3 full next-generation hardware screenshots

I do,but I'll probably have to wait till tomorrow to see it,since I will have about 8-10 pints in me around the time you post the info.

 

It's past midnight here!! c'mon man!

8)

 

bit late to this thread but the AA doesn't look particularly impressive to me.


The few good contrast near hoizontal edges seem to show an ordered grid like pattern (some intermediate shades only for one pixel) and also some blurring.




It's doing well on diagonals but for near horizontal (and probably verticals too, can't tell there aren't any) its worse than a sparse 4xAA.

 

It reminds me a LOT of a high-resolution Quincunx.

It's got that kinda-blur-almost that QC added to a filter. Especially check out the tiles on back wall around the lower level vent panel (only 1 vent on the lower level visible, actually, so you should see what I mean).

You can *almost* not see the seperation between tiles at all, it's so blurry. It could, of course, be some texture filtering doing that (some REALLY CRAPPY texture filtering), but the image as a whole does look a tad on the blurry side.

 

Well 4x OGMSAA with blur? Hmm. Cool.
I mean nVidia cards have that mode since the GF3

It takes 9 samples.
And blurring can be free - hell Quincunx is free over 2x since the GF4.

If that's all nV can do for their next generation, then ...

And since it's 4x MSAA, the FPS should better be high...

 

I think we are over-analysing these shots. Personally I think they look great. A bit soft, but nowhere nearly as extreme as Quincunx. In addition the AA method seems to take care of alpha textures, AND provides 4xAA v. comparable (maybe better?) than 4x ATI FSAA (RGMS).

 

And still the first shot looks very good for 4xAA.

 

Well, if it is Quincunx I know of one b3d regular who'll be overjoyed...
K.I.L.E.R., have you seen this yet?

 

I don't believe it is 4xAA, given the number of intermediate shades it looks more like 9x/16x ordered grid (or something like 8xS maybe)

The original gf256 had the same mode except in supersampling in the det 2 and det 3 series of drivers (refered to as 2x2 (9tap))

ut2003/4 don't have any alpha tested textures do they? It's all blends in those games and they don't show aliasing.

 

neg. I just fired up dm-rankin in ut2004 check out the spider web. it shows aliasing. how do blends work compared to alpha tested textures?

 

I agree. It does look nice.
If the NV4x looks this good and has the power behind it
the NV40 is going to be a big hit for nvidia. Unlike the NV3x.

 

*Cough* GF256/GF2 can't do MSAA at all, so I don't see how that would be possible...

Sure they do... Floor grating and such for example.

 

ATI: I must confess. I'm interested.
________
Xnxx

 

2 hours left

 

In which timezone?

 

I'll say again I believe this is a Qunix still AA mode but with a higher degree of samples. Each pixel will share 4 or more samples with the nieghbouring pixels. Lets wait a couple of weeks and see if I'm right or wrong '


PS Don't PM me for details I have none :/

 

If it is a beefed up Qunn AA it seems to do a nice job.

 

Ooooooh.. You.......

* Natoma taps feet while watching the clock