Digital Foundry Article Technical Discussion Archive [2014]

[zed]

PS4 version has a texture streaming problem yes, not a texture filtering one.

yes from my quickly looking it seems as if what theyre talking about is a straming problem (which resolution effects in no way, texture res yes but not screen res)

 

[FATBOT]

Penalty for AF is next to 0 on modern GPU architectures.

Nope, definitely not:

http://www.computerbase.de/artikel/grafikkarten/2011/test-amd-radeon-hd-7970/7/

(Scroll down to "AF-Skalierung")

 

[London Geezer]

http://abload.de/img/thief28slmz.gif

Watch until the end of the gif, there are indeed shadows pop-ins on... XB1!

This whole article is a mess and should be completely re-written, with a Thief already pre-installed on a clean non-fragmented HDD if possible on both consoles

I always find it funny when people pass judgement on anything without having seen the product in question. In this case no one here has seen the game on ps4, let alone on both consoles and PC, to make any kind of comment on the validity of DF's analysis.

Not speaking of you in particular.

 

[London Geezer]

http://i.imgur.com/uGy0BaH.jpg
http://i.imgur.com/tqopKMt.jpg
http://hqwide.com/wallpapers/l/1920x1080/37/video_games_metro_last_light_1920x1080_36491.jpg

This game has no art direction? seriously?

In fact how can a game have "nonexistent" art direction? Have you even played the game? if yes then please provide a post-apocalyptic game that you think that has art direction because apparently Metro is random geometry thrown together or something like that.

I personally found Metro as one of the best looking games I've played on ps3. Game full of pretty shameful bugs but some aspects of the graphics were really lovely. For a ps3 game, that goes without saying. Lighting, shadows and shaders for sure. Detail textures and normals also quite nice, if not a bit out of place on people's skin for example.
My opinion anyway.

 

[betan]

Having played Last Light recently (too?), I can tell you the game has its visual moments but the discussion is kinda moot since most of the time you are playing underground.

 

[function]

So basically AF > 1080p, better framerate and more comprehensive use of POM. Is this a new low for DF?

This comment is a new low for B3D.

 

[taisui]

This comment is a new low for B3D.

The mindset of if I dislike the result it means the source is wrong this time and only this time fascinates me.

 

[London Geezer]

This comment is a new low for B3D.

Sounds like some people would rather play Crash Bandicoot at 4k than TLOU at 720p...

 

[Strange]

Sounds like some people would rather play Crash Bandicoot at 4k than TLOU at 720p...

We're not comparing two different games.

If we were to simply offer two options:

A 900p, worse frame rate, better AF
B 1080p, better frame rate, worse AF

Which would you vouch for?

However, due to the use of high quality FXAA post-processing on both platforms, the dropped pixel tally isn't a major point of differentiation. Bird cages shimmer, while hair and fur shaders show up aliasing artefacts to a more notable degree on Xbox One in motion - but they are also an issue on PS4 and PC too.

This manifests more aggressively on Microsoft's platform, where drops down to 20fps are possible, as compared to 25fps on PS4 - but in amongst the frame-pacing issues this is trivial. The stimulus seems to be interactions with multiple guards, and travelling at high speeds around the game world - neither of which are necessarily the core tenets of a stealth game. In either case, the PS4 version technically holds steadier at these particular stress points, but the end result still appears choppy.

If you're committed to buying for one next-gen platform or another, the superior texture filtering on Xbox One means it carries itself slightly better in the visual stakes

So basically all the PS4's advantages are overshadowed by larger issues across all three platforms to the point that they don't matter and that the AF stands out so much that they give the XB1 version the recommendation.

I find that difficult to believe.



Also, the oddities of the 30fps is very fishy. two unique frames followed by two duplicates? 122234445666??

It look to me that they can simply just go 60fps...
Putting the issues together it seems like they had more power to spare but wasted it in bad optimization for both platforms. Maybe they'll fix it in a patch like the various other titles.

 

[shredenvain]

I own this game for the X1. The framerate tends to be pretty smooth through most of the gameplay. The drops come mostly in the realtime cut scenes and oddly enough in a video cut scene at the end of the tutorial level. I did see some drops during gameplay while I was beat to death by 4 cops with billy clubs.I have only seen texture popin on a couple of levels at the begining right after the level loads. This happens in quite a few UE3 games. I cant speak on the Ps4 version of the game or the Pc version but as late as Df's faceoff was I would think they would have noticed if the X1 version was missing Occlusion mapping and tesselation. In fact they state all 3 versions share the same settings of Pom and tess. I do find it quite funny though that people constantly use DF results to champion their console of choice but when the results dont match their point of view the are really quick to bash DF and call them out as biased. Everyone can make mistakes but just because you dont like the results dont make them wrong.

 

[Betanumerical]

I cant speak on the Ps4 version of the game or the Pc version but as late as Df's faceoff was I would think they would have noticed if the X1 version was missing Occlusion mapping and tesselation. In fact they state all 3 versions share the same settings of Pom and tess. I do find it quite funny though that people constantly use DF results to champion their console of choice but when the results dont match their point of view the are really quick to bash DF and call them out as biased. Everyone can make mistakes but just because you dont like the results dont make them wrong.

They clearly missed the missing occlusion (parallax?) mapping in there own screenshots, not to mention other comparison videos that show it missing in other areas. How thorough is a technical review when the material they capture shows differences they don't even bother to mention.

 

[Lucid_Dreamer]

They clearly missed the missing occlusion (parallax?) mapping in there own screenshots, not to mention other comparison videos that show it missing in other areas. How thorough is a technical review when the material they capture shows differences they don't even bother to mention.

Exactly! How do you miss what's in the video evidence that you, yourself, provide? How do you choose "winners" in the past based on 4fps alone and yet go the opposite direction on this one (with resolution, etc added on top)?

Never declare any group to be correct or incorrect all the time. It is just ridiculous. No individual or group is correct all the time. Worthiness should be judged on a case by case basis. If your article adds up, based on the evidence you provide, a pat on the back is in order. If not, the article should be condemned. Anything else is asking for trouble and article quality will likely decrease (as seen here).

Gamers have mainly been concerned with three things from last/current gen:

1. Resolution
2. Frame rate
3. Texture work

The first two has always outweighed the third, so why is that different now? This is based on DF's own articles from the past. It's just odd.

 

[DavidGraham]

Nope, definitely not:

http://www.computerbase.de/artikel/grafikkarten/2011/test-amd-radeon-hd-7970/7/

(Scroll down to "AF-Skalierung")

I think they are testing with low resolution for the AF benchmarks. When the GPU is outputting insane amount of frames, any increase in the load of one area of the GPU pipeline is going to be obvious by the loss of some frames. The process is very sensitive in that case.

However when the load is shifted to other areas, like max settings at 1080p, an increase texturing load is not going to matter at all, since the pipeline is now bound by something else, (ie, shaders).

 

[ultragpu]

Think about it, what's the point of having a very small pixel if all the surrounding pixels are the same color? Resolution loses a chunk of it's purpose if it doesn't have good texture filtering, etc, to go along with it.

While it is true the slope angled surface would get blurred but other surfaces would still benefit from a higher native res by being sharper, less jaggies and shimmering. I guess it's also context biased from scene to scene so if you compare a shot looking at a flat wall, it would obviously tip the favor to the ps4 version than a shot when looking at a sloped angle.
In the end I feel like the overall technical prowess in the PS4 version outweighs the AF advantage in the XBone version if only by pure quantity or amount of processing power required. The Xbone version may look more appealing in certain scenes but to say it's more technically advanced than the PS4 version is disingenuous.

 

[djskribbles]

PS4:
+Better framerate
+Native 1080p
+Better use of POM?
-Worse texture filtering

XB1:
+Better texture filtering
-Lower resolution
-Less stable framerate
-More selective use of POM?

It looks like a clear win for the PS4 version. But I suppose poor texture filtering can make quite a difference in an open world game, enough to make some (ie DF) give the edge to the XB1 version. However the lack of AF on PS4 almost seems like a mistake and that they simply forgot to enable it. If the framerate on PS4 was the same or worse, then I could maybe understand. But during gameplay, there are many instances where the delta is 3-5fps in the PS4's favor, which is a considerable amount in a 30fps game.

 

[taisui]

I think the PS4 has texture pop issues, and it sounds like it happens enough to be annoying.
Also while the PS4 has better frame rate, it's not as if one is locked at 30 and the other at 25, both are stuttering frequently.
PS4 also has frame rate issue in some of the cut-scenes when X1 is at 30...

on a side note, this game is not worth even a rental, so who cares...

 

[djskribbles]

But there's also that GIF that shows shadows disappearing on the XB1 version, so I don't know if we can conclude that the PS4 version is worse with limited footage. But I'll give you that. Could be caused by the playgo system, or the particular HDDs used in the systems.

Whenever there's no clear winner I suppose it's debatable which is better. DF are neither right or wrong. But I think people should understand why PS4 owners are questioning the decision. It doesn't help that the issues with the PS4 version seem like they could easily be fixed.

 

[taisui]

I am assuming DF spent hours testing and comparing the different platforms, given 1 GIF doesn't really show the whole picture, neither are the vids and imgs on DF. It could be just bugs that'll get patched, could be coding skills, could be time constraints. Fro example, the red hanging banner show the PS4 has horrible texture, but from the video it gets loaded when the camera moved further back, and the former image doesn't show the whole picture, is what i am saying.

 

[Cjail]

DF should simply correct/update the face-off.
They did it in the past, they can do it now.
BTW DF made mistakes in the past, they are not infallible...they are only human after all

 

[Betanumerical]

Is it possible to have some sort of texture filtering that improves as the camera moves away from the object? because I only just watched the video for the thief article and the texture seems to pop in suddenly, wouldn't this been indicative of a streaming issue and not a texture filtering issue?.

DF should simply correct/update he face-off.
They did it in the past, they can do it now.
BTW DF made mistakes in the past, they are not infallible...they are only human after all

Indeed, the person who wrote this article also thought the AO on the XB1 version of NFS was better because the HBAO on the ps4 was subtle. It was only corrected after Ghosts told them they were wrong.