Printable View
lets not make this another IQ battle because we all know that:Quote:
Originally Posted by Sr7
a) its pointless
b) penny will become rabid in his ban-box if we do
my point earlier about the 64x AA being impossible was that each gpu has to do 16 samples of AA. Even if 4xGPUs worked well in real world situations (ati or NV) this would still be a giant load on the GPU and with the current gen I doubt it is possible to actually play games with this setting enabled.
Oh dear what ppl come up to with all sorts of stuff and mix everything in one huge useless blob.
This has nothing to do with Temporal AA, it has nothing to do with 64 passes per frame.
Besides, i still think 64xQ FSAA is rather pointless when 8xFSAA (Edge Detect) already renders perfectly smooth edges.
I use 4xFSAA EdgeDetect most of the time anyway and i don't see any real need for higher sampling rates except in very very old games that are capped at 640x480 or 800x600 and you want to play them on big screen.
They should invent a completelly new and more efficient FSAA mode instead just trahing out raw increase in samples.
NVIDIA's Quincunx years ago was step in right direction, AMD's Edge Detect repeated that success again. Matrox was using Fragment AA. Thats what i want and am still waiting. Where is nearly free 4xFSAA?
Then I don't get it, Nvidia was doing fine in AA. The only thing I still haven't tried was ATI's new style of AA in the 4xxx series cards.Quote:
Originally Posted by STaRGaZeR
:rolleyes: Obviously, I was generalizing. But between different models of the same generation, the image quality can be identical, if they use the same chip architecture, precision, and compiler. Between series, you will obviously see differences in IQ.Quote:
Originally Posted by joshd
Like I said, you can diff the images and see how little difference there really is in IQ. What people usually flock to is the brighter default colors that many see on ATI cards. Most people generalize as "better" when something is simply different. Both cards can be made to have the same color/brightness as the other. They just have different defaults.
These are NOT true samples.Quote:
Originally Posted by Sr7
These are probably utilizing RGMSAA (with CSAA components) and combining 4 frames from 4 GPUs.
A single frame from FRAPS will give you 16x CS quality. In motion only then it works as "64x"...
Oh well. ATI's Crossfire Edge Detect goes up to 48x, and is 24x MSAA equivalent per frame (while 16x CS is between 8x and 16x mostly)
Following your line of thought, if NV is doing fine then ATI is doing better than fine in AA. The only better thing in NV cards is AF quality, and that has been like that since the 8800 series.Quote:
Originally Posted by Warboy
+1. And even with my current low res 19" CRT @ 1280x960.Quote:
Originally Posted by Shintai
Instead of battling who's got the most gimmicky AA/AF quality etc, rather come up with some new useful improvements or improve the other modes to be faster, 64x is getting ridiculous!
Yeah, cheap 4x AA please ;)Quote:
Originally Posted by RPGWiZaRD
i had that opinion as well, but after trying 8xAA on the radeon 4850 i fell in love with it :p:Quote:
Originally Posted by Shintai
it looks so much better compared to 4xAA especially if you use adaptive aa respectively transparency aa. edge detect aa gives another IQ-increase, but it eats too much performance so imo it's not worth it (even if i use 8xAA edge detect + adaptive AA in bf2, which is quite an old game, the fps go down the drain with all the grass and fences that are rendered with AAA :p:).
but 64xAA sounds like a complete overkill, 8xAA on ati/16xAA on nvidia is more than enough if you ask me :) however, if they manage to keep the fps drop low, i'm all in for higher modes as well :D
Actually if i remember there were slight inperfections in AF algorithms, while Radeons have perfect filtering ever since introduction of HQ AF mode.Quote:
Originally Posted by STaRGaZeR
GTX 260/280 also have perfect AF now.
16AA in crysis wars is nice to see 25fps barely playable 8AA fine and dandy
Why MUST it look good?
Yeeeesss.... alot of old games. Like before 2006Quote:
Originally Posted by Simps
Quote:
Originally Posted by Simps
One of these days some research firm is going to do the graphics card equivalent of the pepsi challenge.
They're going to set up multiple rigs with different video cards and ask the players what video card is in each machine while they play a game and what they fell the level of filtering is being applied.
Hell this should be done at CES!
Thats impossible to perform because i really doubt anyone can tell the difference between modes beyond 8x FSAA, not to mention tens of xQ, xS, Narrow, Wide, Edge Detect. I can spot a Wide/Narrow tent and Quincunx, but i doubt anyone could ever tell for sure what kind of FSAA is used. I could tell which FSAA level is higher if i could diagnose same scene on different computer but no way to spot that if you're running different scenes.
AA :down: :down:
gimmie 3840x2160/2400p and zero/no AA and ill die happy @ 2012
Dan, start a thread on this. Both of us ask the same questions with every gtx285/295 thread/topic. The question will really be, will nvidia's support for this card be the same as the GX2 cards the produced in the 7000 & 8000 series? If so, I'd go 285, but that's me.Quote:
Originally Posted by dan7777
In responce to this topic, 64x? WTH! I don't even think I run AA at all. Where's the setting for that in UT3? lol. I don't have really any other current games to test it with unless I fire up farcry 2, that must have AA, right? :rofl:
The main difference between both algorithms is that on ATI if you don't look straight to a certain object the driver thinks that object shouldn't get the same level of AF as the ones right in front of you. This angle problem is not fixed even with HQ AF unfortunately, and I think ATI specifically wants it to be like that for some strange reason. Maybe perfomance, but nowadays AF is peanuts for every card out there. I can't find the article about the angles with screenshots and all that, but this two links are also interesting:Quote:
Originally Posted by RejZoR
http://www.beyond3d.com/content/reviews/3/5
http://www.beyond3d.com/content/reviews/47/3
And the AF algorithm hasn't changed since R600 AFAIK.
*update*
Fudo just emailed me that he just spoke to Nvidia at CES and they told him the drivers are faulty and should not be used. :(
What is going on??
He's got a point, lol :rofl:Quote:
Originally Posted by Warboy
I've played three same games on same monitor Samsung T220 with a 3850 Agp and 8800 Ultra and 8xaa in CCC still looks better then 8xQaa.
Yep, and R600 AF is just a slightly tweaked/improved algorithm from the R580 AF.Quote:
Originally Posted by STaRGaZeR
Ehh...i thought if you had toooooo much AA it was a bad thing? The game will actually look worse...
Based on what exactly? More samples you use, the more precise the edge is calculated and smoothed. It's similar to resolution itself. Play Quake 3 at 320x240 and at 1920x1600 on same 22 inch screen. While in 320x240 all distant objects would look like moving blocks, you'd actually see all the detail on 1920x1600. The higher you go, the more detail you have. In fact if you'd go high enough with resolution, there would be no need to use FSAA anymore most probably. But we're talking in a 5 digit numbers for horizontal and vertical line...
Decent performance with transparent/adaptive anti-aliasing at 4x (maybe 8x for hardcores) is a lot more important than 64x...
There's really no difference to be seen from 8x upwards.
IMO it's pointless to use 64XAA in any game! Nvidia remains the same
Aside from powerlines in some games being done better than NVIDIA, there is no perceptible difference. You're only exaggerating reality.Quote:
Originally Posted by STaRGaZeR