Mod, please lock, this is no longer on topic.

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
Okay then, but then you claim ATi's AF is broken even though AF has no spec.

Anisotropic Filtering has a definition- sampling in a square pattern doesn't hit that definition- in fact it is by definition the sole thing that makes it not anisotropic. Argue the English language on this one.

You basically claimed nVidia's shaders didn't really have a problem but it was the likes of Gabe and FutureMark involved in a giant conspiracy to make them look bad.

I never stated anything remotely like nVidia's shaders were not significantly slower then ATi's- I did state that Gabe was part of an *admitted* conspiracy against nVidia which has since been proven repeatedly. He was paid $6Million dollars to make ATi look good- that is public record. He claimed he was going to let independent reviewers and home users verify his results within a couple of weeks of shady day- he was a liar there too. You know you could point out that Core was on nV's payroll and was playing PR boy for them and you would be right- denying that Newell flat out lied time and time again to try and make ATi look good and nVidia look bad is delusional.

Sure, and on mathematical basis pre-rendering is superior to realtime rendering so are you going to stop playing games and just watch movies and cinematics instead?

Point filtering is faster then 16x AF. The point is that the current trend towards non filtering is regressive. What if MS introduced shader model 4.0 with a limit of four instructions per pixel and only 4bit per component accuracy- would you be talking about how much faster it is? Of course not, it would a moronic move aimed at lowering IQ for the sake of performance.

And again I'll ask which games you're playing right now and utilizing nVidia's "gift-from-the-gods" 8xAF? Tell me, if this scheme is so viable why has it gone the way of the Dodo?

Tell me which game SLId 7800GTXs couldn't handle 8xAF done properly on. If they offered it- I would be running them now.

That doesn't leave you with a lot of options then, huh? Either disable AF and go back to 1999 rendering quality, or stop playing games completely.

The GeForce was out in 1999- and its AF was certainly better at doing its job then anything released in the last couple of years. The problem is that we now have hardware that is more then powerful enough to do it the right way- they simply won't.

This shift is thanks to ATi's- 'who cares about IQ let's win some benches' mentality that has dragged the industry down. Not to let nVidia off the hook mind you, they have been all too willing to follow this time around which is a let down as they didn't slide down that slope in the 3Dfx days.

I challenge you to find any post where I praised this feature. I've slammed it multiple times and requested the likes of SSAA instead. T(emporal)AA is nothing more than a useless gimmick.

It's just like adaptive AF- only it tends to do more actual work.

and incidentally adding some shimmer depending on the settings you used.

Which games? I looked for it and didn't notice much difference at all(actually, I tended to spot a bit too much conservatism in their later drivers, not aliasing).

That's just utter nonsense. Name one other game where you can do something like using the left strafe key to go up (i.e towards the roof/sky).

D@mn near every space flight sim ever made.

Or how about rotating about x, y, z at the same time while moving backwards and upwards at a 45 degree angle?

Same as above, what on Earth do you think makes that any different from a typical space sim?

All child's play in the Descent series but basically impossible in any other game. Hell, most free-flight games like flight sims can't even do something as simple as flying backwards on the z axis.

Can you name me a singular space sim that doesn't? I haven't seen it that I can recall.
 

BFG10K

Lifer
Aug 14, 2000
22,709
3,002
126
Anisotropic Filtering has a definition- sampling in a square pattern doesn't hit that definition- in fact it is by definition the sole thing that makes it not anisotropic.
If they sampled in a square pattern they wouldn't be able to sharpen anything as the viewpoint from your eyes to the ground is non-square. If you're referring to the box shape then that's related to the fact that mip-maps are square and the fact that the card could only do bilinear AF. Look at the color mip patterns for a Ti4600 running bilinear AF and they'll be square as well.

I never stated anything remotely like nVidia's shaders were not significantly slower then ATi's- I did state that Gabe was part of an *admitted* conspiracy against nVidia which has since been proven repeatedly.
I'm not going through this again, however I do suggest you read some of the archived threads to refresh your memory, like I did yesterday.

Point filtering is faster then 16x AF
Sure, and it also looks like ass compared to 16xAF. It's also not comparible to 16xAF, unlike adaptive 16xAF which looks almost as good but runs far faster.

What if MS introduced shader model 4.0 with a limit of four instructions per pixel and only 4bit per component accuracy- would you be talking about how much faster it is?
That's completely different as that example is simply going backwards for no reason.

Current AF techniques look almost as good as the real thing but run a magnitude faster This is an acceptable trade-off, much like something like lossy compression for audio and video. Or are you against lossy compression too? Do you despise DVD movies as much as adaptive AF? Shall we also kill MP3s on portable players because it's not "real" audio? Or how about FM radio? Hell, even CDs can't compare to DVD-audio so no CDs for you either.

If we followed your logic we'd be living in an Amish paradigm with no technology as none of it is "real" and therefore it's not good enough to be used.

Tell me which game SLId 7800GTXs couldn't handle 8xAF done properly on. If they offered it- I would be running them now.
Such a setup would probably only be able to run NV25 style AF on the games that were top-end during the Ti4600 days. I think you're forgetting just how big the performance hit was.

I mean soft shadows (I guess the only "real" shadows by your definition) are only playable on such a setup @ 1024x768. If you're concerned about regression then you'd have a heart attack at that resolution as you'd be regressing back to the Voodoo2 days. And add "real" AF to those soft shadows and you'll probably be down to 640x480 on a 7800 GTX SLI setup. Wow, let's go back to 1980s VGA just to put ten pixels back into their correct places and get "real" AF. :roll:

The GeForce was out in 1999- and its AF was certainly better at doing its job then anything released in the last couple of years.
No it wasn't. It, for all intents and purposes, didn't have AF. And neither did the NV20 line either. NV25 had it but it was so slow you couldn't even use it. NV30 was slammed heavily but ironically was nVidia's first card to give their users long-overdue free AF.

This shift is thanks to ATi's- 'who cares about IQ let's win some benches' mentality that has dragged the industry down.
Back in the day most reviewers didn't even know what AF was, much less benchmark it. I don't think bechmarking was the reason to introduce it, ATi simply wanted to give their users another viable feature for their cards.

It's just like adaptive AF- only it tends to do more actual work.
It's nothing like adaptive AF. TAA increases workload and introduces a variety of limitations without actually significantly enhancing IQ at all (in some cases it degrades it because of flickering).

Which games?
Unreal and C&C Renegade seem to spring to my mind first. There were others but the details escape as soon thereafter I picked up my 9700 Pro and began enjoying free 16xAF in all titles which looked better than nVidia's hack-job version and almost as good as their "real" version.

Can you name me a singular space sim that doesn't? I haven't seen it that I can recall.
I don't think you understand my examples. Descent 3 allows full 360 degree freedom in full 360 degree space while putting you into an essentially FPS paradigm with the associated geometry and texturing to go with it. For that reason it's the best testbed around for AF.
 

ArchAngel777

Diamond Member
Dec 24, 2000
5,223
61
91
Originally posted by: BFG10K
Anisotropic Filtering has a definition- sampling in a square pattern doesn't hit that definition- in fact it is by definition the sole thing that makes it not anisotropic.
If they sampled in a square pattern they wouldn't be able to sharpen anything as the viewpoint from your eyes to the ground is non-square. If you're referring to the box shape then that's related to the fact that mip-maps are square and the fact that the card could only do bilinear AF. Look at the color mip patterns for a Ti4600 running bilinear AF and they'll be square as well.

I never stated anything remotely like nVidia's shaders were not significantly slower then ATi's- I did state that Gabe was part of an *admitted* conspiracy against nVidia which has since been proven repeatedly.
I'm not going through this again, however I do suggest you read some of the archived threads to refresh your memory, like I did yesterday.

Point filtering is faster then 16x AF
Sure, and it also looks like ass compared to 16xAF. It's also not comparible to 16xAF, unlike adaptive 16xAF which looks almost as good but runs far faster.

What if MS introduced shader model 4.0 with a limit of four instructions per pixel and only 4bit per component accuracy- would you be talking about how much faster it is?
That's completely different as that example is simply going backwards for no reason.

Current AF techniques look almost as good as the real thing but run a magnitude faster This is an acceptable trade-off, much like something like lossy compression for audio and video. Or are you against lossy compression too? Do you despise DVD movies as much as adaptive AF? Shall we also kill MP3s on portable players because it's not "real" audio? Or how about FM radio? Hell, even CDs can't compare to DVD-audio so no CDs for you either.

If we followed your logic we'd be living in an Amish paradigm with no technology as none of it is "real" and therefore it's not good enough to be used.

Tell me which game SLId 7800GTXs couldn't handle 8xAF done properly on. If they offered it- I would be running them now.
Such a setup would probably only be able to run NV25 style AF on the games that were top-end during the Ti4600 days. I think you're forgetting just how big the performance hit was.

I mean soft shadows (I guess the only "real" shadows by your definition) are only playable on such a setup @ 1024x768. If you're concerned about regression then you'd have a heart attack at that resolution as you'd be regressing back to the Voodoo2 days. And add "real" AF to those soft shadows and you'll probably be down to 640x480 on a 7800 GTX SLI setup. Wow, let's go back to 1980s VGA just to put ten pixels back into their correct places and get "real" AF. :roll:

The GeForce was out in 1999- and its AF was certainly better at doing its job then anything released in the last couple of years.
No it wasn't. It, for all intents and purposes, didn't have AF. And neither did the NV20 line either. NV25 had it but it was so slow you couldn't even use it. NV30 was slammed heavily but ironically was nVidia's first card to give their users long-overdue free AF.

This shift is thanks to ATi's- 'who cares about IQ let's win some benches' mentality that has dragged the industry down.
Back in the day most reviewers didn't even know what AF was, much less benchmark it. I don't think bechmarking was the reason to introduce it, ATi simply wanted to give their users another viable feature for their cards.

It's just like adaptive AF- only it tends to do more actual work.
It's nothing like adaptive AF. TAA increases workload and introduces a variety of limitations without actually significantly enhancing IQ at all (in some cases it degrades it because of flickering).

Which games?
Unreal and C&C Renegade seem to spring to my mind first. There were others but the details escape as soon thereafter I picked up my 9700 Pro and began enjoying free 16xAF in all titles which looked better than nVidia's hack-job version and almost as good as their "real" version.

Can you name me a singular space sim that doesn't? I haven't seen it that I can recall.
I don't think you understand my examples. Descent 3 allows full 360 degree freedom in full 360 degree space while putting you into an essentially FPS paradigm with the associated geometry and texturing to go with it. For that reason it's the best testbed around for AF.


Riiiiiiiiiiiiiiiiiiiiiight, because developers mode which are common in FPS games never allow things like "no clipping" and "fly" mode... Right? Uhhmm, no, nice try. Any FPS game can do what Descent 3 does, it just needs the right commands. Far Cry, for one, is a good example.

Edit ** I see that you are still referring to AF as "free"... I am guessing it will take you several years to fix that statement. Perhaps even longer, since you *still* refer to the GeForce 5XXX series daily on these boards.
 

ajaidev

Junior Member
Aug 30, 2005
16
0
0
I had an HERCULES 3DP 4500 with modefied heatsink and fan.... and at that time 175 Mhz was big for me.... And the new TBR feature was like holy water... Playing Quake3 on it was better than anything i had found....!!!!
 

blckgrffn

Diamond Member
May 1, 2003
9,676
4,309
136
www.teamjuchems.com
:laugh:


*Knows better than to jump in here. Grabs long stick, marshmellows & hotdogs, and cooler full of beer to enjoy the show :p lol, next gen console CPU debate, anyone?
 

BFG10K

Lifer
Aug 14, 2000
22,709
3,002
126
Riiiiiiiiiiiiiiiiiiiiiight, because developers mode which are common in FPS games never allow things like "no clipping" and "fly" mode
You make a good point but even in those modes you typically can't do something like fly upside down or roll around the z axis.
 

Stretchman

Golden Member
Aug 27, 2005
1,065
0
0
A friend of mine used to have a "Hercules" branded Kyro card. I forget the actual model type, but he used it to play Quake, Master of Orion and Deus Ex. At the time it was much less expensive then offerings from Nvidia and could be readil store bought at CompUSA.

This was at a time when we both didn't have much computer knowledge, but his reason for purchasing it was a review he read online. In the review, it was stated that the Kyro had a unique way of rendering graphics in real-time. To conserve horsepower (or pixel power, hehe) it actually wouldn't render anything that wasn't visible to the user onscreen, focusing only on what was immediately present in the visual field.

If I remember correctly, this was a unique approach (it may have been called Tiling) that wasn't being used by non-Kyro cards.
 

irwincur

Golden Member
Jul 8, 2002
1,899
0
0
Had an old Hercules 4500. In my opinion it was a great card and served me well until I bought an ATI 8500. It was dirt cheap when I bought it too.
 

reever

Senior member
Oct 4, 2003
451
0
0
Originally posted by: BFG10K
Riiiiiiiiiiiiiiiiiiiiiight, because developers mode which are common in FPS games never allow things like "no clipping" and "fly" mode
You make a good point but even in those modes you typically can't do something like fly upside down or roll around the z axis.


I'm pretty sure you can do this in Serious Sam, and with colored mipmaps too
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
If they sampled in a square pattern they wouldn't be able to sharpen anything as the viewpoint from your eyes to the ground is non-square.

They couldn't do it properly- which oddly enough they don't.

I'm not going through this again, however I do suggest you read some of the archived threads to refresh your memory, like I did yesterday.

I did, your point? Perhaps you can provide a link because I failed to find anything like what you are trying to indicate.

It's also not comparible to 16xAF, unlike adaptive 16xAF which looks almost as good but runs far faster.

You have to be d@mn near blind to think that- it isn't remotely in the same league. Perhaps you should go back and check them out side by side- one is horrific in comparison.

That's completely different as that example is simply going backwards for no reason.

BS- that is faster framerates at the sake of significant degredation to IQ- exactly what you are so fond of.

Current AF techniques look almost as good as the real thing

They aren't remotely close to looking as good- nothing resembling being close.

Or are you against lossy compression too?

Absolutely. All of my music is encoded in lossless format on my HD- explodes the file size but I don't mind spending a couple dollars for a larger storage space instead of listening to the god awful crap like 320 MP3 rips.

Do you despise DVD movies as much as adaptive AF?

Not as much, but I'm certainly no fan of the compression artifacts that MPEG2 introduces.

Shall we also kill MP3s on portable players because it's not "real" audio?

It's very low quality audio. It has its place. Much as I wouldn't expect perfect AF on a portable 3D device as in certain cases serious trade offs have to be made. I think that current adaptive AF implementations would be tollerable for cell phones as an example.

Or how about FM radio?

Never listen to it. I'll buy DVD-Audio when available, CD Audio when its not. All of the FM receivers in my house are disconnected except for in the kid's rooms and my car's stereo shuts off when I take a CD out- much better then listening to the insanely poor quality of open air transmissions.

If we followed your logic we'd be living in an Amish paradigm with no technology as none of it is "real" and therefore it's not good enough to be used.

We are talking about proper anisotropic filtering which has been prove can be done properly years ago- you are just too much a fan of bench charts at the expense of IQ to accept this.

I think you're forgetting just how big the performance hit was.

No, I'm not. I want you to tell me which games the 7800GTX in SLI couldn't handle AF on say running 1600x1200. I alreay have looked over all the benches and am quite familiar with how horribly CPU limited that setup is. You tell me which games it couldn't handle.

I mean soft shadows (I guess the only "real" shadows by your definition) are only playable on such a setup @ 1024x768.

Actually 'real' shadows would be radiosity based- we haven't seen anything remotely close to that yet. I am not asking for off line pre rendered perfection before it is possible. The thing is that right now we have already seen proper AF years ago and it worked just fine. It was performance limited to older titles and lower resolutions just as AA was including MSAA when it came on to the market. Time marches on and we have significantly lowered out standards for acceptable IQ because of people like you. Benchmarks first and second who cares about how bad the IQ is.

No it wasn't. It, for all intents and purposes, didn't have AF.

I have mathematical reality on my side- it is not arguable.

NV30 was slammed heavily but ironically was nVidia's first card to give their users long-overdue free AF.

It's not free on any part with any remotely demaning game, it is just a lesser performance hit.

Back in the day most reviewers didn't even know what AF was, much less benchmark it. I don't think bechmarking was the reason to introduce it, ATi simply wanted to give their users another viable feature for their cards.

I recall seeing tons of benches comparing the R100's AF to the NV10/15's.

increases workload and introduces a variety of limitations without actually significantly enhancing IQ at all

Just like adaptive AF. The difference is that adaptive AF changes things from too blurry to too aliased- TAA is working on reducing not exploding aliasing.

Unreal and C&C Renegade seem to spring to my mind first. There were others but the details escape as soon thereafter I picked up my 9700 Pro and began enjoying free 16xAF in all titles which looked better than nVidia's hack-job version and almost as good as their "real" version.

I've gone back and looked over all the benches I could find and everything I see very strongly disagrees with what you are saying. The performance hit is huge using 8x AF, not 2x.

I don't think you understand my examples. Descent 3 allows full 360 degree freedom in full 360 degree space while putting you into an essentially FPS paradigm with the associated geometry and texturing to go with it. For that reason it's the best testbed around for AF.

What about Forsaken(for something obvious)? That is nigh exactly what Descent3 is in terms of gameplay- although there are so many titles where you can do exactly what you can in Descent3 it probably is close to if not crossing four figures worth.
 

BFG10K

Lifer
Aug 14, 2000
22,709
3,002
126
They couldn't do it properly- which oddly enough they don't.
Then neither did the NV2x since it had the same square pattern when running bilinear AF.

You have to be d@mn near blind to think that- it isn't remotely in the same league.
Yes Ben, perhaps the whole world is blind. Except for you of course.

BS- that is faster framerates at the sake of significant degredation to IQ- exactly what you are so fond of.
Are you trying to be intentionally obtuse? Your childish example is quite frankly ludicrous.

Absolutely. All of my music is encoded in lossless format on my HD
Great, now how about the fact that CDs themselves are compressed to begin with? And how about the fact that CDs are mathematically inferior to a live source?

Not as much, but I'm certainly no fan of the compression artifacts that MPEG2 introduces.
Yet you continue to use DVD movies and no doubt VHS before they were avaialble, yet I don't hear you screaming about the issue each time you watch a movie like you do when AF is discussed.

Much as I wouldn't expect perfect AF on a portable 3D device as in certain cases serious trade offs have to be made.
I wouldn't expect perfect AF anywhere with that kind performance hit, except maybe for pre-rendering.

I'll buy DVD-Audio when available
Except it hasn't been available until now so what have you been doing all these years? Did you only recently start listening to music or something?

We are talking about proper anisotropic filtering which has been prove can be done properly years ago- you are just too much a fan of bench charts at the expense of IQ to accept this.
I'm not a fan of bench charts, I'm a fan of playable speeds at high resolutions in modern games. You however appear to be a fan of running ancient games at VGA resolutions.

No, I'm not. I want you to tell me which games the 7800GTX in SLI couldn't handle AF on say running 1600x1200.
1600x1200 now is it? That's a middling/low resolution for you isn't it? To quote yourself, "you'd have to d@mn blind".

I am not asking for off line pre rendered perfection before it is possible
Yes you are. NV2x AF wasn't viable so the industry moved away from it.

The thing is that right now we have already seen proper AF years ago and it worked just fine.
It only worked fine at VGA resolutions on ancient games so "you'd have to be d@mn blind" to use it.

I have mathematical reality on my side- it is not arguable.
Just like mathematical reality is on my side and states live music and live movie sets are more accurate than any recording, so I guess you'd have to be "d@mn blind" and "d@mn deaf" to use any recorded media then.

I recall seeing tons of benches comparing the R100's AF to the NV10/15's.
I don't.

Just like adaptive AF.
Adaptive AF increases workload? What on Earth are you talking about? It reduces workload significantly while provided almost identical IQ to "real" AF.

I've gone back and looked over all the benches I could find and everything I see very strongly disagrees with what you are saying
I didn't ask you to look at benches, I asked you to look for shimmering when nVidia's optimizations were enabled. I assume you can see them, not being d@mn blind and all.
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
Then neither did the NV2x since it had the same square pattern when running bilinear AF.

Where exactly? The mip transitions were hard lined, not boxed. Big difference.

Great, now how about the fact that CDs themselves are compressed to begin with? And how about the fact that CDs are mathematically inferior to a live source?

Absolutely correct- and when I see a superior source that can more closely replicate the original music I buy it. Of course- you must realize that the source for an enormous amount of existing music was recorded in analog and is clearly inferior to what CDs can replicate- to make the matter worse I don't think I'm ever going to catch a live Led Zeppelin performance in the future as an example.

Great, now how about the fact that CDs themselves are compressed to begin with?

Trunctuated- not compressed. I would expect you to realize the difference.

And how about the fact that CDs are mathematically inferior to a live source?

As I have stated numerous times- when given a superior alternative I will take it.

Yet you continue to use DVD movies and no doubt VHS before they were avaialble

LaserDisc prior to DVDs- show me a superior source available now. I'm waiting on Blu-Ray anxiously- but that hasn't already been widely available in millions of consumer devices and then I was forced back to deal with complete crap.

yet I don't hear you screaming about the issue each time you watch a movie

Ask my wife. She can explain in great detail the finer nuances of MPEG2 artifacts and why we need improvements along with being able to explain to you how the decode hardware in them can amplify the existing problems not to mention how much color data is lost due to the poor compression standard. She can also tell you why it is that the entire generation of 720p/1080i displays was a farce and why we need to move straight to 1080p with the highest available medium storage possible to use the least agressive compression ratios. Take a wild guess why she knows all this.

Except it hasn't been available until now so what have you been doing all these years?

I bought the best available technology. I didn't get a taste of DVD-A then have to revert to horribly worn eight tracks. That is what you are asking.

1600x1200 now is it? That's a middling/low resolution for you isn't it?

I know that you have a low resolution monitor but I don't recall where exactly it caps- is 1920x1440 or ever lower? Change it up to 2048x1536- no AA- and tell me all the titles SLId 7800GTXs couldn't handle with proper AF.

NV2x AF wasn't viable so the industry moved away from it.

Neither was its AA, and the industry stuck with it until it had the power to handle it. For that matter- the superior SSAA that MSAA replaced was nigh dropped until they developed enough power and then they brought it back as it was superior larger performance hit or not.

It only worked fine at VGA resolutions on ancient games so "you'd have to be d@mn blind" to use it.

Final Fantasy The Spirits Within looks a lot better then any PC game I've played even on the incredibly poorly compressed DVD format. Obviously it will look much better on Blu-Ray(I'll be picking that one up quite quickly) but the reality is that increasing ugly rendering only gives you sharper quality ugly rendering.

Just like mathematical reality is on my side and states live music and live movie sets are more accurate than any recording, so I guess you'd have to be "d@mn blind" and "d@mn deaf" to use any recorded media then.

Given a choice I hit iMax, movie sets don't work quite right as the movie kind of has to be finished. If you mean I choose live performances over recorded the answer is of course- what kind of person wouldn't? Given that it isn't viable to see live performances all of the time due to logistics I take the best medium available and I don't step back in anything outside of PC graphics unfortunately.

Adaptive AF increases workload? What on Earth are you talking about? It reduces workload significantly while provided almost identical IQ to "real" AF.

Temporal AA only increased the workload versus non corrected images. Not versus the comparable peak quality settings it offers.

I didn't ask you to look at benches, I asked you to look for shimmering when nVidia's optimizations were enabled. I assume you can see them, not being d@mn blind and all.

On my GF4? Nope, can't see them. Which driver revision should I be using?
 

ArchAngel777

Diamond Member
Dec 24, 2000
5,223
61
91
Originally posted by: BenSkywalker
Then neither did the NV2x since it had the same square pattern when running bilinear AF.

Where exactly? The mip transitions were hard lined, not boxed. Big difference.

Great, now how about the fact that CDs themselves are compressed to begin with? And how about the fact that CDs are mathematically inferior to a live source?

Absolutely correct- and when I see a superior source that can more closely replicate the original music I buy it. Of course- you must realize that the source for an enormous amount of existing music was recorded in analog and is clearly inferior to what CDs can replicate- to make the matter worse I don't think I'm ever going to catch a live Led Zeppelin performance in the future as an example.

Great, now how about the fact that CDs themselves are compressed to begin with?

Trunctuated- not compressed. I would expect you to realize the difference.

And how about the fact that CDs are mathematically inferior to a live source?

As I have stated numerous times- when given a superior alternative I will take it.

Yet you continue to use DVD movies and no doubt VHS before they were avaialble

LaserDisc prior to DVDs- show me a superior source available now. I'm waiting on Blu-Ray anxiously- but that hasn't already been widely available in millions of consumer devices and then I was forced back to deal with complete crap.

yet I don't hear you screaming about the issue each time you watch a movie

Ask my wife. She can explain in great detail the finer nuances of MPEG2 artifacts and why we need improvements along with being able to explain to you how the decode hardware in them can amplify the existing problems not to mention how much color data is lost due to the poor compression standard. She can also tell you why it is that the entire generation of 720p/1080i displays was a farce and why we need to move straight to 1080p with the highest available medium storage possible to use the least agressive compression ratios. Take a wild guess why she knows all this.

Except it hasn't been available until now so what have you been doing all these years?

I bought the best available technology. I didn't get a taste of DVD-A then have to revert to horribly worn eight tracks. That is what you are asking.

1600x1200 now is it? That's a middling/low resolution for you isn't it?

I know that you have a low resolution monitor but I don't recall where exactly it caps- is 1920x1440 or ever lower? Change it up to 2048x1536- no AA- and tell me all the titles SLId 7800GTXs couldn't handle with proper AF.

NV2x AF wasn't viable so the industry moved away from it.

Neither was its AA, and the industry stuck with it until it had the power to handle it. For that matter- the superior SSAA that MSAA replaced was nigh dropped until they developed enough power and then they brought it back as it was superior larger performance hit or not.

It only worked fine at VGA resolutions on ancient games so "you'd have to be d@mn blind" to use it.

Final Fantasy The Spirits Within looks a lot better then any PC game I've played even on the incredibly poorly compressed DVD format. Obviously it will look much better on Blu-Ray(I'll be picking that one up quite quickly) but the reality is that increasing ugly rendering only gives you sharper quality ugly rendering.

Just like mathematical reality is on my side and states live music and live movie sets are more accurate than any recording, so I guess you'd have to be "d@mn blind" and "d@mn deaf" to use any recorded media then.

Given a choice I hit iMax, movie sets don't work quite right as the movie kind of has to be finished. If you mean I choose live performances over recorded the answer is of course- what kind of person wouldn't? Given that it isn't viable to see live performances all of the time due to logistics I take the best medium available and I don't step back in anything outside of PC graphics unfortunately.

Adaptive AF increases workload? What on Earth are you talking about? It reduces workload significantly while provided almost identical IQ to "real" AF.

Temporal AA only increased the workload versus non corrected images. Not versus the comparable peak quality settings it offers.

I didn't ask you to look at benches, I asked you to look for shimmering when nVidia's optimizations were enabled. I assume you can see them, not being d@mn blind and all.

On my GF4? Nope, can't see them. Which driver revision should I be using?


Ok, I am calling a time out between you guys. This is getting ridiculous. Both of you have made some ridiculous statements. Such as a "low resolution monitor being 19** X 14**", such as "free AF" and etc... It is time to stop being such an extremist. You guys could argue forever and no one is gaining ground.

Learn to disengage.

For the record, 1600 X 1200 is no middling resolution. In addition to that, 19** X 14** is not a "low resolution". You both keep making statements that are more ludicrous than the next. All for the sake of arguements.

 

xtknight

Elite Member
Oct 15, 2004
12,974
0
71
It does suck that AA/AF quality has been "optimized" as of late. Video card enthusiasts like myself get bored when we're stuck with a lowly 1280x1024 LCD...I need something to challenge my video card. No, I can't afford anything bigger, likely. No income, just birthdays and holidays. Doesn't it suck being young?
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
is getting ridiculous. Both of you have made some ridiculous statements. Such as a "low resolution monitor being 19** X 14**"

Top resolution displays are pushing 2560x1600 for consumer displays, 1920x1440 is well down on the resolution scale from there. 1920x1440 for that matter barely edges out TVs as of now, do you consider TVs high resolution....?

You guys could argue forever and no one is gaining ground.

BFG and I have been arguing since prior to you registering on this board, and more then likely we will still be arguing when you have stopped posting here. We agree on roughly 99% of issues- the 1% we don't agree on we like to argue until they are no longer relevant, and even then we will bring those up in retrospect :D

No income, just birthdays and holidays. Doesn't it suck being young?

Get a part time job- you will have more disposable income being young enough to still be in HighSchool and having a part time job then you likely will for the next six or seven years or so.
 

ArchAngel777

Diamond Member
Dec 24, 2000
5,223
61
91
Originally posted by: BenSkywalker
is getting ridiculous. Both of you have made some ridiculous statements. Such as a "low resolution monitor being 19** X 14**"

Top resolution displays are pushing 2560x1600 for consumer displays, 1920x1440 is well down on the resolution scale from there. 1920x1440 for that matter barely edges out TVs as of now, do you consider TVs high resolution....?

You guys could argue forever and no one is gaining ground.

BFG and I have been arguing since prior to you registering on this board, and more then likely we will still be arguing when you have stopped posting here. We agree on roughly 99% of issues- the 1% we don't agree on we like to argue until they are no longer relevant, and even then we will bring those up in retrospect :D

No income, just birthdays and holidays. Doesn't it suck being young?

Get a part time job- you will have more disposable income being young enough to still be in HighSchool and having a part time job then you likely will for the next six or seven years or so.

I am glad you both are willing to waste your short time on earth going back and forth for what seems like the rest of your life. Part of becoming an adult is stating your position and disengaging from it. Hence, this is my last post here on this thread.
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
Part of becoming an adult is stating your position and disengaging from it.

Actually, that is more the stance of someone with cowardice tendencies. If you were to some day make it into an 'adult' school and study up a bit on the majority of the greatest 'adults' ever to live they were the ones who would battle endlessly in one manner or another either to make their point or to complete their tasks. You will find this throughout history, science, philosophy and even mathematics in some instances. From Ghandi to Genghis Khan, from Thales to Hawking- the ability to engage and remain that way until your point is heard it what seperates the great minds from the should have beens.

William James Sidis was excellent at disengaging- Einstein was not.
 

ArchAngel777

Diamond Member
Dec 24, 2000
5,223
61
91
Originally posted by: BenSkywalker
Part of becoming an adult is stating your position and disengaging from it.

Actually, that is more the stance of someone with cowardice tendencies. If you were to some day make it into an 'adult' school and study up a bit on the majority of the greatest 'adults' ever to live they were the ones who would battle endlessly in one manner or another either to make their point or to complete their tasks. You will find this throughout history, science, philosophy and even mathematics in some instances. From Ghandi to Genghis Khan, from Thales to Hawking- the ability to engage and remain that way until your point is heard it what seperates the great minds from the should have beens.

William James Sidis was excellent at disengaging- Einstein was not.

I was done with this thread until you had to post this garbage. You quote
From Ghandi to Genghis Khan, from Thales to Hawking- the ability to engage and remain that way until your point is heard it what seperates the great minds from the should have beens.

You have posted on this subject how many times in this thread? You were heard the first time. Yet you continue to beat a dead horse over and over and over and over. Insanity is defined as "Someone who does or says the same thing over and over again, yet expects different results". Now, we have heard you reply to BFG10K about 10 times in this thread and what do you hope to accomplish? Additionally you are putting yourself in the same class as Einstien! That is absurd and laughable.

The ability to disengage comes from a radically different world view. You are willing to fight about image quality while the rest of the world does more important things. A discussion is one thing, but you making your entire life work to discuss image quality does nothing, absolutely nothing for humanity.

To sum it up, people who defend tooth and nail on important matters are the true heroes. This is in contrast to those who defend tooth and nail over trivial non-essential matters that just don't get it. You may have a great mind, I do not doubt that, but to waste it, that is the true crime.


This thread is now completely off topic due to several people, including myself hi-jacking it. We can either let this thread die, or get a mod to close it.

 

vshah

Lifer
Sep 20, 2003
19,003
24
81
Originally posted by: Schadenfroh
/recalls the tile based rendering kyro fanboys versus geforce flamewars
/shutters at the thought

:) good times....i was a lurker back then

i still have my old geforce 2 ultra
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
You were heard the first time. Yet you continue to beat a dead horse over and over and over and over.

If you are reading the discussion and comprehending it you would realize that it evolves as most conversations do when there is discourse.

Insanity is defined as "Someone who does or says the same thing over and over again, yet expects different results".

In what language? In English that isn't even close to the definition. The inability to comprehend the nature and consequences of one's actions; or a serious mental disability are what the word means in English.

Using your definition Ghandi was insane by expecting different results doing the same thing over and over- although that is exactly what ended up happening.

Additionally you are putting yourself in the same class as Einstien! That is absurd and laughable.

It is pointing out the erroneous assertion by you that disengaging is somehow a more mature stance then continuing discourse until a full comprehension of the opponents view can be understood. All of the great individuals throughout history have done the exact opposite. Also if you are under the impression that I am comparing myself to Einstein then that would indicate that I would be comparing you to William James Sidis(look it up). That is an incorrect assertion but one that you will find would certainly not be an insult were it true.

The ability to disengage comes from a radically different world view. You are willing to fight about image quality while the rest of the world does more important things.

If you pay close attention then you will find BFG and I don't have radically different 'world views' at least as you use the term for this particular topic. When it comes to the subject at hand we agree on so many fundamental issues that we tend to spend enormous amounts of time in discourse on the small areas we do not agree on.

A discussion is one thing, but you making your entire life work to discuss image quality does nothing, absolutely nothing for humanity.

I can assure you that my posting disagreements in this thread to BFG once a day is time far better spent then watching the latest sitcom.

To sum it up, people who defend tooth and nail on important matters are the true heroes.

And nowhere will you find either of us pretending that we are trying to do anything of great importance in the global sense- you are talking about a singular post per day by each of us. A few minutes worth of typing.

This thread is now completely off topic due to several people, including myself hi-jacking it. We can either let this thread die, or get a mod to close it.

Why would a mod close it? There is no flame war going on here- I have far too much respect for BFG to partake in such behaviour towards him(I may tongue in cheek insinuate he is blind, much as he does to me, but that is about it). I have responded to all the points brought up about PowerVR and the conversation has evolved from that point. There is no reason to close this thrad.
 

BFG10K

Lifer
Aug 14, 2000
22,709
3,002
126
The mip transitions were hard lined, not boxed. Big difference.
Not that big. This isn't like a simple LOD adjustment like you initially claimed it was.

Of course- you must realize that the source for an enormous amount of existing music was recorded in analog and is clearly inferior to what CDs can replicate
I don't need to realize anything; you need to realize that you aren't practicing what you preach.

Trunctuated- not compressed. I would expect you to realize the difference.
In a nutshell reduced quality which is mathematically inferior to a live source. Again if you practiced what you preached you wouldn't have any recorded media at all. Unless you're d@mn deaf that is.

As I have stated numerous times- when given a superior alternative I will take it.
Just like you're taking a Ti4600's 8xAF? Tell me Ben, in what games are you "taking this alternative"?

I know that you have a low resolution monitor but I don't recall where exactly it caps- is 1920x1440 or ever lower?
What my monitor caps at has nothing to do with your backpedaling from 1600x1200 being a low resolution to using it as a valid example for "real AF". If you want to be obtuse as least be consistent.

Change it up to 2048x1536- no AA- and tell me all the titles SLId 7800GTXs couldn't handle with proper AF.
Why no AA? You must be d@mn blind not to notice jagged edges at 2048x1536. I love it how you claim how viable "real" AF is but you continue to present ridiculous examples (1600x1200, no AA) to come up with proof for your claims.

the superior SSAA that MSAA replaced was nigh dropped until they developed enough power and then they brought it back as it was superior larger performance hit or not.
Developed enough power? We've seen 8xAA and 16xAA benchmarks on a 7800 GTX SLI at 2048x1536 and I can assure there isn't enough power to run it. Again it's a good option for older games but to advocate dropping MSAA because you'd have to be d@mn blind to use it is quite foolish. If we dropped MSAA you'd either be using SSAA with low resolution mush or not getting any AA at all (in either case you'd need to be d@mn blind).

On my GF4? Nope, can't see them. Which driver revision should I be using?
Maybe you're just d@mn blind then? :p

This is why these discussions never go anywhere. Now it doesn't even matter now that it's not "real" AF when those optimizations are enabled, because you can't see them they don't count so therefore it's still "real".
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
Not that big. This isn't like a simple LOD adjustment like you initially claimed it was.

Very close to it- and yes there is a huge difference between proper AF and what the R100 was doing- there is a huge difference between what the R100 was doing versus the R200 for that matter.

I don't need to realize anything; you need to realize that you aren't practicing what you preach.

I want to use the best available method of reproduction- how is that not practicing what I preach?

In a nutshell reduced quality which is mathematically inferior to a live source.

That would depend on what the live source was created with. If they are using equiptment that has a lower S/N ration and inferior balancing to what a CD can produce then a CD recording from a studio would be more accurate. Now if you are talking about the artists using optimal equiptment I can speak from an awful lot of experience saying that they don't always do that for live performances, not even close for most artists actually.

Just like you're taking a Ti4600's 8xAF? Tell me Ben, in what games are you "taking this alternative"?

You are excluding every other factor in your tangent here while they are relevant. I have not criticized the progress in other aspects that newer parts have made- if I could purchase a Ti4999 that offered performance comparable to a 7800GTX(although with a larger hit for AF) I would. Given that I can't- I have to try and find a part that can play upcoming games with the best quality possible. Unfortunately that appears to mean I have to deal with another reduction in IQ for basic filtering.

What my monitor caps at has nothing to do with your backpedaling from 1600x1200 being a low resolution to using it as a valid example for "real AF". If you want to be obtuse as least be consistent.

Try and find benches north of 16x12 on most sites- I'd be glad if they dropped everything but 20x15 but that isn't going to happen. I can't find any benches of no AF 2048xx1536 to offer as comparison.

Why no AA?

Resolution- filtering- AA- I use the same standards you do there. Increasing the resolution works as a function comparable(not in terms of functionality but in terms of output) to AF as the increased pixel density impacts the LOD such to increase the clarity of the textures a considerable amount. AA chews up bandwidth which is needed for proper AF(moreso then GPU time with current parts).

Developed enough power? We've seen 8xAA and 16xAA benchmarks on a 7800 GTX SLI at 2048x1536 and I can assure there isn't enough power to run it.

But there is to run nV's TAA modes and ATi is also getting into the game of offering SSAA modes at some point(supposedly when Crossfire ships). There is also plenty enough power to run 4x SSAA on a 7800GTX SLI setup.

Now it doesn't even matter now that it's not "real" AF when those optimizations are enabled, because you can't see them they don't count so therefore it's still "real".

That is why I asked which driver revision? I don't tend to update drivers on my nV machines very often- they don't need it. Perhaps it is something in driver revisions later then what I am using now in my GF4 rig(seriously outdate, XP2100, 512MBs- 31.80 are the Dets currently installed).
 

BFG10K

Lifer
Aug 14, 2000
22,709
3,002
126
Very close to it- and yes there is a huge difference between proper AF and what the R100 was doing- there is a huge difference between what the R100 was doing versus the R200 for that matter.
Sure there was a difference, just like there was a big difference between a LOD adjustment and what the R100 was doing. The R100 offered 64 tap anisotropic for near-zero performance hit and that was revolutionary at that time, especially since nVidia's 16 tap method barely had any affect on IQ at all and the performance hit was shocking.

I want to use the best available method of reproduction- how is that not practicing what I preach?
That would depend on what the live source was created with.
By real actors and real props at a live movie set. Since this real life scenario is mathematically superior to a lossy DVD you are not practicing what you preach since you've admitted you both own and watch DVD movies.

You are excluding every other factor in your tangent here while they are relevant.
No, I am focusing solely on AF performance and how optimizations have resulted in almost no IQ degradation while maintaining near-zero performance hit. You for some reason find that unacceptable and would rather go without. Ironic really - someone who doesn't like optimized AF will gladly take the blurry trilinear mess or run at VGA resolutions instead.

I can't find any benches of no AF 2048xx1536 to offer as comparison.
Here you go - near zero performance hit at 16xAF courtesy of optimized AF. Why anyone would want to go backwards to 2xAF cutting their framerate in half and providing essentially zero IQ gain is beyond me. It's like dropping cars for horses because horses are "real".

Resolution- filtering- AA- I use the same standards you do there.
Not quite. I keep 16xAF on all the time, then I crank the resolution, and then AA when I've maxed out the resolution. The issue here is that I sure as hell wouldn't want to go to a middling resolution like 1600x1200 with 16xAF and 0xAA to get "real" AF when I can get something like 1920x1440, 16xAF and 4xAA with optimized AF, with better performance as well no doubt.

But there is to run nV's TAA modes and ATi is also getting into the game of offering SSAA modes at some point(supposedly when Crossfire ships).
nVidia's TAA modes are not the same as straight SSAA. SSAA is hideously inefficient because it uses brute force to blindly AA everything but it really only makes a difference in alpha texture situations. nVidia's TAA is like an optimized version of SSAA so I guess you can't class that as "real" AA either and therefore you won't be using it.

Regardless, full SSAA isn't viable on even SLI/Crossfire setups and if we had been following your standards we'd be in the dark ages with no AA since MSAA isn't "real" AA according to you. Of course that would be to the detriment of everyone since alpha textures aren't really an issue in most cases and MSAA is both really fast and does wonders for the core problem of jagged edges.

There is also plenty enough power to run 4x SSAA on a 7800GTX SLI setup.
Sure, in old games. Again benchmarks show 8xAA and 16xAA are not viable in new games, even on a 7800 SLI setup. If we had been following your standards we'd have no AA at all or we'd be running SSAA at slideshow low resolution mush.

That is why I asked which driver revision?
What difference does that make? If they're reducing texture samples on stages that's mathematically inferior to "real" AF and therefore it's unacceptable, right?

I'm not sure why you're putting the "it looks good to me" spin in it since your objection to this very comment is the sole basis of this whole thread with regards to real AF.

Likewise you claim 8 bit filtered AF is "real". If however somebody were to claim that 16 or 32 bit filtered AF is "real" and 8 bit is garbage because it's mathematically inferior to the previous two I'm sure you'd put some "8 bit looks good to me" spin on it as well.
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
The R100 offered 64 tap anisotropic for near-zero performance hit and that was revolutionary at that time, especially since nVidia's 16 tap method barely had any affect on IQ at all and the performance hit was shocking.

The R100 didn't actually offer 64 tap anisotropic filtering. It offered a filtering hack.

By real actors and real props at a live movie set. Since this real life scenario is mathematically superior to a lossy DVD you are not practicing what you preach since you've admitted you both own and watch DVD movies.

You even quoted me-

I want to use the best available method of reproduction

No, I am focusing solely on AF performance and how optimizations have resulted in almost no IQ degradation

I consider considerably increased aliasing and misaligned LOD selection on adjacent textures very pronounced IQ reduction.

Here you go - near zero performance hit at 16xAF courtesy of optimized AF.

Depends on the particular game- Pacific Fighters is showing ~10% performance hit.

Not quite. I keep 16xAF on all the time, then I crank the resolution, and then AA when I've maxed out the resolution. The issue here is that I sure as hell wouldn't want to go to a middling resolution like 1600x1200 with 16xAF and 0xAA to get "real" AF when I can get something like 1920x1440, 16xAF and 4xAA with optimized AF, with better performance as well no doubt.

That is what you would like to do, not me.

nVidia's TAA is like an optimized version of SSAA so I guess you can't class that as "real" AA either and therefore you won't be using it.

Haven't had enough hands on time with it to make a call there- but it obviously is much better then the poor hack of MSAA.

Regardless, full SSAA isn't viable on even SLI/Crossfire setups

BS.

Sure, in old games. Again benchmarks show 8xAA and 16xAA are not viable in new games, even on a 7800 SLI setup.

Using that logic none of the current parts can run 256x AF so obviously they can't run AF at all. I mentioned 4x SSAA- and SLId GTXs have plenty of power for it. Crossfire is still vaporware so who knows about that.

What difference does that make? If they're reducing texture samples on stages that's mathematically inferior to "real" AF and therefore it's unacceptable, right?

If it changes the ouputted pixel yes. I want to see it changes the outputted pixel or not. Without HDR support trunctuating non visible data can result in the same mathematical output.

I'm not sure why you're putting the "it looks good to me" spin in it since your objection to this very comment is the sole basis of this whole thread with regards to real AF.

Likewise you claim 8 bit filtered AF is "real". If however somebody were to claim that 16 or 32 bit filtered AF is "real" and 8 bit is garbage because it's mathematically inferior to the previous two I'm sure you'd put some "8 bit looks good to me" spin on it as well.

It depends on the accuracy of the drawn pixel- Trunctuating non visible data isn't an issue. When you are off by several color levels on the outputted pixel and your LOD selection is erratic across side by side textures then you have serious issues.