PDA

View Full Version : Trilinear vs Trylinear vs Brilinear analysis


Ruined
06-02-04, 11:11 AM
ixbt put up a nice analysis, linked to on the front page:

http://www.ixbt-labs.com/articles2/gffx/nv40-rx800-3.html

Their conclusions:

"We can draw the following intermediate conclusions basing on the statements above: (a) the percentage ratio of areas in which ATI and NVIDIA use only bilinear filtering during "optimisation" is practically the same;

using optimised trilinear filtering brings about a significant increase in filtering speed;

both manufacturers get approximately the same gain from using optimised trilinear filtering;

ATI images will sometimes be sharper than those of NVIDIA, but in the case of a moving camera, ATI images are more likely to feature moire and dithering that result from under-filtering;

NVIDIA enables the user to enable/disable optimisation on current drivers for NV40;

ATI doesn't provide this opportunity: the driver itself decides about optimisation enabling/disabling. For some texture classes, the current driver version never uses optimisation; for others, the driver enables optimisation when it thinks that less detalised MIP levels are minimised copies of the previous ones. "

mikechai
06-02-04, 12:41 PM
Since X800's trilinear optimisation can't be turned off, I guess its only fair to compare against 6800 with trilinear optimisation turn on.

Jarred
06-02-04, 12:55 PM
yeah, we came to that conclusion a while back on a different thread, but people still aren't convinced that there's anything wrong with Ati's filtering, and their waiting for American reviews, as they don't trust "dodgy Russian sites that may have hidden agendas" LOL.

muzz
06-02-04, 02:24 PM
It states that with NV you can control on/off in drivers.
Wasn't that the last set of drivers, AKA 60.72?
If so why did they stop allowing the use of this feature in 61.11 set that is newer?

jbirney
06-02-04, 02:30 PM
Since X800's trilinear optimisation can't be turned off, I guess its only fair to compare against 6800 with trilinear optimisation turn on.

Agreed its fair. But what I found funny were people demained that all of the benchmarks be changed to make it fair. If thats the case might as well take ALL OF THE FX benchmarks and throw them out as they don't allow to turn this off so its also not fair in that case either.

but people still aren't convinced that there's anything wrong with Ati's filtering

I am one of those people that is not convinced that there is anything wrong with ATI's method. I have yet to see a crediable reproductable case in a real game where it shows any IQ loss. I did not say I was happy for the way it was forced. And I would like that option to turn it off.

Jarred
06-02-04, 02:39 PM
Agreed its fair. But what I found funny were people demained that all of the benchmarks be changed to make it fair. If thats the case might as well take ALL OF THE FX benchmarks and throw them out as they don't allow to turn this off so its also not fair in that case either.



I am one of those people that is not convinced that there is anything wrong with ATI's method. I have yet to see a crediable reproductable case in a real game where it shows any IQ loss. I did not say I was happy for the way it was forced. And I would like that option to turn it off.

what about the whole, oh, we forgot to tell everyone about it thing?

I'll restate my statement,

people still aren't convinced that what Ati did was wrong.

jbirney
06-02-04, 02:44 PM
what about the whole, oh, we forgot to tell everyone about it thing?

I'll restate my statement,

people still aren't convinced that what Ati did was wrong.

Thats a different issue. You said with ATIs method. There is nothing wrong with it. Now their maybe something wrong with ATI's PR that I can agree too. To different things :)

evilchris
06-02-04, 03:34 PM
There is nothing wrong with NVIDIA's method either.

jimmyjames123
06-02-04, 04:02 PM
FWIW, apparently the NV4x mode with trilinear optimizations "on" (the so-called brilinear filtering mode) in the newer ForceWare 60.72/61.11 drivers is much improved in quality vs the brilinear filtering mode used with the NV3x with the older driver sets.

In that sense, it might be "fair" to compare the NV4x cards with trilinear optimizations "on" to the ATI R4xx cards with their adaptive filtering algorithm.

muzz
06-02-04, 04:04 PM
I was under the impression from folks that they disabled the turning off of OPS in 61.11, yet you state that they stil give the option in 61.11

Do they give you the option to disable OPS in 61.11?

jimmyjames123
06-02-04, 04:07 PM
The trilinear optimizations cannot be disabled in v61.11, but moving forward with official (non-beta) drivers, there will presumably be an option to disable trilinear optimizations. In fact, there was someone from NV who claimed that users would have even more filtering choices to choose from in the future.

Note that some reviewers used v60.72, some used v61.11, and some used both in their most recent tests of the GeForce 6800 cards.

muzz
06-02-04, 04:09 PM
Thanks JJ

Jarred
06-02-04, 06:49 PM
Thats a different issue. You said with ATIs method. There is nothing wrong with it. Now their maybe something wrong with ATI's PR that I can agree too. To different things :)

I think I mentioned Ati's Filtering, talking about it as a whole, and there are some problems with it, or at least there are now,

like, not allowing the user to disable it, and review sites noticed some "dithering"