Nvidia FXAA in Unreal Samaritan Demo (running on Kepler)

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

taltamir

Lifer
Mar 21, 2004
13,576
6
76
they zoom in on the metal cylinder thing.
But looking at the picture as a whole, especially the leather, FXAA is just horribly blurry and unimpressive.

Also why do they compare it to MSAA instead of CSAA?

And something must be wrong with their driver for MSAA on keplar because that doesn't look like 4x MSAA at all.
 

railven

Diamond Member
Mar 25, 2010
6,604
561
126
they zoom in on the metal cylinder thing.
But looking at the picture as a whole, especially the leather, FXAA is just horribly blurry and unimpressive.

Also why do they compare it to MSAA instead of CSAA?

And something must be wrong with their driver for MSAA on keplar because that doesn't look like 4x MSAA at all.

Glad I'm not the only one who saw that. The 4xMSAA side looked like 1xMSAA or none at all.

Sometimes to make a new feature look good, companies gimp an old and tried feature.
 

LOL_Wut_Axel

Diamond Member
Mar 26, 2011
4,310
8
81
:confused: I don't see any benchmarks in this thread... :hmm:

I guess you didn't see any in the links in the OP, either, or read about how the GTX 680 "is 3x faster than the GTX 580 on the same demo".

NVIDIA's extreme cherry picking of benchmarks gets really old.

What also gets really old is a particular person (not you, but you know who it is) that keeps screaming on top of his lungs that they're not biased towards NVIDIA and yet was screaming about tessellation and DX11 when NVIDIA was a bit better at it than AMD and now that some rumors are saying the tables have turned they're whining about how "NVIDIA is SO much better at DX9" even though months ago they were saying DX9 was "irrelevant". A truly hilarious display of hypocrisy, and the person isn't even mature enough to admit it.

This person also never gets tired of mentioning his setup, which according to him was made with price/performance in mind, yet he spend a few hundred on more extravagant cooling that he could've simply used to get a card that was much faster stock to begin with. He also doesn't seem to understand the second law of thermodynamics, equating temperature to heat output.

Disregarding that, I'll keep the same, reasonable estimate regarding the GTX 680: anywhere from 10% slower to 10% faster overall than the HD 7970. Since it was launched somewhat late, GK104 will need aggressive pricing.
 
Feb 19, 2009
10,457
10
76
Does it really matter whether the GPU Mark Rein is holding is real or if it's Kepler? The card is going to be introduced next week and reviewed the week after. It exists, it's real, it's release is imminent.

If its introduced next week, apparently on the 12th, why not reviews too?
 

BallaTheFeared

Diamond Member
Nov 15, 2010
8,115
0
71
I guess you didn't see any in the links in the OP, either, or read about how the GTX 680 "is 3x faster than the GTX 580 on the same demo".

NVIDIA's extreme cherry picking of benchmarks gets really old.

What also gets really old is a particular person (not you, but you know who it is) that keeps screaming on top of his lungs that they're not biased towards NVIDIA and yet was screaming about tessellation and DX11 when NVIDIA was a bit better at it than AMD and now that some rumors are saying the tables have turned they're whining about how "NVIDIA is SO much better at DX9" even though months ago they were saying DX9 was "irrelevant". A truly hilarious display of hypocrisy, and the person isn't even mature enough to admit it.

This person also never gets tired of mentioning his setup, which according to him was made with price/performance in mind, yet he spend a few hundred on more extravagant cooling that he could've simply used to get a card that was much faster stock to begin with. He also doesn't seem to understand the second law of thermodynamics, equating temperature to heat output.

Disregarding that, I'll keep the same, reasonable estimate regarding the GTX 680: anywhere from 10% slower to 10% faster overall than the HD 7970. Since it was launched somewhat late, GK104 will need aggressive pricing.

Sounds like someone needs to go on a fact finding mission as their conceived notions couldn't be further from actual fact.

It also sounds like that person gave others more credit than they deserved.

I hear tubing is short, and rooms are huge.

I predict that won't actually happen, and IPC has clearly gone up.
 

superjim

Senior member
Jan 3, 2012
293
3
81
Wow you're getting excited at.... wait for it.... a bag?! How many gigawatts is that bad boy, how many shaders does it have? Hey, I have Kepler too, look guys!

180c6a93-c49e-1ee3.jpg

I pi$$ed myself laughing....
 

Grooveriding

Diamond Member
Dec 25, 2008
9,147
1,329
126
It looks terrible. You cannot even quantify fxaa/smaa/mlaa by saying it looks like AA from 10 years ago, because it doesn't, it looks worse. It is the worst looking type of AA to date. o_O Blurry textures and a washed out image, greaaaaaat. Would rather more work so the MSAA used applies to all edges and a big performance hit.

At least it would not look like crap.
 

Grooveriding

Diamond Member
Dec 25, 2008
9,147
1,329
126
And yet one of the best looking games released in 2011 uses FXAA, namely Crysis 2.

You kidding ? Crysis 2 was a terrible game first off, both in gameplay, story and MP - all crap. Subjective though.

Per the FXAA used, it looks absolutely horrible and makes the game a blurry mess. Had it turned off and was forcing 2x SGSSAA. This is all also ignoring the fact that Crysis/Warhead released in 2007 look better than the sequel. They perform better too.

It's appropriate a terrible game like C2 used a terrible fake AA mode.
 

Vesku

Diamond Member
Aug 25, 2005
3,743
28
86
Oh, so it was FXAA giving Crysis 2 that blurry feeling? In the Unreal demo I have to say it was hit and miss for me whether I thought it looked like an improvement over what current AAA titles have had. But I recognize that this is a stopgap update while they continue to hammer away at the Unreal 4 engine.
 

Arzachel

Senior member
Apr 7, 2011
903
76
91
You kidding ? Crysis 2 was a terrible game first off, both in gameplay, story and MP - all crap. Subjective though.

Per the FXAA used, it looks absolutely horrible and makes the game a blurry mess. Had it turned off and was forcing 2x SGSSAA. This is all also ignoring the fact that Crysis/Warhead released in 2007 look better than the sequel. They perform better too.

It's appropriate a terrible game like C2 used a terrible fake AA mode.

I preferred Crysis 2 sp to any other "Modern" style shooter, and found the multiplayer more fun than any other shooter that has come out lately barring BF3, but that's a different beast altogether, so yeah, definitively subjective.

I also think that post-processing AA modes are far superior than regular AA in games with deferred rendering and I don't mind slight blurring, because I find it much preferable to artificially "sharp" looking models.
 

BallaTheFeared

Diamond Member
Nov 15, 2010
8,115
0
71
Oh, so it was FXAA giving Crysis 2 that blurry feeling? In the Unreal demo I have to say it was hit and miss for me whether I thought it looked like an improvement over what current AAA titles have had. But I recognize that this is a stopgap update while they continue to hammer away at the Unreal 4 engine.

Could have also been motion blur.

I'm not seeing an issue with IQ blurring across the board, but then I was never good at this sort of thing.

Also how do you disable FXAA in Crysis, there is no "None" setting that I can find.

b5e40d26.jpg


On another note, uploading 5900x1080 is a pain, nothing accepts it.

Here is Batman AC

FXAA off:

e0512a7f.png


FXAA High:

97ab8e77.png



The nice thing about FXAA, for me personally is not only do I not notice any IQ blurring or detail loss, but I also don't inncur a huge vram increase, or performance hit.

No AA to FXAA didn't increase vram usage at all, and only cost me 4 fps @ 5900x1080... Not bad imo.
 
Last edited:

ocre

Golden Member
Dec 26, 2008
1,594
7
81
You kidding ? Crysis 2 was a terrible game first off, both in gameplay, story and MP - all crap. Subjective though.

Per the FXAA used, it looks absolutely horrible and makes the game a blurry mess. Had it turned off and was forcing 2x SGSSAA. This is all also ignoring the fact that Crysis/Warhead released in 2007 look better than the sequel. They perform better too.

It's appropriate a terrible game like C2 used a terrible fake AA mode.

i absolutely loved crysis2. Millions of ppl did.
Unlike the original crysis, i finished Crysis2. Crysis was beautiful but it lacked the spice to keep me interested enough to finish it. Thats a pretty bad thing. Crysis2 was much more interesting. From the start it sold tons of copies. Millions jumped all over crysis2.

Crysis2 was a terrible game? Your statements are intriguing. It was a smash hit????
wiki said:
-Crysis 2 received critical acclaim from critics. The reviewers praised various graphical attributes
-Gamereactor reviewed all versions simultaneously and awarded the game a 9/10
-The amount of detail is insane, and the effects are incredible.
-They concluded that "it would simply be a shame not to call this the best action game so far this year."
-. OPM calls Crysis 2 "excellent - technically strong, visually outstanding and full of welcome fresh ideas."
-"It doesn't just look good, it looks different. The Manhattan mix of crooked concrete spires and green urban spaces is refreshing after the relentless dark khaki backgrounds of Call of Duty and Medal of Honor."

http://en.wikipedia.org/wiki/Crysis_2#Reception

There is tons of praise for the game. Much more praise than criticism. But haters are gonna hate!

You need to realize that your opinion is just your opinion. You consistently state it as if it were fact. Its not. You arent speaking for a majority. Most of the stuff you complain about is not the most commonly shared view. your opinion often is skew to a minority. A tiny group that tries to make the loudest stink. Its not the majority view, no matter how much you try to convince ppl it is.
 

Arzachel

Senior member
Apr 7, 2011
903
76
91
i absolutely loved crysis2. Millions of ppl did.
Unlike the original crysis, i finished Crysis2. Crysis was beautiful but it lacked the spice to keep me interested enough to finish it. Thats a pretty bad thing. Crysis2 was much more interesting. From the start it sold tons of copies. Millions jumped all over crysis2.

Crysis2 was a terrible game? Your statements are intriguing. It was a smash hit????

http://en.wikipedia.org/wiki/Crysis_2#Reception

There is tons of praise for the game. Much more praise than criticism. But haters are gonna hate!

That said, popularity definitely != quality and gaming press is even worse way to judge a game.
 

BallaTheFeared

Diamond Member
Nov 15, 2010
8,115
0
71
I enjoyed it, but lets keep personal opinions on something subjective that has no point in this thread out of it.
 

ocre

Golden Member
Dec 26, 2008
1,594
7
81
I guess you didn't see any in the links in the OP, either, or read about how the GTX 680 "is 3x faster than the GTX 580 on the same demo".

NVIDIA's extreme cherry picking of benchmarks gets really old.


.

yeah if they are anything like this:

http://chinese.vr-zone.com/wp-content/uploads/2012/03/nEO_IMG_7800_3-665x333.jpg

But strange enough you didnt have anything to say about this utter trash!
this would but the 7870 at the same performance of the 7970, hahaha!!!

Anyway where are these extreme cherry picking benchmarks from nvidia on kepler???

Its hilarious!!!
 

railven

Diamond Member
Mar 25, 2010
6,604
561
126
Could have also been motion blur.

I'm not seeing an issue with IQ blurring across the board, but then I was never good at this sort of thing.

Also how do you disable FXAA in Crysis, there is no "None" setting that I can find.



On another note, uploading 5900x1080 is a pain, nothing accepts it.

Here is Batman AC

FXAA off:



FXAA High:




The nice thing about FXAA, for me personally is not only do I not notice any IQ blurring or detail loss, but I also don't inncur a huge vram increase, or performance hit.

No AA to FXAA didn't increase vram usage at all, and only cost me 4 fps @ 5900x1080... Not bad imo.

In Crysis 2 it was a CVAR setting, and disabling it made the shadows look way better. I'm not sure if still images would do it justice, but disabling the "blurr" affect made the game look tons better (at least to me.)

i absolutely loved crysis2. Millions of ppl did.
Unlike the original crysis, i finished Crysis2. Crysis was beautiful but it lacked the spice to keep me interested enough to finish it. Thats a pretty bad thing. Crysis2 was much more interesting. From the start it sold tons of copies. Millions jumped all over crysis2.

Crysis2 was a terrible game? Your statements are intriguing. It was a smash hit????


http://en.wikipedia.org/wiki/Crysis_2#Reception

There is tons of praise for the game. Much more praise than criticism. But haters are gonna hate!

You need to realize that your opinion is just your opinion. You consistently state it as if it were fact. Its not. You arent speaking for a majority. Most of the stuff you complain about is not the most commonly shared view. your opinion often is skew to a minority. A tiny group that tries to make the loudest stink. Its not the majority view, no matter how much you try to convince ppl it is.

Funny, the console version got a lot of praise, because frankly it was a step up for them. I believe the console version accounted for 80% of sales.

On the PC side, I only read slams against it ranging. It had taken lots of steps back that you'd only have really noticed if you played Crysis 1 indepth. Just things I can name off the top of my head, they gimped the physics (from destructible environments to simple interaction with placed objects), the foilage looked way more flat and wasn't as interactive, the light took a huge hit, from washed out sun flares to shadows that looked like a blurred mess at times and even lacked shape.

The PC version received mix feelings, and I personally felt it seemed most hated it versus liked it. The console crowd gobbled it up. But that was Crytek's intentions - cater to the giant mass and dumbed their product down to the lowest denominator. Selling out hardcore.
 

Grooveriding

Diamond Member
Dec 25, 2008
9,147
1,329
126
i absolutely loved crysis2. Millions of ppl did.
Unlike the original crysis, i finished Crysis2. Crysis was beautiful but it lacked the spice to keep me interested enough to finish it. Thats a pretty bad thing. Crysis2 was much more interesting. From the start it sold tons of copies. Millions jumped all over crysis2.

Crysis2 was a terrible game? Your statements are intriguing. It was a smash hit????


http://en.wikipedia.org/wiki/Crysis_2#Reception

There is tons of praise for the game. Much more praise than criticism. But haters are gonna hate!

You need to realize that your opinion is just your opinion. You consistently state it as if it were fact. Its not. You arent speaking for a majority. Most of the stuff you complain about is not the most commonly shared view. your opinion often is skew to a minority. A tiny group that tries to make the loudest stink. Its not the majority view, no matter how much you try to convince ppl it is.

Note my original post said 'it's subjective' You just wasted a whole lot of rant. Crysis 2 failed hard on PC, if you are a 360/PS3 gamer, then I guess I understand why you liked it.
 

tviceman

Diamond Member
Mar 25, 2008
6,734
514
126
www.facebook.com
I guess you didn't see any in the links in the OP, either, or read about how the GTX 680 "is 3x faster than the GTX 580 on the same demo".

Back up there with the spin big chief. You are spinning the words so much I think you made everyone dizzy.

The EXACT words used were
Today GDC 2012 is upon us, and once again Epic has shown the Samaritan demo, but this time with a twist - instead of three GeForce GTX 580s, the demo was shown running on a single next generation NVIDIA graphics card.

In no way did they say Kepler is 3 times faster than the gtx580. The article didn't say how well, what settings, and what resolution Kepler ran the Samaritan demo. Further, the article never said how well 3 gtx580's ran the demo, nor did they say how well Kepler ran the demo. And when Firebird said "I don't see any benchmarks in this thread..." He is absolutely correct. Samaritan wasn't benched, it was just demoed. They weren't doing speed runs. Please quit trying to prove your points by making up crap to fit your argument.
 
Last edited:

nitromullet

Diamond Member
Jan 7, 2004
9,031
36
91
121038gdgsossqedoly3hq.jpg


Geforce 670TI

Blower type fan, she must scream something aweful!

GTX 580 was pretty quiet, so not necessarily. One thing you have to give NV credit for is that they know how to built a quiet cooler (now). They have years of experience with hot gpus and it shows.
 

tviceman

Diamond Member
Mar 25, 2008
6,734
514
126
www.facebook.com
GTX 580 was pretty quiet, so not necessarily. One thing you have to give NV credit for is that they know how to built a quiet cooler (now). They have years of experience with hot gpus and it shows.

LOL! This is true. I think their entire 500 series lineup was quieter tit-for-tat than AMD's hd6000 cards.
 

MrK6

Diamond Member
Aug 9, 2004
4,458
4
81
Wow you're getting excited at.... wait for it.... a bag?! How many gigawatts is that bad boy, how many shaders does it have? Hey, I have Kepler too, look guys!

180c6a93-c49e-1ee3.jpg
Hilarious :thumbsup:
Ditto. The bag was a smart move on their part. Its much harder to see if its real or not if you can't see it at all!
That and an inanely giant NVIDIA sticker covering it so you can't make out the part. They didn't try too hard, did they? :p
 

tviceman

Diamond Member
Mar 25, 2008
6,734
514
126
www.facebook.com
Note my original post said 'it's subjective' You just wasted a whole lot of rant. Crysis 2 failed hard on PC, if you are a 360/PS3 gamer, then I guess I understand why you liked it.

Do you think Oblivion failed hard for not supporting DX11 and not releasing a high resolution texture pack until 2 1/2 months later?