Digital Foundry Article Technical Discussion Archive [2014]

[function]

Easy there, Shifty's daddy! You're dangerously close to suggesting there might be better things to do with a platform's power than always chase native res.

 

[London Geezer]

Bannable offence!

 

[Shifty Geezer]

I read that as bananable offense...

 

[AlNom]

So you would drop the resolution to avoid a short drop that happen once during the opening sequence? The XB1 also slightly drops during this moment (an explosion) so we should reduce the resolution on both versions?

I'm not sure what your argument is really when many developers already reduce the Xbox resolution to a larger degree.

edit: More profiling would be needed to even determine if it is worth it, but framerate stability is a pet peeve.

I'm not too sure why you're getting upset about a suggestion.

That's really not worth it. Drops on both platforms are really, really occasional (and short), even in the framerate stress test provided by DF. The first is caused by an explosion (CPU bottleneck?) and the 2 significant others are most probably usual stutters caused by loading data so any drop in resolution probably wouldn't resolve things.

Yes. I clearly stated there was an assumption of GPU limited scenario. The rest is obvious.

 

[Svensk Viking]

Sorry if this is the wrong thread for this, but Far Cry 4 apparently refuses to run on dual core CPUs. Core i3 with hyperthreading seems to work though.
Could this be a consequence of the infamous DX11 overhead and next-gen games starting to use the six cores available in the consoles?

 

[AlNom]

My impression is that the overhead will put a higher load on single threaded perf, so they must be doing a fair bit of multithreading aside from that.

 

[Globalisateur]

Well I don't find the occasional and totally acceptable slowdowns in games like Far Cry 4 anything to worry about, I was just thinking aloud that if a very well AA'ed 1440*1080 or whatever other resolution gives real-life results as good as DF says, then it would be interesting to see the games running on PS4 at that resolution and how that would affect performance. It's been a standard staple lately to see face-off's where PS4 looks sharper but XO runs smoother. Perhaps the reason has nothing to do with the lowered resolution but it would still be nice to see.

But for Far Cry 4 I disagree. Except for one case shown previously (cleverly put a the beginning of the video, because most people just watch the start of a video...remember Unity video?) with one explosion making a very slight advantage on the XB1 version (but the XB1 still stutters here) but other that the video shows that both games runs very similarly when CPU heavy stuff and during streamed loading, with sometimes a 1fps difference between both versions.

My point is that they mostly selected only CPU heavy moments (or during invisible loading) in the video and that the PS4 may well run slightly better in GPU heavy scenes.

As a matter of fact I even found one small evidence about my theory in the video (DF must have missed it and didn't cut it from their scene selection), a GPU heavy scene where there are a lot of fire and smoke (no explosion) where we can clearly see one torn frame on the XB1 side, no torn frame on the PS4.

It's not much, but it shows on certain scenes (GPU heavy), the PS4 has the advantage and definitely runs smoother:

So I disagree with the DF conclusion of FC4 because clearly on some scenes PS4 has a slight advantage, (if there is one torn frame here, there is probably more on other scenes). But in the end the difference should be really negligible and drops on each hardware may only happens like every 2 hours of gameplay so they really shouldn't had the need to write paragraphs about it, particularly in the PS4 preview where the unique significant fps drop they previously shown was in fact from an invisible loading and they almost made the whole article around it.

Finally Far Cry 4 is not pure shooting game but also a exploration & hunting game so it would have been fair to include heavy GPU scenes heavily present when exploring & hunting on some spots.

It's as if they really needed a framerate advantage on XB1 and made the video around that conclusion.

 

[function]

But for Far Cry 4 I disagree. Except for one case shown previously (cleverly put a the beginning of the video, because most people just watch the start of a video...remember Unity video?) with one explosion making a very slight advantage on the XB1 version (but the XB1 still stutters here) but other that the video shows that both games runs very similarly when CPU heavy stuff and during streamed loading, with sometimes a 1fps difference between both versions.

This isn't true. The video also shows the Xbox one pull ahead at 0:59, 2:38, 2:41, PS4 pulls ahead briefly at 2:45, then Xbox pulls back ahead again at 2:48.

So no, that's not true, it happens at several points in the video. And if they'd put even more such moments, you'd be accusing them of filling a video unfairly with moments of the PS4 performance suffered - something you've accused them of a great deal in the past.

Also, a point which seems to be lost in every face off is that this is a few minutes from many hours of testing.

My point is that they mostly selected only CPU heavy moments (or during invisible loading) in the video and that the PS4 may well run slightly better in GPU heavy scenes.

You have no knowledge whatsoever of what they left out, as you don't have the data from the other many hours of testing.

And the initial explosion that drops on PS4 may well be GPU bound be rather than CPU.

As a matter of fact I even found one small evidence about my theory in the video (DF must have missed it and didn't cut it from their scene selection), a GPU heavy scene where there are a lot of fire and smoke (no explosion) where we can clearly see one torn frame on the XB1 side, no torn frame on the PS4.

DF don't count tearing in the overscan area because it isn't noticeable. If DF counted tearing in the overscan area, they would have gone to town on Metro for PS4:

http://www.eurogamer.net/articles/digitalfoundry-2014-metro-redux-face-off

"One small technical note concerns a constant tear at the very top of the screen on both PS4 games, tucked into the overscan area and thus invisible on many screens and mostly unnoticeable on others, unless you're actively eyeballing for it. It's a curious one - some games flip the framebuffer late in order to give themselves some flexibility on rendering time (mostly to reduce input lag - why wait another 15ms for a new frame if you can render it 1ms later?), but the PS4 Metro titles stick doggedly to their 16.67ms refresh. It's a technical curiosity and nothing more really, but worth pointing out for those wondering why there's a tear on every one of our PS4 shots."

It's not much, but it shows on certain scenes (GPU heavy), the PS4 has the advantage and definitely runs smoother:

That doesn't register on their metric for the same reasons they didn't rip apart the PS4 version of Metro.

It's as if they really needed a framerate advantage on XB1 and made the video around that conclusion.

I doubt most people will see it that way.

 

[Davros]

Unless Ive got it wrong the seems to be a tear on the xb1 version (am I right ?)

edit: was the above poster pointing this out with the red line ?

 

[shredenvain]

So you would drop the resolution to avoid a short drop that happen once during the opening sequence? The XB1 also slightly drops during this moment (an explosion) so we should reduce the resolution on both versions?

That's really not worth it. Drops on both platforms are really, really occasional (and short), even in the framerate stress test provided by DF. The first is caused by an explosion (CPU bottleneck?) and the 2 significant others are most probably usual stutters caused by loading data so any drop in resolution probably wouldn't resolve things.

When playing the game I haven't noticed any of those drops during explosions or shootout or even loading, they must be really, really rare. Most interestingly the only drops I noticed are not during shooting/explosion/loading but in exploration phases in some rare spots on the map where vegetation is really dense + alphas from smoke behind the vegetation. I am really surprised that DF didn't record those spots that are here most probably GPU bottlnecked (you can hear the fan spins significantly faster in those areas).

But DF didn't select them in their video, they only showed CPU heavy and loading moments, not purely GPU heavy moments which is very odd... or not.

I can't speak for the Ps4 version, but one situation in Far Cry 4 that causes judder on the X1 version is the use of the camera zoom function. It occurs mostly when looking over bodies of water from the side of a mountain that has a lot of vegetation.
From the DF video and article it seems both console versions have a pretty steady frame rate. I have to say that I am quite impressed with the amount of on screen detail without constant pop in from LOD.
The only time I see any pop in is when flying which can be expected.

 

[iroboto]

Unless Ive got it wrong the seems to be a tear on the xb1 version (am I right ?)

edit: was the above poster pointing this out with the red line ?

Yes. Red lines show tears

 

[London Geezer]

Ewwwww TEARING! So gross.

 

[function]

Yeah, tearing isn't so hot, but at least it's mostly confined to the overscan area and there's very little of it in the game.

There's no doubt from the F/O that the PS4 comes out ahead in almost every way. The occasional frame rate advantages for the X1 are treated more as a technical curiosity than any kind of game changer - as indeed they are in this particular game.

The really interesting takeaway from this face off, for me anyway, is just how little the X1 version loses in terms of perceived detail while operating at only 75% of the resolution. The horizontal scaling and AA combine to make a 25% difference appear much, much smaller in motion.

I think this bodes well for future PS4Bone games, as it makes pushing better pixels as opposed to more pixels, and using more GPGPU for simulation, seem very acceptable.

 

[Globalisateur]

This isn't true. The video also shows the Xbox one pull ahead at 0:59, 2:38, 2:41, PS4 pulls ahead briefly at 2:45, then Xbox pulls back ahead again at 2:48.

So no, that's not true, it happens at several points in the video. And if they'd put even more such moments, you'd be accusing them of filling a video unfairly with moments of the PS4 performance suffered - something you've accused them of a great deal in the past.

Also, a point which seems to be lost in every face off is that this is a few minutes from many hours of testing.



You have no knowledge whatsoever of what they left out, as you don't have the data from the other many hours of testing.

And the initial explosion that drops on PS4 may well be GPU bound be rather than CPU.







I doubt most people will see it that way.

0.59 is a stutter from an invisible loading (already present in the PS4 preview), and watch closely the fps meter: XB1 indicates min 25fps and PS4 26fps, I mean you can't give any advantage here, both versions stutter similarly during an invisible loading, just look at the similar frame-time.

I don't see any drops at 2:28.

2:38, 2:41 and 2:45 display in fact almost exactly the same frame-pacing issue on both hardware, only that one has a few seconds of advance.

The only notable area is at 2:48 where it shows a mere 1fps advantage (3 torn frames on ps4) probably from the same previous invisible loading during fast traversal.

The torn frame displayed on a GPU heavy scene on XB1 (not during a loading like just previously) in itself is only proving the fact that in some cases the PS4 version has the some (GPU) advantage, even tiny.

Metro is completely different because the game has a programmed screen tearing in the overscan which doesn't depend of the load of the scene. You can see this tearing even in the main menu.

But let's go back at the beginning of the video (the opening sequence) which is interesting because this long scene was recorded in the FC4 preview too.

First we can notice that the very beginning displaying a locked framerate on PS4, even with tons of alphas (GPU heavy), has been completely cut off from the face-off. Why?

Secondly, we can see here the exact same spot where PS4 drops at 26fps and XB1 has some frame-pacing issue in the full face-off, guess how this PS4 scene performs on this particular play: Almost exactly like the XB1 performed, with only a short frame-pacing issue + one torn frame at 29fps:

Compare this with the exact same spot on the full face-off:

So now I am seriously beginning to worry. First they cut the very beginning of the scene, displaying some very heavy GPU moment and then they replayed this particular scene on PS4 until they found a version where the fps drops at 26fps instead of 29fps? WTF?

So the real conclusion of this suspicious article should be that both versions performs similarly during CPU load, the PS4 has a slight advantage during GPU load, and the XB1 has the slight advantage during invisible loading with fast traversal (like GTA5, really, maybe like here they replayed some traversal scenes in the PS4 version until they find the lowest fps to be compared to the XB1 version?). But in the end PS4 has the biggest advantage in image quality, sharpness and clarity.

 

[London Geezer]

That's a bit rude.

 

[DrJay24]

The really interesting takeaway from this face off, for me anyway, is just how little the X1 version loses in terms of perceived detail while operating at only 75% of the resolution. The horizontal scaling and AA combine to make a 25% difference appear much, much smaller in motion.

I'm not sure how you could draw such a subjective conclusion without having both versions hooked up to the same device in front of you. You are looking at screenshots and compressed videos.

 

[Davros]

DF don't count tearing in the overscan area because it isn't noticeable.

surely it is if you have a 1080p panel, unless tv makers are putting the bezel over some of the pixels

 

[function]

I'm not sure how you could draw such a subjective conclusion without having both versions hooked up to the same device in front of you. You are looking at screenshots and compressed videos.

Well it would still be a subjective conclusion if both games were hooked up infront of me.

Also, DF's zoom-in screenshot comparison tool is pretty good. If we can't judge anything from these, then just why do we bother looking at them?

Looking at the DF zoom tool and to a lesser extent the videos, I've seen bigger reductions in integrity and detail simply from going from a plasma to a "motion dissolve" LCD. And that's a good point, actually, because most panning in (a FPS) game is horizontal, so you lose horizontal detail most frequently on an LCD ...

... horizontal upscale ftw?

 

[function]

surely it is if you have a 1080p panel, unless tv makers are putting the bezel over some of the pixels

You know that's basically what overscan is, right?

Edit: with overscan on, pixels in the overscan area of the incoming image are cropped as they would appear "outside" the bounds of the panel. You are effectively talking about a small amount of zooming/scalling.

This means you lose the edges of the native image *and* the image you see is slightly upscaled.

By default, almost all TV's are set to overscan.

Even with overscan off and a 1:1 image, it is extremely unusual to notice tearing in the overscan area as you are almost never looking there.

Edit 2: it's why games have a "safe zone" of the display area where health bars / score / essential visible FOV have to go.You can't design a game for display on a TV with any guarantee that the entire image you output will be displayed.

 

[London Geezer]

Are you sure about that? I thought that these days with digital connections and 1080p displays, a 1080p image is displayed pixel by pixel on my TV with zero overscan?