Well I still think its AF fault by those UT3 screens, this is what im seeing and by other games too So AMD doesnt has the same wavy look? And normal tunnel mode
I know that the 5850 that I have has a hardware limitation that didn't show up in these test you guys are using , but showed up as distinct bands when the tests were combined. It still exists in my 5850 to this day. AMD tried to reduce it in the 6k series and did a much better job, but it still existed. http://alienbabeltech.com/main/?p=12648&page=2
Wow that is so wrong on so many levels... Your texel coordinates are shifted way off. It's center is off and it's somehow rotated. No idea why, but that is wrong. Edit: You moved it with mouse, did you? well then it's no good for comparison. Default angle has value. This is incorrect It was present on HD 5xxx series, till it got fully fixed by drivers. My notebook w/ HD5870 (desktop 5770) had it and it was impossible to overlook. It's long time fixed.
I have a 5850 currently and the band still exists in many games. I am not sure how you define fixed. infact here is AMDs press release showing the 6000 fixed it. http://img217.imageshack.us/img217/7882/355595ds.png
5850 Spoiler Note the harsh transition on my 5850, which CAN'T be fixed because it was a hardware limitation. catalyst 13.2 beta 6 FYI. This is why I struggle to buy another ATI card.
Yeah, note the distinct bands. On complex repeating textures those show up as shimmering rows. They don't show up in most scenarios, but on games where you have say.. racer with asphalt.. you will see those bands. In some games like Trackmania it is pretty disctinct. In other games where the floor is grated you will see it. But most games don't repeat complex floor textures that much. So it isn't always noticeable. I would like to see these top end cards in action though. Because right now you have Nvidia nearly identical to reference and ATI trying to improve on reference. Is the end result even noticeable.
The 6000 also had AF issues. http://translate.google.com/transla...l/grafikkarten/2010/bericht-radeon-hd-6800/5/ http://translate.google.com/transla...modifiziertem-AF-Update-AMD-Statement-808746/ Videos: AMD: http://www.file-upload.net/download-7311912/Barts-High-Quality.avi.html Note the shimmering lines. Nvidia: http://www.file-upload.net/download-7311959/GF100-High-Quality.avi.html
yes i moved it, when its around 35-45° it makes that transition or wavy look, it goes back to normal by higher or lower angles. speaking of 35-45° angle, driver forced AmbientOcclusion sometimes disappears by this angle, mostly U3E games (sometimes TF2 or L4d2) it depends on driver. Sucks, because it can make a nice difference.. Ok, thanks. Although I saw nvidia goes a little further - distant filter.. Imo both firms need improvements or invent new filtering technique without those waves, this would own
Alright experts. Look carefully at the 7k shots. Look at the inner circle and then move left out of the circle.. you will note a lighter colored band that then transitions in a harsh line to a more normal band in comparison to the alu. The effect is subtle, but there. Also.. if you would all load up ground 2 texture, turn off tunnel, set y movement to -5 and x movement to -5 and see how much shimmer you get I get a lot on my 5850 in distinct bands.
@ Pillmonster There is still a slight LOD filling deep inside the tunnel, nvidia's filtering goes a little further before you see a solid colour. Look at ATI HD5850 and you will see what i mean. @yasamoka TMU is the important one, its a texture filtering pipeline. For example by nvidia you see those small offsets by colour tunnel, ALU is angle independent. http://en.wikipedia.org/wiki/Texture_mapping_unit
the bad thing is that i spoke with developers from both sides and they do not intent of using higher AF, 16x will be the max, even the jump to 32x would be a huge hit and tremendous amount of calculations, can we sign a petition for 32xAF.?