AF again

An illustration of what I mean in Mafia about aniso :

NoAniso
mafia_noaniso.jpg


4xAniso
mafia_4xaniso.jpg
 
Play a game like Mafia and you may just revise your statement.

When I have do a lot of slow driving around in Mafia, I appreciate aniso more than AA. Which means to say I appreciate aniso more than AA in a game like Mafia. Not being able to see the white dividing lines on the road is more disturbing than aliasing.

Of course, AA is still preferred but AA + aniso in Mafia is a no go with a GF4Ti4600 due to performance reasons. I haven't checked which results in a greater performance hit (various AA vs various aniso) but I didn't bother because I felt that aniso was more important in Mafia to me than AA.

Actually, I finished Mafia on my gf4ti4600 and I played it with Quincunx AA+Aniso 8x all the way through and it was very playable (around 20-25fps on car and about 30-50fps on foot). I have to disagree though, when playing with Aniso 8x enabled (just that), that didn't do much difference to the overall IQ (although less shimmering was obvious), but when playing with AA enabled, I could see almost no jaggies anywhere, which counts more for me than "removing some of the bluriness far away". Definetly, a combination of both at the highest quality is the most preferable way to play, but unless you own a R9700pro, no card can provide you with a fluid enough experience for that.

A combination of AA & Aniso is the way to go, using them seperately kinda ruins the experience. One technique accounts for the faults of the other and vice versa.
edit:
I can definetly see your point with these pics, but let's be fair here: while you're playing the game, it's not THAT noticable and doesn't distract from the overall IQ, while seeing the bumper of your car jagged as hell does indeed destroy it (at least for me it does).

Conclusion: A combination of AA & Aniso at the highest settings is the way to go and does indeed have a great impact on the overall IQ, but when using Aniso & AA seperately (either that or that), I prefer to use AA.

P.S
Keep in mind that the impact Aniso has on Mafia is quite substantial, while the impact it has on other games is much less visible (in Q3A for instance, it's only good for taking pics).
 
Different strokes for different folks then. The blurriness isn't "far away" to me :)

Again, it's different for you - the blurriness stood out more than the jaggies in this game for me... more to the point, the blurriness detract more because it is unreasonable to expect the lines on the road to blur so quickly (and disappear altogether a little further on!). IOW, the blurriness at such (near) distances is less realistic than a jaggies.

BTW, since you mentioned the game is playable with a Ti4600 - even during those gun battles without truecolor textures + lightmap and QC+8xAF ?!? Not with my Ti4600+XP2000+ !!
 
BTW, since you mentioned the game is playable with a Ti4600 - even during those gun battles without truecolor textures + lightmap and QC+8xAF ?!? Not with my Ti4600+XP2000+ !!

To tell you the truth, the game was unplayable for me with Quincunx + Aniso 8x with the 30.82 detonator set, but when I tried to play it with the same settings with 40.71, it was smooth as silk (compared to the slideshow with the 30.82 detonators :) )

So maybe you should try those...
 
Dave Salvator is just plain wrong in his article by claiming that NVIDIAs anisotropic filtering is not adaptive. It IS adaptive too, so his table claiming the number of tabs beeing always 16, 32 and 64 is just wrong.

Nvidia's marketing is good. I read severals reviews where I read the same wrong claim about NV2x anisotropic filtering using always the maximum degree of anisotropy.
 
alexsok said:
unplayable for me with Quincunx + Aniso 8x with the 30.82 detonator set, but when I tried to play it with the same settings with 40.71, it was smooth as silk (compared to the slideshow with the 30.82 detonators :) )

So maybe you should try those...

I tried it with the new drivers, it is unplayable with the settings you claim you played the game.
 
I tried it with the new drivers, it is unplayable with the settings you claim you played the game.

Well, I donno, but it was smooth all the way for me at 1024x768@32bit, everything set to max with aniso set to 8x and quincunx AA.
 
I don't know whar alexsok's agenda is but AF is extremely noticeable.

I know from several of my non-gaming friends that AF is something that the average Joe notices a lot more easily than edge AA.
 
Games with large outdoor enviroments Anisotropic filtering to me is far more important, games like Tribes and some UT 2003 maps Anisotropic filtering is better for sure and the Quake 3 screen shots and MAFIA above show why.

But why debate over which is better FSAA and Anisotropic, the Radeon 9700 eliminates having to choose, you can run with both on easily..which is the way it should be.
 
You should definitly rush to the next shop and get the first Radeon 9700 you can get, 'cause your Ti4600 is too slow for both. =)

Well, in most games, Aniso 4x+Quincunx AA is a good balance of IQ & Speed... but if we're talking about Aniso8x+4xS AA - can't argue! :)
 
Quincunx AA is a BLUR filter..nothing more and IMO should not be classed as FSAA, you can get the same effect by buying a pair of glasses and rub baby oil on the lenses when playing or a cheap Komodo monitor...and screenshots are not capturing that effect as proven by Sharkfood, so don't go there.
 
Doomtrooper said:
Quincunx AA is a BLUR filter..nothing more and IMO should not be classed as FSAA, you can get the same effect by buying a pair of glasses and rub baby oil on the lenses when playing or a cheap Komodo monitor...and screenshots are not capturing that effect as proven by Sharkfood, so don't go there.

You have a point there, but since the perfomance hit is too big for me, I can't use anything higher than that in conjuction with Aniso (4xS for instace). Using Quincunx AA in conjuction with Aniso 8x can remove most of the bluriness, so I rely on that! :)
 
Doomtrooper said:
Quincunx AA is a BLUR filter..nothing more and IMO should not be classed as FSAA, you can get the same effect by buying a pair of glasses and rub baby oil on the lenses when playing or a cheap Komodo monitor...and screenshots are not capturing that effect as proven by Sharkfood, so don't go there.

Doom, Alex is standing in line for a 9700 like you're in line for a NV30, so let's please just drop this before it gets ugly. ;)
 
I don't know if this has been mentioned yet...

But in that article, they showed 2 Serious Sam shots...Here's where Serious Sam can sorta screw you (but it's a good thing it's in there)...

If you're not careful, the game _will_ set the degree of anisotropy, regardless of what the driver is set to. I did a similar thing some time ago, when benchmarking Serious Sam. I set the driver to, say, 2x...benchmarked it...quit the game...set the driver to 8x...fired up the game, benchmarked it, etc. I was getting the same results, regardless of what the driver was set to.

As it turned out, the game was setting the anisotropy level to "2"...and this is what I was benchmarking, irregardless to what the driver was set to.

Has anybody else experienced this? That's the only thing I can think of that would explain the non-AF screenshot looking the way it did.
 
Has anybody else experienced this? That's the only thing I can think of that would explain the non-AF screenshot looking the way it did.

Also depends on what the drivers are doing. For instance enable any mode of Aniso on a Radeon and it will override what the game says, this is different from a GeForce.
 
Yeah, good point. That screenshot looks like it has a very low level of filtering...and I'm banking on the fact that the author used a GeForce to show the difference, but didn't realize the game engine was defaulting to "2x."
 
But in that article, they showed 2 Serious Sam shots...Here's where Serious Sam can sorta screw you (but it's a good thing it's in there)...

If you're not careful, the game _will_ set the degree of anisotropy, regardless of what the driver is set to. I did a similar thing some time ago, when benchmarking Serious Sam. I set the driver to, say, 2x...benchmarked it...quit the game...set the driver to 8x...fired up the game, benchmarked it, etc. I was getting the same results, regardless of what the driver was set to.


This is what bit me when doing this piece. The AF setting is buried in the bowels of the Advanced Rendering settings, and it supercedes whatever driver setting you're trying to force on (on nVidia's 40.41 driver anyway). We've updated the story to reflect the new screenshots (http://www.extremetech.com/article2/0,3973,559039,00.asp), and did find that for shooters where you're looking at long, sloped surfaces, AF certainly can make a difference.

I also came across a place in NASCAR 2002 where it helps, which is when you coming up to pass another car, AF helps clean up the side-panels of the car and the decals and car number. It's much more subtle than the effect seen in Sam, but it is there.

I'm still of the opinion that in most cases, if you have to choose between AF or FSAA, I'll take the latter, though as someone pointed out on this thread (and I said as much in my piece), Radeon 9700 often doesn't make you decide which you want, since you can often get both and playable frame rates.

Cheers,

Dave Salvator
Senior Technology Analyst
ExtremeTech
 
Back
Top