Digital Foundry Article Technical Discussion Archive [2014]

Status
Not open for further replies.
http://www.gamespot.com/articles/ho...-off-ps4-and-xbox-one-potential/1100-6419785/

The resolution will maybe change on PS4 1080p and 2xMSAA incoming???

Hayes also adds that the game's tech guys were looking at the possibility of having the game run at a higher resolution on PS4 in time for the game's launch: the UFC demo runs at 1600x900, according to a Digital Foundry report. "They were looking at whether or not we could have the PS4 [version run] at a higher resolution," he says, "but whether or not their investigations were successful? I don't know at this time."
 
People couldn't tell that the games wasn't 1080p until Digital Foundry told them. Even after playing the demo.

It happened ... again.
 
Or EA is intentionally avoiding the 1080p Vs. 900/768/720p debates.

This is clearly in the conspiracy theory territories but also I'm thinking similarly. It clearly doesn't benefit from that controversy when they could just chuck in some extra MSAA at an identical res.
 
Well it wouldn't be the first game having same resolution but better AA on PS4:

I think NBA 2K14 was the first: 1080p@2xMSAA on PS4, 1080p NoAA on XB1.
 
At the moment NBA 2k14 looks vastly better due to native 1080p res also with 2xaa and without a shit ton of motion blur. And that's a game with more than just two people on screen all the time.
Like everyone said It's a very strange decision they've made, 1080p + SMAA would be globally better no questions about it. I did find the demo blurrier than usual on my 65" but then I thought it was the excess amount of motion blur and that in itself is a wasteful feature.
 
Your just being deliberately obtuse.

I assure that I'm not. I just think that your comment is completely useless for a technical forum unless you have some specific knowledge of their renderer and MSAA usage that you can use to back up your statement. MSAA on a modern GPU encompasses a spectrum of possible configurations that you can use, and each makes various trade offs in terms of resulting quality, performance, and memory overhead. On top of that the resulting cost and quality depends heavily on the details of how the renderer is setup, and how exactly it makes use of MSAA throughout the pipeline. It would probably worthwhile to talk about other recent games, and perhaps point out some general trends among them. But it would be a huge mistake to assume that all games render the same way and that a performance cost observed in a PC game will directly apply to a different title.

So if you happen to be knowledgeable of the specific low-level details of this game's rendering as well as the associated cost of MSAA, then please feel free to share them so that we can have an actual technical discussion. If not, then I don't see what your comment does to actually provide value to the conversation. In fact it comes across as needlessly derisive and inflammatory, which would probably do more harm than good even if you had some actual facts to back up your point.
 
I assure that I'm not. I just think that your comment is completely useless for a technical forum unless you have some specific knowledge of their renderer and MSAA usage that you can use to back up your statement. MSAA on a modern GPU encompasses a spectrum of possible configurations that you can use, and each makes various trade offs in terms of resulting quality, performance, and memory overhead. On top of that the resulting cost and quality depends heavily on the details of how the renderer is setup, and how exactly it makes use of MSAA throughout the pipeline. It would probably worthwhile to talk about other recent games, and perhaps point out some general trends among them. But it would be a huge mistake to assume that all games render the same way and that a performance cost observed in a PC game will directly apply to a different title.

So if you happen to be knowledgeable of the specific low-level details of this game's rendering as well as the associated cost of MSAA, then please feel free to share them so that we can have an actual technical discussion. If not, then I don't see what your comment does to actually provide value to the conversation. In fact it comes across as needlessly derisive and inflammatory, which would probably do more harm than good even if you had some actual facts to back up your point.
Edit; Nevermind, I understand why you take issue with my issue. :)
 
Last edited by a moderator:
looked at a video for the game
http://www.youtube.com/watch?v=d2xLW0NZCLw action happens 2:40 onwards

seems roughly 1/3 to 1/2 screen is just the mat!
MSAA does nothing for this, sure if you had lots of edges in the game yes perhaps you could make a valid reason for choosing this

different game/diff GPU/CPU and all that
4_14_Antialiasing_Graph.png

but MSAA 4xAA (which Ive read is what theyre using) is expensive
Im betting with SMAA 2x they could bump up the res to 1080p (the mat looks better) and the performance would no doubt increase as well
 
but MSAA 4xAA (which Ive read is what theyre using) is expensive
Im betting with SMAA 2x they could bump up the res to 1080p (the mat looks better) and the performance would no doubt increase as well
What do you mean by SMAA 2x? SMAA with temporal supersampling from the last frame? SMAA with 2xMSAA? SMAA with 2xSGSSAA? SMAA with quincunx-blended coverage samples?

With many of the possible options, I'd be astounded if switching to it meant being able to afford a 44% increase in pixels, though the exact characteristics are going to depend a lot on implementation details.

Would raising the resolution increase the load/work the GPU has to do by any decent amount vs just raising the MSAA level?.
Generally, yes. This is especially true with classic forward-rendered games (where MSAA doesn't directly hit the GPU front-end and theoretically has no impact on the shaders), but tends to be true with games using newer schemes; there are various funky costs that could come up, but you're still doing far less texture sampling and shading with MSAA versus full general-purpose spatial sampling.
 
Last edited by a moderator:
looked at a video for the game
http://www.youtube.com/watch?v=d2xLW0NZCLw action happens 2:40 onwards

seems roughly 1/3 to 1/2 screen is just the mat!
MSAA does nothing for this, sure if you had lots of edges in the game yes perhaps you could make a valid reason for choosing this

different game/diff GPU/CPU and all that
4_14_Antialiasing_Graph.png

but MSAA 4xAA (which Ive read is what theyre using) is expensive
Im betting with SMAA 2x they could bump up the res to 1080p (the mat looks better) and the performance would no doubt increase as well

You may have an ever so slightly and mostly unnoticable increase in texture sharpness afforded by the increase in resolution but a potential decrease in texture sharpness as well as a decrease in edge aliasing quality depending entirely on what version of 2x SMAA you chose to implement. More advanced SMAA implementations will also carry a higher performance penalty, at which point you may not end up with enough performance to allow for a resolution increase. And you'd likely still have worse edge AA quality which could be noticeable.

Although for the PS4 it might possibly be a win as it apparently has a worse hardware scaler than the XB1. So going 1080p would allow you to bypass the scaling hardware. Although, I'd imagine the developers actually tested various resolutions combined with various AA techniques to find which gave the best picture quality during gameplay and still determined that better edge AA trumped higher resolution for this game.

Unfortunately, this being a console game (the game we're discussing and not the one the graph applies to), it's not like we can test this ourselves to come to our own subjective opinion on which looks best at the target frame rate (30hz).

Regards,
SB
 
Guys, this discussion could soon be obsolete; the PS4 will get a day 1 1080P patch most probably.

http://www.cinemablend.com/games/EA-Sports-UFC-Could-1080p-PS4-900p-Xbox-One-64534.html

there are currently 2 released games in existence for PS4/Xbox that had a sub-1080P resolution parity;
Assassins Creed 4, and Call of Duty: Dogs. In all of these games, there was a day one 1080P PS4 patch.

Also, SB; I noted your scaler remark: but to be honest, I don't think Penello is a reliable source for specs (or advantages).

edit: we can form a subjective opinion in 10 days on wether: 1080P 4xMSAA 100% obm looks beter than 900P 2xMSAA 50% obm.
Though there really are only 2 possible opinions: 1) "1080P version looks better" and 2) "it's a dark game, fighting games don't need that much resolution anyway, I don't think I see any difference, to be honest: the 1080P looks a little bit too sharp, and the better obm makes it look, I don't know, too smooth? they are both great, comparisons are useless anyway."

Resolution, image quality and shader quality is something that we can measure objectively if we leave platform preference out of it.
 
Last edited by a moderator:
about DF watch dogs downgrade or not article...

i think,
the NPC reaction in the demo is the best.
they shocked by the crash and then help the car driver get out of car

in retail version, they just angry at the car driver. <--- the other comparison not made by DF also shows this same behavior. NPC did not come to help.

and about the graphic... even with downgrades it doesn't run well on PC (according to my CPU and GPU usage, it never nearly reach 100%). I cant imagine its spec requirement if the graphic is the same as E3 demo.
 
they shocked by the crash and then help the car driver get out of car

in retail version, they just angry at the car driver. <--- the other comparison not made by DF also shows this same behavior. NPC did not come to help.
I've seen a variety of actions by NPCs to crashes, including NPCs rushing to help. PS4 version but that should not make any difference.
 
I've seen a variety of actions by NPCs to crashes, including NPCs rushing to help. PS4 version but that should not make any difference.

I haven't witness that yet (PC edition). Usually it's screaming, or them standing around looking perplexed.
 
we can form a subjective opinion in 10 days on wether: 1080P 4xMSAA 100% obm looks beter than 900P 2xMSAA 50% obm.
I think you mean the other way around, then again some ppl cant tell the difference at > 720p ;)

But seriously. Ive only got one game demo on my PC (quake3arena) Now Is there a game demo I can download, that will let me try out FXAA/MSAA/TSAA/SMAA etc whatever?
Im interested to see 900p 4xMSAA vs 1080p (2xAA/orSMAA etc) and seeing which looks better at
1. edges, wires,
2. plain textures
 
Status
Not open for further replies.
Back
Top