The Technology of GTA IV/RDR *Rage Engine*

[MazingerDUDE]

The fact is that they are there, but certainly not in anywhere near the quantity to meaningfully affect the test.

This is so strage, also many others claim that they've never noticed any tearing after many hours of playing, while for me it happened so obviously that I was able to capture it with cam

Wish I had a capture device so I could capture a more clean shot of tearing from both in game and cut scenes

 

[pjbliverpool]

Great article. Its amazing to see framerate figures on a console. I would love to see more!

 

[Scott_Arm]

This is so strage, also many others claim that they've never noticed any tearing after many hours of playing, while for me it happened so obviously that I was able to capture it with cam

... image cut ...

Wish I had a capture device so I could capture a more clean shot of tearing from both in game and cut scenes

That seems really odd. If it's happening that often, and remaining on screen long enough to capture it with a camera, I'd say you're seeing abnormal tearing.

 

[Neb]

Great article. Its amazing to see framerate figures on a console. I would love to see more!

Though that is some low average framerates on the PS3.

 

[MazingerDUDE]

That seems really odd. If it's happening that often, and remaining on screen long enough to capture it with a camera, I'd say you're seeing abnormal tearing.

It happened about 3~4 times so far (now I'm playing PS3 version only) As for the shot, I did rotate the in game camera to get tearing, however in cut scenes, the tear just stays in the middle of screen. I do think this is abnormal, since I don't see this normally. It only happens in certain time/weather (I think) and stays this way until time/weather changes.

 

[DUALDISASTER]

I;ve tried telling you guys that the blurriness could be removed on the ps3 verison, just turn the flicker filter off in options. The shadows seem to be identical when you do that. Also what about ped counts in each version?

 

[patsu]

I generally don't notice the blur but I have been trying to find that option to check it out. It's not in my game. Where are you located ?

 

[DUALDISASTER]

I generally don't notice the blur but I have been trying to find that option to check it out. It's not in my game. Where are you located ?

Pause the game. go to display amnd then scroll down to the second to last option, 'flicker filter". Hmm so i guess the ps3 version is the inferior one, that is unless ps3's has more peds and cars...I wonder if it's a development issue instead of the ps3.

Edit: only on the ps3 version the options seems to be missing on the 360's.

 

[dabomb665m]

Dualdisaster, you are really confusing the masses here. Flicker Filter only becomes available when the game is set to sub HD levels, which is not what we are even discussing here.

Unless you have a freak copy that has flicker filter during 720p+ output, everyone else that I've read/talked to don't have this function unless they go below that resolution. The option is there to help fix sub HD rendering issues on non-HD tvs.

Please check your display settings, your TV's output... and confirm you are running at HD before repeatedly telling people to do something that doesn't even pertain to the discussion at hand. It may be possible that 1080i has flicker filter since supposedly that's 480p upscaled... but again that isn't an HD resolution.

Flicker Filter, as far as I know, does not work with HDMI no matter what resolution you have your game set at (at least for me). I don't have component cables to test this.

 

[Philip42]

That would be a stupid move, rather go for the 640p and kick the dither in the nuts. And obtain 100% equal versions incl V-Sync.

This i agree with.

I know it's best if it's possible to obtain 720p rendering and 2XFSAA, but i'd take the PS3 style of non-dithering in 640p any day.

On the flip side, i wish the blurryness can be turned down a notch.

On my PS3 version there is no option to shut off the flicker filter, and i'm not sure how other are getting this in the're options.

One more thing, the 360 was the lead platform and it's style of direction was probably the way the team doing the PS3 version was going to go with. But maybe they realized the PS3's CPU/GPU could not handle 720p rendering, so they opted for 640p and better non-dithering to give it smooth out the jaggies.

I'm sure Rockstar knew that if the same style was used for each, they would have had lesser PS3 units sold.

Because besides that , there are only a few PS3 advantages(bit less pop-in, slightly better load times(1-3 sec's), and it's larger TV view).

I really think this was a move done by Rockstar to make sure both versions sold well, especially considering how well Live is doing, the achievements, and especially the 360's upcoming DLC's.

One thing is for sure, everyone can guess as to why and how the two styles contrast, but until Rockstar says anything about this, we may never know why.

Hopefully the 360 version gets a patch soon for the dithering (bug?), and it would be ironic if that patch happens with the first DLC content.

 

[Mintmaster]

Xenos Hi-Z cache is only sufficient to hold 720p w/ 2xAA. If they are doing a z-prepass, 4x becomes more expensive then simply the cross tile vertex cost.

You can update the Hi-Z when switching tiles. There's a .pptx somewhere talking about loading a 720p screen of Z values from memory and updating the Hi-Z in 0.17ms. About 1/100th of GTA4's render time, and the "lazy" method is only twice that.

Personally, I would be more happy if they did 640p with 4xAA and better shadow filtering (and of course no texturing bug/whatever). Would probably have higher framerate, too.

 

[nAo]

There's a .pptx somewhere talking about loading a 720p screen of Z values from memory and updating the Hi-Z in 0.17ms. About 1/100th of GTA4's render time

1/200th

 

[Mintmaster]

1/200th

 

[joker454]

I wonder if people could bully T2 to patch in a VSync toggle for PS3 version one day.

I wonder if there is some kind of synchronization limitation with their PS3 engine which prevents them from disabling v-sync on that box. Perhaps due to how their integrated spu use in their geometry processing. Just speculating of course, but why else would they not even at least offer an option to disable vsync?

I'm curious, do you think the game would still suffer visually on PS3 if it was the lead version?

Lead platform always benefits, no question about that. We don't really know though if they lead on 360 or did parallel development. Assuming they lead on PS3 though, the vague answer is 'it depends'. Some examples:

- If they dropped to 640p due to being tight on memory then leading PS3 would not have helped them since they still would be memory constrained one way or the other compared to the 360 build, unless they just decided to leave that extra 360 memory unused.

- They could not leverage blu-ray because they want the game to be workable on 360.

- They likely wouldn't go with a deferred rendering scheme again to ultimately be more multi platform friendly.

My 'guesstimate' is that leading on PS3 would not have benefited them as much as people may expect because GTA is a multi platform product, and thus has to be built as such. In any case, given the astronomical budget of the game I have to assume they were able to dedicate a reasonable amount of resources to the PS3 version. If 100 million bucks isn't enough to hire enough programmers, then how much is?

 

[Rockster]

You can update the Hi-Z when switching tiles. There's a .pptx somewhere talking about loading a 720p screen of Z values from memory and updating the Hi-Z in 0.17ms. About 1/100th of GTA4's render time, and the "lazy" method is only twice that.

In order to create a 4xAA depth render target to use the method your describing, you would have to tile the Z only pass. ie. effectively running the z-prepass twice since there would be no established tags identifying which commands are for which tile. And milliseconds matter.

 

[-tkf-]

If 100 million bucks isn't enough to hire enough programmers, then how much is?

Maybe they just didn´t hire those 1337 programmers and relied to much on the input/help from Sony and Microsoft.

 

[MonkeyLicker]

My guess is parallel development like Assassins Creed and Call Of Duty 4 had. I think the differences in artistic details lend this theory credibility.

 

[grandmaster]

The think the only actual difference in artistic detail is the colour balance. The rest - no texture dithering/noise, lower resolution etc - are omissions or compromises made to get the PS3 version running at an acceptable speed. I'm inclined to think that the PS3 version looking better in places is down to a happy accident of the upscaling and lack of noise working nicely with the more processed blur of the background items.

I don't think parallel development was an option simply because for a game of this scale, development work would've started long before the PS3 dev kits were out and about.

By the way, I put together all the rough cuts of the Eurogamer videos into one big, eight minute edit, so they might be going up in the next few days. The edits in the current piece probably show around 50-60% of everything that was rendered.

Interestingly, I also did a version where the video was returned to full speed and was surprised at how well the h264 compression held up considering it's dealing with twice as much motion with the same bandwidth. So that one might go up too, even though its actual comparison value is limited, it just looks cool.

 

[jonabbey]

I
Lead platform always benefits, no question about that. We don't really know though if they lead on 360 or did parallel development. Assuming they lead on PS3 though, the vague answer is 'it depends'. Some examples:

- If they dropped to 640p due to being tight on memory then leading PS3 would not have helped them since they still would be memory constrained one way or the other compared to the 360 build, unless they just decided to leave that extra 360 memory unused.

On this topic, I've a question regarding the typical development cycles for console games. It's a rather open secret now that Sony has driven down the memory requirements of their OS in recent firmware releases, which came out before GTA4.

Is that the kind of thing that a developer would have been able to count on and develop against ahead of time, or would they have been locked into a devel/test cycle against the prior memory constraints?

I know that Criterion is talking about putting a few new features into the PS3 Burnoutaradise thanks to the reduction in memory usage that came about with a recent PS3 firmware release, but I don't know whether GTA4 is likely to be benefiting from the same, given development cycles, and so forth.

Anyone have a feel for that?

 

[catisfit]

if PS3 version always have V-sync and X360 not (over 2 VBL) then it's more easy to have a better average fps in X360 version if GPU don't Wait V-sync whereas PS3 version lost GPU time to wait V-sync at every frame
With no V-sync PS3 averafe FPS would be little better (but with tearing of course)

20% different average is standard fare for what happens when you run two identical tests, one with VSync off, one with on. In fact it's almost exactly the difference I've observed when doing similar tests in my own work.

If anything this suggests that internal fps is really close. I wonder if people could bully T2 to patch in a VSync toggle for PS3 version one day.

It would have been nice if the article explained what the consequences of vsync actually are on a game with a variable framerate.

I did love the bit where "obvious pop-in on 360 that the PS3 version coped better with" concludes that any pop-in whatsoever is "something you wouldn't expect from the PS3 code". That made me chuckle

Props to the people who put it together, it just seemed to be lacking in a little bit of explanation and objectivity.