SCISSORS]

Actually no. Don't know what technique PD is using here but often the tears don't travel across the whole screen. Example, there might be tears on the windshield wiper but it doesn't affect the rear view mirror which is on the same plane as the screentear of the windshield wiper. A lot of the times you see screen tear *spots* in GT5.

That's interesting. I never really noticed that.

 

[sajj316]

Triple buffering isn't free. It sometimes seems that way on PC where you often have 2-10x as much video RAM as you need, but on consoles, developers usually have better things to spend RAM on than 2 extra frame buffers.

Don't disagree. Again, it is a "hope". I attribute frame dips and screen tears to two things in all games.

1. Bad code -> as seen in bad port jobs
2. Taking on more than the hardware can handle (assuming good code)

I don't think its 1. PD have some of the best code monkeys in the business. The stress this game puts on the physical hardware has to be taken into consideration on some of the design decisions. I applaud the high benchmarks that PD is trying to deliver. 16 cars at once! Its those design decisions that could have easily freed up CPU cycles and RAM for other on screen performance functions like tripe buffering and MLAA.

EDIT: Easy for me to say in hindsight so take my comments as wishful thinking.

 

[skyfinch]

http://blog.livedoor.jp/ps360/archives/51645053.html

new analysis of framerate cockpit vs bumper cam

Special Stage Route 7
通常：Avg:43.400fps Min-Max:35.5-53.5fps Tear:70.222%
車内：Avg:37.483fps Min-Max:31.0-45.5fps Tear:60.500%

Chamonix
通常:Avg:52.733fps Min-Max:43.5-58.0fps Tear:84.306%
車内:Avg:47.900fps Min-Max:40.0-54.5fps Tear:78.250%

London
通常:Avg:57.110fps Min-Max:48.5-60.0fps Tear:53.002%
車内:Avg:48.400fps Min-Max:39.5-57.0fps Tear:78.385%

 

[Painraze]

:lol

 

[knee]

EXACTLY! :lol :lol :lol

 

[sdornan]

Don't disagree. Again, it is a "hope". I attribute frame dips and screen tears to two things in all games.

1. Bad code -> as seen in bad port jobs
2. Taking on more than the hardware can handle (assuming good code)

I don't think its 1. PD have some of the best code monkeys in the business. The stress this game puts on the physical hardware has to be taken into consideration on some of the design decisions. I applaud the high benchmarks that PD is trying to deliver. 16 cars at once! Its those design decisions that could have easily freed up CPU cycles and RAM for other on screen performance functions like tripe buffering and MLAA.

EDIT: Easy for me to say in hindsight so take my comments as wishful thinking.

What I've been mulling over is how much having 16 cars on the track actually matters. Typically, in GT5, you're either in first position, last position (if you messed up), or close to the first position. Does it matter that there are eight more cars on the track? When do you even see them other than at the beginning of the race? Wouldn't additional graphical fidelity improve the experience of playing the game more than having eight additional cars on the track that you never see? This isn't an opinion, they're questions.

 

[iam220]

Does anyone know how it the tearing is on 1080i? Because I hardly see any at all :/

1080p here, playing cockpit view. Haven't noticed any tearing at all. Not one. ... and I usually notice that sort of thing.

 

[skyfinch]

What I've been mulling over is how much having 16 cars on the track actually matters. Typically, in GT5, you're either in first position, last position (if you messed up), or close to the first position. Does it matter that there are eight more cars on the track? When do you even see them other than at the beginning of the race? Wouldn't additional graphical fidelity improve the experience of playing the game more than having eight additional cars on the track that you never see? This isn't an opinion, they're questions.

It comes down to the AI. Hell, you can have 40 cars on the screen, but if the AI sucks, they're just moving pylons. If the AI is really good, even a one on one battle can be challenging.

 

[Sykra]

I have played the game for a few hours and haven't noticed that much tearing, I am even quite sensitive to it. They must be hiding it well

 

[cgcg]

78% tearing? There's no way these numbers are correct.

 

[ShapeGSX]

It has to be well hidden. Screen tearing is one of the few things that irritates me and I'm not having a problem with GT5. Sure, I see it here & there (especially piling into a corner with 4 cars ahead, all tyres smoking), but nothing like Moto GP '08 (where the track split through every corner, every time) which those percentages suggest ought to be the case.

I see the screen tear at every single corner in London, even if I'm way out in front. It is mid to high on the screen, so you can't miss it.

 

[Stallion Free]

78% tearing? There's no way these numbers are correct.

I think it makes more sense if it is 78% of frames without tearing.

 

[Fëanor]

Laguna Seca looks way better in Forza:

What's with the difference in looks?

 

[iam220]

What's with the difference in looks?

Well, Forza 3 has it right. GT5 does not.

I think the GT5 version is ported, LS is one of my favorite tracks and I was a bit disappointed with how it looks.

 

[Fëanor]

I would say is missing a few things, like a fence and trees.:lol

 

[ShapeGSX]

A link to Laguna Seca on Google Maps

 

[skyfinch]

What's with the difference in looks?

Time of year? In all the Gran Turismo games, Laguna Seca has always had greener scenery, while in Forza's, it's been more of the brownish tones.

 

[bj00rn_]

Time of year? In all the Gran Turismo games, Laguna Seca has always had greener scenery, while in Forza's, it's been more of the brownish tones.

Yeah, it's only at certain times of year houses and stuff starts to grow.

 

[sajj316]

It comes down to the AI. Hell, you can have 40 cars on the screen, but if the AI sucks, they're just moving pylons. If the AI is really good, even a one on one battle can be challenging.

Good point. Its still crunching the numbers and computing for the other 15 cars off screen. Might not seem like a big deal but I bet there's a good bit going on. Assuming the AI is progressively getting better, does that mean performance (frame rates, tears) degrades on higher levels?

 

[C-Jo]

Yeah, it's only at certain times of year houses and stuff starts to grow.

:lol

 

[Brashnir]

Time of year? In all the Gran Turismo games, Laguna Seca has always had greener scenery, while in Forza's, it's been more of the brownish tones.

Laguna Seca is surrounded by sand year-round.

 

[skyfinch]

Yeah, it's only at certain times of year houses and stuff starts to grow.

Summer months are when houses blossom.

 

[skyfinch]

Laguna Seca is surrounded by sand year-round.

Looks pretty green to me.

 

[Jonsoncao]

Yeah, it's only at certain times of year houses and stuff starts to grow.

:lol

EDIT: LS was one of my favorite tracks in GT4, the other is Nurburgring, it is really a shame that PD didnt revamp LS and just used a lot PS2 resources...

 

[Windom Earle]

QAA in 1080p mode remember tho, yuck

Sharpen option replaces QAA with MSAA (according to Quaz51). He was not completely sure but it looks like it on my set.

http://www.lazygamer.net/lazygamer-feature-lap-time-comparison-gt5-vs-forza-3-vs-real-life/

Goddamn GT

 

[antiquegamer]

I know how it works.

What I'm saying is that even having the option to rewind should be segregated from those who choose to avoid having the option altogether.

It's already is .... people that rewind get move to the bottom. Not sure how much more segregated you want it to be. I think it's the best way to handle all the assist and still have a single meaningful leaderboard.

All this talk about framerate just make me remember before Forza 3 came out and everyone was railing on it because the pre-release built of Forza 3 did not run at 60fps in cockpit mode. Now we have GT5 that couldn't even keep steady frame even without the cockpit (let not get started on no cockpit for the standard car) and people said it's not a big deal... how time have changed.

 

[sajj316]

Sharpen option replaces QAA with MSAA (according to Quaz51). He was not completely sure but it looks like it on my set.
D

So if sharpen is turned off, and resolution is 720p, is it still QAA or does it only display 4x MSAA?

 

[iam220]

Looks pretty green to me.

 

[chifanpoe]

Well, Forza 3 has it right. GT5 does not.

I think the GT5 version is ported, LS is one of my favorite tracks and I was a bit disappointed with how it looks.

I believe you are correct that it was ported over from GT4. However I would take the more accurate track and cars then a few extra trees and houses...

 

[REMEMBER CITADEL]

I think it makes more sense if it is 78% of frames without tearing.

If that was the case, then their previous results would mean that there's practically 100% of tearing in replay.

 

[Death Dealer]

That's as stupid as people who say Traction Control Management in Sim racers is ruining them.

Traction Control Management is a huge part of modern motorsport. Many people, myself included, think it is getting too ridiculous at the degree to which it can level the difference in pure skill. Many riders in MotoGP like Valentino Rossi have expressed this belief. It makes sense though to have that in a sim, since it's become an important factor in racing at the world class level. You can't accurately simulate racing without it. However rewind is not on the same page. It is a built in cheat mode. It's like putting in a code for unlimited credits.

 

[cuevas. PhD.]

So which game is better? The one that took 5 years to make or the one that took over 2 and released a year ago? I am going to guess the former...

 

[Windom Earle]

So if sharpen is turned off, and resolution is 720p, is it still QAA or does it only display 4x MSAA?

720p always uses MSAA from what we know. Not sure what sharpen does in 720p.

 

[Stallion Free]

If that was the case, then their previous results would mean that there's practically 100% of tearing in replay.

Holy shit these percents are confusing then. Why don't they just say the average number of frames torn for 60 seconds of gameplay then. It would be a lot easier to figure out and get understandable data out of.

 

[ShapeGSX]

However rewind is not on the same page. It is a built in cheat mode. It's like putting in a code for unlimited credits.

So is running a race again. Races don't really get organized and run every time you want to run against a certain set of cars at a particular track. In real life, once you run a race, that's it. If you didn't win any money, you can't hit restart until you do.

You're already playing a game. It isn't real life. Neither of these things is a cheat mode. Hitting rewind is only just slightly easier than hitting Start then Restart.

You: "Hey, all you guys in the other cars? I totally blew that last corner and it doesn't look like I'm going to win."
Competitior: "WTF?"
You: "Yeah, I'm going to have to ask all of you to line up again at the start for a re-do. Really sorry."

 

[See You Next Wednesday]

Well, Forza 3 has it right. GT5 does not.

I think the GT5 version is ported, LS is one of my favorite tracks and I was a bit disappointed with how it looks.

Shame too, in GT3 it looked amazing.