Digital Foundry v GranTurismo 6

  • Thread starter Thread starter phil_75
  • 123 comments
  • 13,033 views
Well we already knew what to expect from this much awaited evaluation. It doesn't take away any of my enjoyment of GT6. Since GT6 came out I could easily spot drops in performance, I am glad DF think like me in that although clearly present then any dips are less impacting than in GT5, I have consistantly described it as 'less jarring'. I wondered if my fanboism blinded me to the extent of GT6's flaws.
I think I will set my PS3 to 720 now. It doesn't matter what htz it runs at, it is when it fluctuates that causes me distress, Zuzuka in GT5 really used to get on my nerves, better in GT6.
Are you reading the same article?

The performance is worse than GT5, the only part that is "less jarring" is the screen tear.
Every other part of performance is worse.

"Comparisons with Gran Turismo 5 reveal a sequel that has added some genuinely impressive new effects - but there is a drop in frame-rate that impacts the fluidity of the driving experience."


"The frame-rate dips here are just too jarring at times, impacting the interface between player and game, introducing too much inconsistency into the way the cars handle from one race to the next."

Whoops, double post!
 
Last edited:
Yes, think I took that as describing the whole-obviously flawed- game rather than just screen tear. For me it is still better than GT5. Again I come back to Suzuka, runs better, but then I have not had 10 cars in front of me in GT6 yet? Worrying.

XB1 it is then. Goodbye PD.

I'm back. PD get one more try on PS4.
 
How is RAM tied to car sounds?

That's probably a very long post I'm not really willing to research and write right now (and almost no-one would be willing to read).

But, in short: in the same way that graphics needs memory to store the frame buffer (that it's alternately drawing into and sending to the display; actually there are usually two of these), models, textures, shadow maps and working buffers for things like the smoke effects, sound needs memory to store the output bus, control buffers, samples and several working buffers for mixing and effects.

The thinking is, then, that the samples (being the largest usage of memory for sound, much as textures and models are for graphics) could not be improved without extra RAM usage, which the game is probably already maxing out.
I guess it's a good thing PD seem to be using a new method that typically uses less RAM per source, at the expense of computational power. ;)
 
That's probably a very long post I'm not really willing to research and write right now (and almost no-one would be willing to read).

But, in short: in the same way that graphics needs memory to store the frame buffer (that it's alternately drawing into and sending to the display; actually there are usually two of these), models, textures, shadow maps and working buffers for things like the smoke effects, sound needs memory to store the output bus, control buffers, samples and several working buffers for mixing and effects.

The thinking is, then, that the samples (being the largest usage of memory for sound, much as textures and models are for graphics) could not be improved without extra RAM usage, which the game is probably already maxing out.
I guess it's a good thing PD seem to be using a new method that typically uses less RAM per source, at the expense of computational power. ;)
Ok. Thank you for curbing my stupidity before the finger pointing and laughing began 👍 :cheers:
 
Honestly I don't think GT7 will be that much of an upgrade in visuals besides being smoother in frames and detail on the trackside.

I disagree.

While its a very different style of racing game, NFS Rivals shows that while the car models from GT6 don't need to change much, they still have a far bit to do in other areas for the move to GT7 on the PS4. Something as simple as how grime accumulates on the bodywork and glass in Rivals is way ahead of what we have in GT6.

If PD make the most out of the PS4 then I expect to see quite a big jump in these kinds of areas.

On the wider subject, I am one of those who had hoped for GT6 that PD went in the way DF suggested, which was to reduce the res to 720 to lock the frame-rate. Now while I am impressed in quite a few areas with what PD have managed with GT6 and am enjoying it far more than I ever did with GT5 (mainly down to the revised physics), I still wish they had gone down that route (along with 30fps for mirrors, etc.).
 
I disagree.

While its a very different style of racing game, NFS Rivals shows that while the car models from GT6 don't need to change much, they still have a far bit to do in other areas for the move to GT7 on the PS4. Something as simple as how grime accumulates on the bodywork and glass in Rivals is way ahead of what we have in GT6.

If PD make the most out of the PS4 then I expect to see quite a big jump in these kinds of areas.

On the wider subject, I am one of those who had hoped for GT6 that PD went in the way DF suggested, which was to reduce the res to 720 to lock the frame-rate. Now while I am impressed in quite a few areas with what PD have managed with GT6 and am enjoying it far more than I ever did with GT5 (mainly down to the revised physics), I still wish they had gone down that route (along with 30fps for mirrors, etc.).

PD showed a video of such "accumulation" effects before GT5 came out, in relatively high quality (it looks texture based: RAM); I think it's part of the long game thing. They seem to have been prototyping systems (presumably on beastly hardware), ironing out the kinks, optimising and then bringing it into the game when they can shoehorn it into the overall hardware budget. No doubt that particular system has had more work and is being prepared for the PS4.

Not to pick holes, since I agree in total, just something I've been noticing over the years. Kinda makes the "GT7 this year" claim more plausible (if not quite believable yet). I agree with your other points, too.
 
PD showed a video of such "accumulation" effects before GT5 came out, in relatively high quality (it looks texture based: RAM); I think it's part of the long game thing. They seem to have been prototyping systems (presumably on beastly hardware), ironing out the kinks, optimising and then bringing it into the game when they can shoehorn it into the overall hardware budget. No doubt that particular system has had more work and is being prepared for the PS4.

Not to pick holes, since I agree in total, just something I've been noticing over the years. Kinda makes the "GT7 this year" claim more plausible (if not quite believable yet). I agree with your other points, too.
I'm not holding my breath for it this year, that would be quite a major surprise if in did launch in 2014. I do however expect to see something GT7 related this year, even if its just in video form.
 
The above is from the article's conclusion, sounds a bit more serious than .01 seconds of input lag to me, and it's something that I noticed myself while playing through the game.
Because it's not the 40fps, the problem is that it's not constant.
From DFs controller latency article:

The lowest latencies a video game can have is 50ms (three frames) - the PS3 XMB runs at this rate, but few games reach it.

Most 60FPS games have a 66.67ms latency - Ridge Racer 7, for example.

30FPS games have a minimum potential lag of 100ms, but many exceed this.
Ok, it's 0.03-0.04 in this case, not 0.01.
 
Last edited:
Because it's not the 40fps, the problem is that it's not constant.

Ok, it's 0.03-0.04 in this case, not 0.01.

The fact is there's pretty bad lag, and when the lag occurs FFB is affected, and when that happens in the middle of a high speed turn when the car is trying to rotate that can be a real piss off.

BTW, you quoted me for the .01 comment, that was axl54 :sly:
 
I personally enjoyed the comment about RAM and sounds, but yeah, good article. Glad I'm not the only one who thought PD were playing the long game.
Yep, well beyond what this article covers are a wealth of other tell-tail signs.
 
You mean the framerate varies depending on any number of things with nothing beyond anecdotal evidence suggesting that it is console revision.

So, according to you, the entire thread about which PS3 console version you own affects FPS is BS?
 
BWX
No it doesn't! That is bunk.

So someone posts a detailed review of different PS3 console versions and their affect on GT6's framerate, and it's bunk because you say it is?

Seems legit.

...snip...

All I did was I observed something and reported it here giving with as much information as I could. Take from it what you will. Maybe all varying ps3 models perform the same, and each console differs on an individual basis like some are saying. I don't know. I did not test more than 1 of the same model as I have said all along.

The point is there is definitely a difference from one console to another because I did a back to back test in many identical curcumstances. I ran the tutorial race on my slim and jiggled the car, and MASSIVE ammounts of tearing. I did the same on the 3002b slim and there was no screen tearing.

I've bought the same car on my 2001 slim and it jerked as I made the purchase animation and it jerked badly. I bought the same car on the 3002b slim and it was smooth. Thats proof enough for me. Whether you beleive it or not I don't really care. I dont have access to a capture card to record at 60fps and analysis software. Nor do I feel the need for it as I the case is clear in my mind.

...snip...

I think I'll take the word of the person providing proof.
 
So someone posts a detailed review of different PS3 console versions and their affect on GT6's framerate, and it's bunk because you say it is?

Seems legit.
Because it is an illogical conclusion... with no proof whatsoever to back it up.

----------------
-------------------
unrelated to that.....


It is interesting that DF did not do ANY cockpit view FPS/ tearing tests. Was it too bad to show?
 
BWX
Because it is an illogical conclusion... with no proof whatsoever to back it up.

LOL. See my edited post above. Snake provides plenty of proof. Tornado was even smacked down in that thread as well.

:rolleyes:
 
So, according to you, the entire thread about which PS3 console version you own affects FPS is BS?

I like how you're pretending that this isn't just some attempt to make some clever cheap shot because of your irrational personal grudge.




But yes. I'm saying that the evidence that the OP provided is insufficient to prove the premise stated by the OP of that thread. Had you read beyond the first and last page you would see the specific issues people took with the proof you're so quickly throwing yourself behind; for starters that the test had so many variables that were ignored that it can't be called scientific in the least. If you're interested in actually discussing the test rather than just blindly agreeing with it because I disagreed with it, perhaps you can take it to that thread and refute the issues people had with it there.
 
LOL. See my edited post above. Snake provides plenty of proof. Tornado was even smacked down in that thread as well.

:rolleyes:
Just stop. There is NO PROOF. It is ALL speculation.

This thread isn't about that thread.

If D.F. thought different console revisions provided different performance in GT6, then he would have tested them! I will not respond to any more of your comments on that subject here.


Back on topic.

It has also been suggested that GT6 employs a form of adaptive tessellation that dynamically adjusts the polygon mesh based on camera proximity. With the limited resolution of the game and the speed at which everything moves, it's difficult to say whether this pursuit was worthwhile in the end. Perhaps it is limited to the 30fps replays because we still see LOD popping on the car models in-game - the process where lower detail models are swapped in for higher poly versions as you move closer to them.

I see LOD popping in and out in the replays and in the races alike. I think this adaptive tessellation only happens on the wheel fenders.. and looks pretty poor most of the time. It is definitely not replacing any L.O.D. models.
 
I like how you're pretending that this isn't just some attempt to make some clever cheap shot because of your irrational personal grudge.




But yes. I'm saying that the evidence that the OP provided is insufficient to prove the premise stated by the OP of that thread. Had you read beyond the first and last page you would see the specific issues people took with the proof you're so quickly throwing yourself behind; for starters that the test had so many variables that were ignored that it can't be called scientific in the least. If you're interested in actually discussing the test rather than just blindly agreeing with it because I disagreed with it, perhaps you can take it to that thread and refute the issues people had with it there.

No personal grudge, although it's pretty enlightening you think that. It's also pretty comical you assume I haven't read the entire thread. You know the saying about assuming, I assume :lol:

Griffiths response to your last post in that thread, which you even liked, about how that OP is accurate.

All things considered, the title is accurate. The thread has already elicited (and the OP has agreed) that there is not necessarily a concrete relationship between model and "performance"; but what is certain is that some people are clearly having a better time of it than others. It varies from console to console. The performance depends on which console you got, which seems to be true even amongst equivalent models.

...snip...
 
BWX
Just stop. There is NO PROOF. It is ALL speculation.

This thread isn't about that thread.

If D.F. thought different console revisions provided different performance in GT6, then he would have tested them! I will not respond to any more of your comments on that subject here.

Jesus... Get worked up much? There was testing done by the OP, so it definitely is NOT all speculation, even though you continue to scream in that thread as well as this one now. Just because Digital Foundry hasn't done it doesn't mean it doesn't hold water.

Anyways, bedtime for this guy. Night gents.
 
No personal grudge, although it's pretty enlightening you think that.
Perhaps you're right. This time you didn't even randomly make a post for no other purpose than to threaten me.


But here's an experiment for you:
Tornado was even smacked down in that thread as well.
I've only made 4 posts in that thread, and one of them was a detailed breakdown for how the OP could be correct; so that leaves 3 posts where I was debating about the standards of proof required to justify the two viewpoints in that thread.



Explain where the "smackdown" occurred regarding those 3 posts that proves the accuracy of the OP's original statements.


It's also pretty comical you assume I haven't read the entire thread. You know the saying about assuming, I assume :lol:
Well, there were two options in my mind: Either you didn't read the thread beyond a few select posts at the beginning and end, and thus missed the several pages of posts back and forth that discussed why the testing criteria used in the OP had issues (as well as the discussions over the hardware capabilities of the PS3 and how the OP's assertions compare). Or you did read them and deliberately misrepresented the overall content of the thread anyway to try to defend a misleading point simply because I pointed out it was misleading when brought up in this thread.


I decided to give you the benefit of the doubt.


Griffiths response to your last post in that thread, which you even liked, about how that OP is accurate.
I like how you quoted the part that says "The thread has already elicited (and the OP has agreed) that there is not necessarily a concrete relationship between model and 'performance'".
 
Last edited:
I'm just wondering which PS3 model these guys used to do the framerate tests. Remember that the frame rate of GT6 varies depending on which version of the console is used.

https://www.gtplanet.net/forum/threads/gt6-framerate-depends-on-which-console-you-have.295306/
If you believe that you will believe anything!

It depends on how much of the game was installed, what harddrive, size, cache, fragmentation etc. Thats all it is.

But lets just say it was true, they would have to test it on the worst of those systems because that's the bottom line.

Wouldn't look good if DF had used the "best" of those systems to do the tests in the article!

I would love to see it though to put it to bed once & for all.


Would be very interested what @Griffith500 thinks?
 
Last edited:
Ive had some moments before GT6 was released where i questioned some of the GT6 pics as real life. The video of GTR GT1 and R8 GT1 going at it in the green hell which were called the first of the GT6 gameplay footage had me mistaken for real life footage for an hour. Either they did an outstanding job on the graphics and camera angle or i'm just getting old lol. Even i can still see FM5 graphic as a video game (not hating the game though) whereas im confused when it comes to comparing GT6 and real life in some aspect.

Anyway, I don't know if i'm ready for GT7, i might change my mind on not getting a PS4....
 
And because of the blood rage most readers fail to read the last sentence of the article:

"Despite falling short of the original goals, the end result is still remarkable - this is the most complete Gran Turismo package released to date."
 
And because of the blood rage most readers fail to read the last sentence of the article:

"Despite falling short of the original goals, the end result is still remarkable - this is the most complete Gran Turismo package released to date."
That's their opinion. Most people here are discussing the facts they presented us before that last sentence.
 
The article was disappointing, it covered LESS details than the GT5 article from years ago, and even contradicts at times...

"Intriguingly, the 1080p mode, processing 50 per cent more pixels than 720p, manages to keep pace very well indeed - sometimes it even outperforms the 720p mode.... So how is this possible? There isn't just a resolution difference between 720p and 1080p, of course, there's the matter of the anti-aliasing too. Dropping down from the 4x MSAA of the 720p to the 2x QAA of 1080p will help in bridging the gap, so one of the remaining differentiating factors will be in pixel-shading the cars."

That was the GT5 analysis for example, but in the GT6 analysis...

"As with GT5, an optional, faster 1280x720 mode can be engaged by selecting this as the maximum supported resolution from the PlayStation 3's XMB."

They need to get back on their A-Game next time...
 
Turning off music by clearing the playlist also slightly improves performance during gameplay, I played with music default playlist and some music from my HDD on 1.0, then I disabled music in all modes ( race, replay and menu ), and I see some improvement in loading times, and game play frame rate. This alone would cause variable on older PS3 models ( fat ones ) with their aging BR lens reading from the disc or old HDD playing the music ( digital version ).

If wanted to test, use 3 PS3 from 3 generation ( good condition - cleaned inside for the older ones), last model from fat, last model from slim and latest super slim, install the same HDD model + size on all of them, same 4.53 FW, format the HDD, same display setting under XMB ( Full color range, deep color on etc ) and install/run GT6 Download version only on latest update 1.03, run them on 1080p, same car, same track ( Bathurst arcade race, pro, 10 AI car, 100% weather, 100% water ), same ingame settings - display and sound ( sharpen, screen size, disable music, DTS 5.1 audio on small theatre ) - Save a race replay from one of them, copy the replay to the other 2, capture the video of the replay in cockpit cam and track view cam from all 3 PS3, then analyze the frame rate. I wonder if DF do this, simple really. DF can then show if different generation PS3 can give different result, best, worst and average fps on all 3 PS3 would be interesting to know.

I think there will be variances on all 3 PS3 generation, even if it might be only 1 to 2 fps on average or it could be 5 or more :)
 
Last edited:
I'm just wondering which PS3 model these guys used to do the framerate tests. Remember that the frame rate of GT6 varies depending on which version of the console is used.

https://www.gtplanet.net/forum/threads/gt6-framerate-depends-on-which-console-you-have.295306/


That was a nonsensical analysis. Easily discarded when you see a PS3 slim filled to the brim with games, videos and other things perform worse than a cleaned out PS3 fat. So it's not just the chipset that affects performance but the content on the HDD, amount of dust inside the console etc etc.
 
Back