Radeon 8500 ansiotropic filtering stuff..
- Thread starter Rancidm
- Start date
What fps put you off? I havent noticed it to a great degree in the ones I have played (CS, Op. Flash, UT, 'Giants', RTCW, SOF2 Demo I tried, UT2003 demo, SS:FE). Of course I usually play with 2xSV which helps eliminate the bilinear lines. I cant say I've noticed mip-map lines in UT2003, but that may be down to the extra texture detail that stands out more than filtering change-overs, even without SV.
In UT I have seen moire a bit in metal grate textures, but you have to look for it.
Other FPSs's like AVP2. NOLF etc I havent played or reach the end of SS:SE so didnt see the revolving rooms.
In UT I have seen moire a bit in metal grate textures, but you have to look for it.
Other FPSs's like AVP2. NOLF etc I havent played or reach the end of SS:SE so didnt see the revolving rooms.
Hi Randell,
more specifically, the fps I found Radeon's bilinear/AF very annoying were SS:SE, AvP2 and RtCW. But that's probably down to personal preference; I find it funny, though, that ATi PR claims that there are NO visual benefits with trilinear/AF when compared to ATi's anisotropic filtering implementation.
ta,
-Sascha.rb
more specifically, the fps I found Radeon's bilinear/AF very annoying were SS:SE, AvP2 and RtCW. But that's probably down to personal preference; I find it funny, though, that ATi PR claims that there are NO visual benefits with trilinear/AF when compared to ATi's anisotropic filtering implementation.
ta,
-Sascha.rb
G
Guest
Guest
nggalai said:
I don't see what is so difficult to understand here. It is obvious here that the Radeon 8500 can use a higher level of AF without the performance hit the Geforce card has. Nvidias method is poor at best.
NVIDIA GeForce4 @ 16-tap
ATI Radeon 8500 @ 16X
http://www.anandtech.com/video/showdoc.html?i=1645&p=12
http://www.anandtech.com/video/showdoc.html?i=1645&p=13
"The race is very close between the Radeon 8500 and the GeForce4 Ti 4600, but the image quality of the Radeon give it the edge. If you were to crank up the anisotropy level on the GeForce4 its performance would plummet. The Parhelia is hanging in there but a definite third place in this chart"
Is this the new complaint from nvidia fans? I guess they really can't complain about the drivers, so they must find something. Personally I think ATIs method is far better then nvidias and I also think nvidia should adopt it.
Hi Geek_2002,
I think you might have misunderstood my post. I am not saying that the quality per se is inferior, nor do I compare it to a Geforce or other competitors' hardware.
I object to the argument as presented in this ATi PR text: that there will be no visual advantage, on a Radeon board, when going from bilinear to trilinear anisotropic filtering because those "bilinear samples" already "smooth out" the MIP-map levels "just as well." The MIP-map boundaries are very visible at low AF settings, and can still be seen at higher degrees in certain situations. Mind, I don't say that this is RELEVANT for each and every actual game situation as you will crank up anisotropic filtering to the max anyway which reduces the effect tremendously--again, depending on the situation, to be non visible.
I personally also agree that a similar "performance-mode" would be a nice feature on future NV boards--as has been mentioned before, it depends a lot on the game (and user) in question if and how quality differences are perceived. So the user should be able to chose, IMO.
ta,
-Sascha.rb
I think you might have misunderstood my post. I am not saying that the quality per se is inferior, nor do I compare it to a Geforce or other competitors' hardware.
I object to the argument as presented in this ATi PR text: that there will be no visual advantage, on a Radeon board, when going from bilinear to trilinear anisotropic filtering because those "bilinear samples" already "smooth out" the MIP-map levels "just as well." The MIP-map boundaries are very visible at low AF settings, and can still be seen at higher degrees in certain situations. Mind, I don't say that this is RELEVANT for each and every actual game situation as you will crank up anisotropic filtering to the max anyway which reduces the effect tremendously--again, depending on the situation, to be non visible.
I personally also agree that a similar "performance-mode" would be a nice feature on future NV boards--as has been mentioned before, it depends a lot on the game (and user) in question if and how quality differences are perceived. So the user should be able to chose, IMO.
ta,
-Sascha.rb
Geek_2002>
I don't know if you are comparing the 2 pics from the GF4 and R8500, but bear in mind that 16 tap <> 16 X. 16 tap = 2 X which is the max on parchelia right now.
But you are right, the R8500 can give a good quality improvement without a big performance hit contrary to the GF4 (but under specific conditions).
I don't know if you are comparing the 2 pics from the GF4 and R8500, but bear in mind that 16 tap <> 16 X. 16 tap = 2 X which is the max on parchelia right now.
But you are right, the R8500 can give a good quality improvement without a big performance hit contrary to the GF4 (but under specific conditions).
nggalai said:
Isn't this pretty much what we have in the 29.42 drivers (using Rivatuner) ?
There's a discussion about this here:
http://www.beyond3d.com/forum/viewtopic.php?t=1417&start=0
Doomtrooper
Veteran
nggalai said:
Are you talking Radeon or Radeon 8500 as the implementation has improved greatly over the original Radeon.
I think we have gone over this before Trilinear and ansiotropic is NOT the answer, the performance hit is too great.
JF_Aidan_Pryde
Regular
You people really need to get a LIFE.
Don't you have anything better to do than to nick pick mip map borders pushed almost beyond visibility and hunt for triangles at 45 degees?
Go out and kick a ball or something!
And Direct3D can't do Trilinear With Anisotropic anyway.
There goes 50%+ of the games out there.
Don't you have anything better to do than to nick pick mip map borders pushed almost beyond visibility and hunt for triangles at 45 degees?
Go out and kick a ball or something!
And Direct3D can't do Trilinear With Anisotropic anyway.
There goes 50%+ of the games out there.
Hellbinder
Banned
Edited by JR: Tone it down! I agree that this whole 'revisiting' of ATi's AF is annoying but posts like this do no good whatsoever.
Hi Hellbinder,
woah. Strong words. Have you actually read my postings?
Also, let me be so rude as to quote myself:
I do apologise if I didn't make myself clear. And I will refrain from a meta-discussion in this thread; should you feel like attacking me personally in the future, please use that handy PM button at the bottom of each of my postings. Thank you.
ta,
-Sascha.rb
woah. Strong words. Have you actually read my postings?
Full ACK. But please tell me where either one of us actually posts his opinions as fact. I think I made it rather clear that I am talking about personal preferences.
Also, let me be so rude as to quote myself:
All I'm saying is that _I_ found ATi's implementation distracting in SOME fps I played--and that was as a by-line. The rest of my postings is concerned about the PR-blurb as published on Firingsquad, NOT about the quality of Radeon's AF.
I do apologise if I didn't make myself clear. And I will refrain from a meta-discussion in this thread; should you feel like attacking me personally in the future, please use that handy PM button at the bottom of each of my postings. Thank you.
ta,
-Sascha.rb
Remember that Mip Map boundries occur because of undersampling. You transition from a level that is slightly undersampling to a level that is undersampling even more. Anisotropic filtering is meant to sample more times so the undersampling doesn't occur.
That's not true. You can use D3DTEXF_ANISOTROPIC as minification and magnification filter, and D3DTEXF_LINEAR for filtering between mipmap levels.JF_Aidan_Pryde said:And Direct3D can't do Trilinear With Anisotropic anyway.
There goes 50%+ of the games out there.
Hellbinder
Banned
Sorry JR... I will will be more *tollerant* frm now on.
To back up what my deleted post said. ANYONE who says that No ansio is higher quality than ATi's ansio... and claims not to be pushing an agenda...
Well...
Please....
To back up what my deleted post said. ANYONE who says that No ansio is higher quality than ATi's ansio... and claims not to be pushing an agenda...
Well...
Please....