Digital Foundry Article Technical Discussion [2017]

I saw someone criticise the game's graphics the other day for not being much of an upgrade over the original, and it quite irked me that they didn't give any recognition to the technical achievement of such large scale battles without any downgrade from the much smaller scale first game.

 

Pretty big differences there. Having seen the Pro version in person, the low res textures really look out of place at times.

 

Looks great! Sort of selling me on the title the more i watch it in action as well. Impressive engine to fit so many different enemies on the field at once, with friendly and enemy AI battling it out. The improvement to textures was something that caught my eye from E3. It was nice that DF did the compare, the thing that sold me was the top of the Troll's head (in E3) it was just so well defined in terms of texturing.

They did a good job there, the increased detailed textures really start to make this game look quite beautiful, without it, it seems like more of what we're used to.

 

Interesting that you can toggle "Dynamic Resolution" on/off...wonder if that will be in the final build??...

 

I don't see why not. but gamescom was in august? So 3 months, not sure if they can optimize that much. Likely it will stay in

 

On XBX Shadow of war has better textures and a resolution usually between 1900p and 2000p, exactly what we could expect based on GPU Tflops specs alone (no 4K native res during most of the gameplay, which is fine and expected). Framerate is OK when indoor or with few enemies...

But what about the framerate in big open spaces with plenty of enemies ? well, it's hovering at ~25fps :



Now apparently resolution and sharpness of textures are the new must have for DF ? They waited near the end of the video to show us that the game runs at ~25fps when outdoors with plenty of enemies ? Odd. BTW in similar scenes the game is basically locked at 30fps on Pro.

 

performance is more likely to change prior to release than anything else.
they gave a good indication where that build.was at, so I'm not sure what the problem is?
that's not even taking into account what they was told about the build.

once they have retail code they will do more performance tests.

some videos focus on different aspects than others, that's always been the way. This was more about texture detail and the affect it has.
expect at least 6 more vids/articles about the game.

 

The difference is massive wow. But i wonder if we are in a similar case as Metal Gear ? The demo was 720p vs 1080p on PS4.

I mean, this game will be an unusual difference or a pretty standard one ?

IMO, nothing should stop the X to get much better textures unless there is some kind of forced parity.

 

Something to take note of is that this build is from Gamescom. Which was released in August, but required at least 3-4 weeks cert before being released, so possibly a July build. Between July to this week the game has been undergoing heavy optimization for all platforms for release, so it's a bit pre-mature to think that these frame rates at this scene is going to make it to the final build. It's entirely possible that they couldn't get this scene working on the final build for any console and tweaked it so that it could for instance. But the build does manage to make it to 4K native resolution in lighter loaded scenes.

As for resolution and textures, this something DF normally covers; the only difference between then and now, generally X1 and PS4 ran the same settings for almost all games with the exception of resolution. There wasn't much difference to report on, so they didn't bother. Now that there are differences to be spotted they are now reporting on it. I don't think is a trend that DF is trying to start up. Most of the time they did resolution and frame rate differences.

Honestly, looking back, it's actually sort of silly how much we debated the differences back then. Aside from resolution there were basically none. I feel silly to even have engaged in it.
Politically though, now that there is going to be visual differences with the textures, I think people will accept resolution gate. It's not nearly as important anymore as people once thought it was.

I mean I used to read all sorts of replies that 900p was pure Vaseline and they couldn't see anything at all compared to 1080p. The exaggerations were quite major.
Now... both resolution and texture detail is operating at a completely different level, and frankly I think most people won't be able to see the 4K as much as they'll see the differences with the textures. It's going to be an interesting next 3-4 years, or at least until PS5 arrives.

Just remember, that DF offers a 4K service on Patreon.
https://www.patreon.com/digitalfoundry

If you really want to see the differences between the 2 without compression issues you'll need to pay $5 per month. Please do not share their 4K HQ videos however, but I think screen shots are okay. I think paying for the service makes sense if you are really into the games and the graphics.

I think you'll see an even larger discrepancy between the two platforms when you run it through their service.
You can see how much quality is lost between YT and DF's encoder


Sorry if this seemed like a plug. But the level of detail I saw on 4K HDR is what sold me on 1X

 

Resolution at a completely different level ? I don't know if the difference between 2160CB vs native 4K is more apparent than 1080p vs 900p or 900p vs 720p.

 

4K v 1080p I mean, I think people will see the difference in textures more often than they would see the difference between 4K and 1080p; 4Pro has resolution but not textures (in this example). 1X has resolution and textures. I dunno, maybe that English didn't come out right.

 

The difference wasn't that big, like iroboto wrote before. X1 and PS4 are very much similar. Event the bandwidth to the memory (thx to the esram which was harder to do, but at the end it was "just" a slower gpu). Biggest difference was a slight hit in resolution but everything else was almost exactly the same. Now the x1x has much more memory and much higher bandwidth so the textures in memory can use much more space. The ps4 pro has only 512MB more memory for games, which isn't that much if you also think that the render target is bigger because of the resolution. A few better textures may be packed inside this, but not much.
If you think of the massive texture-packs that got released with some games. Sometimes it was >30GB just for higher res textures and somehow the needed textures must fit into RAM. so the optimal answer is to just get more RAM so higher res textures can sit in there (and of course bandwidth to transfer them).
And if you believe it or not, pixel count was really never an issue of the current gen. Yes it improves clarity, but higher res textures and better allaround assets do a much better job in that regards. Pixel-count is less of an issue thx to the better AA modes we got in the last few years, as long as they don't blur to much.

e.g. I was really suprised how good even newer games look at 720p (experimented a bit yesterday with that) on PC if you set everything else to the highest setting. Only the UI was ... well not that good

 

I find 720P pretty blurry even on big screen HDTV. On PC it would be horrendous because you sit so close. Anything non native is terrible on PC whenever I tested it. Even just a little. And I'm far from an IQ snob.

That said I get what you're saying. Just think 720P is a bridge too far, IMO it just looks bad. Maybe 900P or 1080P is where resolution starts to scale down in importance a bit.

I was waiting for Global to pounce on that one cruddy texture in the corner or whatever he would find on Xbox...

 

At least when it comes to living room tvs, the importance of resolution really does scale depending on your screen's diameter and viewing distance. I sit about 2.5 meters away from a 50 inch 1080p Plasma tv, and I accidentally completed Prey in 720p. It definitely looked a little soft, by I attributed this to the post fx pipeline and it didn't really bother me. It certainly didn't look terrible either. Plasmas produce a slightly softer picture than LCDs due to the way their subpixels work, so this might have worked in Prey's favor.
Either way, the improved draw distance and sharper textures are a heck of a lot more noticeable than the resolution bump.

 

Sure, but their conclusion is as if all framerate problems on XBX were going to be solved at release and that the current XBX version is better in all aspects than the Pro game. Which is not the case. Also why not also think that developers could also improve the textures on the Pro games then ? Because those textures in some places look awful and worse than in the first game.

When they compare both games here and compare the higher res textures on XBX, it's obvious the game is running quite bad on XBX and smoother on Pro. But they just compare the textures here, not the framerate...

We can remember their analysis of the unreleased Battlefield 1, they weren't bothered to compare the framerate of the alpha code and showed that it ran differently on both machines. The XB1 had the advantage, apparently and it was because of better CPU for them:

And also at the time they also totally ignored for their conclusion one area where the PS4 code had the advantage (during heavy fog conditions), we knew about that area later in another video, but was also put aside as an oddity.

But now the consistent bad framerate ~25fps (as low as 23fps) on XBX when there are plenty of characters on screen is a small detail, an oddity, that can be ignored for their whole narrative and conclusion because it's going to be magically resolved at release ?

There is no other way to put it: they carefully selected some aspects of the comparison (resolution and sharpness), totally ignored other aspects (framerate) and build their whole analysis and final conclusion based on that with the title of their article, their final verdict, not a comment hidden in a paragraph:

In their previous Pro article about Shadow of War their final take of the game was very different: "Shadow of War highlights the strengths and weaknesses of PS4 Pro in our cross-platform breakdown. " And that was comparing with XB1 and PS4, apparently the game had weaknesses on Pro against an hypothetical unreleased XBX version: the lack of higher res textures that could be seen on an unreleased and more expensive machine. What a weakness !

For the Witcher 3 patch on Pro they didn't forget the only known area where the game runs worse than on base PS4, in the Hog, their article and video was mainly based on that in the framerate video and several paragraphs at the beginning were also dedicated to that area. Even if the game runs on average much better on Pro (during streaming everywhere, in the cities etc.), based on one area, their twitter conclusion was more divided and nuanced: "An impressive 4K Pro upgrade for The Witcher 3 but performance is simultaneously better and worse than base hardware ". And the devs have being known for improving their game continuously, they could have thought that the game could be improved in the hog, well they didn't think that.

 

I'm just glad that Mordor runs sweet on the peasant PS4

 

The Digital Foundry conspiracy continues ... But in all seriousness, if they don't fix the frame rate for retail release, Digital Foundry will spell it out, just like they did for the Evil Within 2.

 

I hope this all highlights to Sony the foolishness of not sticking any more memory in the Pro. An extra couple of GB would have made it a much easier sell, as an upgrade beyond just resolution.

 

I've made a picture to illustrate why DF would give XBX the reason of the doubt. And as Scott Arm writes, if they don't fix the framerate issue, they will call it out on the final compare video.

Both versions PS4 and X1 were compared at the same time. DF is actually comparing old code to release code here. That's quite a different compare.

Don't take that personally.

Once again, DF is not here to compare known quantities. The hardware is locked, it never changes. The only thing that ever changes is the software, and how it's coded to run on the hardware.
We should be using the hardware and seeing how it performs on the hardware as a way to peer into how the game is coded.

We shouldn't be using how great the game looks and performs as a measure of how powerful the hardware is. The hardware never changes, from start to end. The only thing that can change is how devs use it. With different types of hardware now, we can see where the engine hits bottlenecks, and we can see what the game is trying to accomplish. Don't take it personally if your platform is not performing the best.

 

I would hope that the framerate is tighter to 30 in demanding areas with the release build...particularly with the Dynamic Res mode...I mean that's sort of the whole point of it...