Nvidia Altered its 8xS Sample Pattern with 61.11 drivers

[ChrisRay]

I probably wouldnt have noticed it earlier, But it doesnt appear to function in OpenGL, But works in Direct3d,

I havent had any chance to really test it. But I am pretty unsure about the pattern. It appears to be 6 samples taken.

http://www.nvnews.net/vbulletin/showthread.php?t=28215

Since I cant seem to upload it here, I will link to where I could. It appears to be 4 samples of Multi sampling and 2 samples of Super Sampling.. I'm gonna do some quality tests in a bit, Personally I dunno if its going to look better or worse,

But its definately less brute force than before. (But its all ordered)

 

[ChrisRay]

I hate to bump my own thread, But 8xS and 8x are totally different than what they were before.

8x Is now used in Control Panel. But theres a new 8xS as well. Nothing at all like the old 8xS.

Check out my thread at nvnews, I intend to post Quality Comparisons between 4x, 6x (Still 12x) The new 8x (6x??) and the new 8xS.

 

[cho]

does the nv40 8x AA screenshot capture on 61.11?

 

[ChrisRay]

Keep in Mind that these could just be broken Modes for the NV35, that are supposed to function on the NV40, Heh.

 

[Lezmaka]

does the nv40 8x AA screenshot capture on 61.11?

And you know for sure those screenshots were taken with 61.11?

Chris said "It appears to be 4 samples of Multi sampling and 2 samples of Super Sampling" which is the opposite of what is in the pic you posted (4xSS + 2xMS)

 

[3dilettante]

Could someone explain to me why the Radeon has really obvious aliasing in the car shadows? It looks like there isn't any AA at all on the shadows.

 

[ChrisRay]

does the nv40 8x AA screenshot capture on 61.11?

And you know for sure those screenshots were taken with 61.11?

Chris said "It appears to be 4 samples of Multi sampling and 2 samples of Super Sampling" which is the opposite of what is in the pic you posted (4xSS + 2xMS)

It could just be 4x SS and no Multi Sampling. I really dont know whats going on the sample Pattern. It has me kinda confused :|

One thing is for certain. Its WAY faster than the old 8xS mode for Nvidia cards, heck its even faster than 6x. (12x)

 

[Eronarn]

Could someone explain to me why the Radeon has really obvious aliasing in the car shadows? It looks like there isn't any AA at all on the shadows.

THG is misinformed (NOT LYING, NO FLAMES PLEASE), perhaps? That seems somewhat ridiculous looking at the image, 2x AA usually looks better, and I find it VERY hard to believe that they dropped image quality.

 

[3dilettante]

Actually, it's not just the car shadows. The cars themselves, the rails, and the rail shadows all show pretty obvious stair stepping. It's something I wouldn't expect a 2x mode to do, much less 6x.

What's up with that?

 

[WaltC]

does the nv40 8x AA screenshot capture on 61.11?

And you know for sure those screenshots were taken with 61.11?

Chris said "It appears to be 4 samples of Multi sampling and 2 samples of Super Sampling" which is the opposite of what is in the pic you posted (4xSS + 2xMS)

I can't vouch for the nVidia FSAA screen shots, of course, but I can tell you the ones representing the 9800 are highly suspect in my view. Take a look at the fencing wire on the outside of the track. They look like they should in the nVidia shot, but in the purported 9800 screen shot they look terrible--broken, jagged, and at times entirely missing from the frame. At 6x FSAA on my 9800P running MTM2 at 640x480 I get nice, clean lines for things like that--like telephone pole wiring, fence rails, curb lines, and so on.

The 9800 screen almost looks to me as if it could have been done with no FSAA at all, compared to what I'm used to (although I don't have this particular game.) Look as well at the car shadows, the cars themselves, and the white curb lines--just doesn't appear to be AA'ed much at all, from this shot, as the jaggies are everywhere, even in this tiny reduction of the frame. Certainly doesn't seem representative to me of 6xFSAA on my 9800P.

Edit: Ooooops, I just noticed this shot purports to be from an x800, not a 9800--looks totally bogus to me, without a doubt...

 

[Pete]

I can see attacking Lars for an error in a published review, but accusing him of devilishness with a leaked image is a step too far, even for a fan of comeuppance like me. It's possible he'd have spotted that error--if it is one (it's hard to tell on such a tiny shot)--in the time he has until the NDA lifts.

 

[Joe DeFuria]

I can see attacking Lars for an error in a published review, but accusing him of devilishness with a leaked image is a step too far...

I agree and disagree...

It TOM'S had been more careful and not had it's sh*t leaked, he'd have nothing to worry about, right?

 

[DemoCoder]

Joe, where's your umbrage over DrivenHeaven's false attack piece on NV (dispelled by Microsoft)? And did you bother with similar outrage when 3dCenter showed images with superior isotropic filtering for the 6800 and claimed it was a huge reduction?

 

[ChrisRay]

heh either these modes are bugged or Nvidia totally screwed FX users with them,. I think its the first, so I am back to the 5x.0 series again.


Nice to see my thread has turned into a THG slamming thread

 

[Joe DeFuria]

Joe, where's your umbrage over DrivenHeaven's false attack piece on NV (dispelled by Microsoft)? ?

Hello? Did you actually read my posts in that thread, or are you just making silly assumptions?

And did you bother with similar outrage when 3dCenter showed images with superior isotropic filtering for the 6800 and claimed it was a huge reduction

Didn't even know that existed.

 

[Doomtrooper]

I can see attacking Lars for an error in a published review, but accusing him of devilishness with a leaked image is a step too far, even for a fan of comeuppance like me. It's possible he'd have spotted that error--if it is one (it's hard to tell on such a tiny shot)--in the time he has until the NDA lifts.

That image looks strangely familiar with a different label:

http://www.tomshardware.com/graphic/20040414/images/fsaa-2.jpg

 

[Althornin]

Joe, where's your umbrage over DrivenHeaven's false attack piece on NV (dispelled by Microsoft)? And did you bother with similar outrage when 3dCenter showed images with superior isotropic filtering for the 6800 and claimed it was a huge reduction?

pfft.
so speaks the most biased "unbiased" person on these boards.

 

[Doomtrooper]

 

[DemoCoder]

Joe, where's your umbrage over DrivenHeaven's false attack piece on NV (dispelled by Microsoft)? And did you bother with similar outrage when 3dCenter showed images with superior isotropic filtering for the 6800 and claimed it was a huge reduction?

pfft.
so speaks the most biased "unbiased" person on these boards.

Pottle. Kettle. Black.

 

[WaltC]

I can see attacking Lars for an error in a published review, but accusing him of devilishness with a leaked image is a step too far, even for a fan of comeuppance like me. It's possible he'd have spotted that error--if it is one (it's hard to tell on such a tiny shot)--in the time he has until the NDA lifts.

My initial thought was that anyone could have manufactured these images and contrived them with the THG logo in an attempt to provide them with a degree of credibility they obviously don't deserve (apparent just from looking at them.) I guess we'll see whether THG intends on releasing anything remotely similar fairly soon now...