My R9700 experince and Aniso still "flawd"?
- Thread starter jb
- Start date
Oratius said:
Look, clearly there really isn't a problem with ATIs AF at all. As far as the IQ goes it was just an observation besides AF effects IQ so in effect it really isn't OT. IMHO the topic of discusstion that Radeon 9700 AF being "flawed" is a poor observation(at best) from what I can tell. Clearly ATIs Radeon 9700 is a far better solution for the high end gamer any who would suggest otherwise are attempting to misslead others opinions. Of course that is just MO.
Mize said:
First of all I haven't said anything about my opinion on the matter, i've only said that during a discussion the better thing is knowing exactly which is the topic.
Besides, when I said to search which is the better implementation I was speaking simply from a technological point of view, because I'd like to know if in the same condition they offer the same IQ; but be careful, I'm not saying that we should judge which card is better comparing them in the same condition, it would be stupid, exactly as you've said.
Galilee said:
On that point, is this only something that occurs exactly @ 45 degrees? What if it is at 43 degrees or 47 degrees? My point here is simple. A person in a FPS moving around quickly spinning looking for targets would never see this issue without looking for it specifically. If this is the case why is such a matter made about it? IMHO the subject is ridiculous.
I only assumed that 45 degrees were the worst angle. There might be a big difference between 30 and 45 degrees.
But I have to agree, noone should and would notice this. Same reason I use the "optimized" aniso in Rivatuner on my GF4. Can't see the difference, but the performance sure is better
But I have to agree, noone should and would notice this. Same reason I use the "optimized" aniso in Rivatuner on my GF4. Can't see the difference, but the performance sure is better
Galilee said:
Sorry, I fail to understand why this is a ‘bug’. To me a bug suggests something that occurs erroneously, outside of its design specifications. Where is the evidence that states that this is not within the design specifications? The idea that ‘because x does it this way and y does not ergo y has a bug’ is slightly absurd – they have different algorithms and hence work differently and may produce different results but in no way does that suggest to me that this is a ‘bug’. Now, if ATI were to come out and say “Oh, well we aimed for this…. but in fact the hardware is doing this….” Then you could say ‘it’s a bug’!
However, going back the hypothetical if hardware y had a greater degree of anisotropy and its outputted IQ never falls below that of hardware x then personally in terms of IQ I would say that y has a better hardware implementation of Anisotropic filtering in terms of quality because it has the ability to always look better.
Galilee said:I only assumed that 45 degrees were the worst angle. There might be a big difference between 30 and 45 degrees.
But I have to agree, noone should and would notice this. Same reason I use the "optimized" aniso in Rivatuner on my GF4. Can't see the difference, but the performance sure is better
Thanks for that. But is it as finicky as too only occur @ 45 degrees? What about a 1-2 degree difference? Is this matter even noteworthy? Or is it nit-picking about a matter that few would ever realize unless it was conveniently brought to their attention?
Geez
I had a funny feeling that this would happen. Sorry Dave I was only going to PM this to you to see if you could validate what I saw. Did not mean to turn it into a hot topic. I just thought that maybe a few others that I know have a 9700 could help. But so far the 9700 has been a champ and thought that was worth sharing. My bad Most site said that the aniso issue was fixed. So when I saw it in SS:SE I was confused.
Folks I ran that level a few minutes and did saw that "biggest" diff was around 45 degrees. There was not a "big" change till then.
However I have ran it though all the other FPS shooters I have and not once did I see any other issues with Aniso. This is a very nice card and I am happy with it.
I had a funny feeling that this would happen. Sorry Dave I was only going to PM this to you to see if you could validate what I saw. Did not mean to turn it into a hot topic. I just thought that maybe a few others that I know have a 9700 could help. But so far the 9700 has been a champ and thought that was worth sharing. My bad
Most site said that the aniso issue was fixed. So when I saw it in SS:SE I was confused. Folks I ran that level a few minutes and did saw that "biggest" diff was around 45 degrees. There was not a "big" change till then.
However I have ran it though all the other FPS shooters I have and not once did I see any other issues with Aniso. This is a very nice card and I am happy with it.
I generally dont play any flight sims, I dont have the full SS:SE game, only the demo, so I have never experienced the 'flawed' aniso implementation on my 8500 denigrating my 'IQ' except in pcchens programme.
I played the IL Stormovik demo and never noticed aniso going in and out.
in pcchens proggy, the aniso 'flaw' is very noticeable as the mip-map lines jump up and down the rotating screen - but this is a worst case scenario.
The only way I can think reasonably think, if this issue is minimised on the 9700, that you would find it 'distracting', is if the mip map borders could still come in and out in extreme circumstances in a continuous rotating screen, such as pcchens programme because the movement could still be there albeit on a much more limited basis.
hmm not too much to get concerned about is it. I dont think I'll buy a Gf4, NV30 and radeon9700 and rotate screens in order to make my buying decision.
I played the IL Stormovik demo and never noticed aniso going in and out.
in pcchens proggy, the aniso 'flaw' is very noticeable as the mip-map lines jump up and down the rotating screen - but this is a worst case scenario.
The only way I can think reasonably think, if this issue is minimised on the 9700, that you would find it 'distracting', is if the mip map borders could still come in and out in extreme circumstances in a continuous rotating screen, such as pcchens programme because the movement could still be there albeit on a much more limited basis.
hmm not too much to get concerned about is it. I dont think I'll buy a Gf4, NV30 and radeon9700 and rotate screens in order to make my buying decision.
It was my understanding that one of the notations (NVIDIA, I believe) was the degree of anisotropy achieved. (i.e. 8x meant 8:1 anisotropy) and the other (ATI, I believe) was the number of samples taken
Of course, I might be completely off my rocker, as I'm going off of a vague recollection of something posted somewhere a long time ago.
Hello everybody! I got news for you! This is a technical board!
Am I the only one interested in this for technical reasons?
I am not interested in pro ATI ppl telling me it's not noticeable so it's not important! Also, I'm not interested in anti-ATI ppl bashing ATI for what appears to be a perfectly fine anisotropic filtering implementation.
But even if it looks great I'm still interested in knowing if a lower level of anisotropic filtering is used on 45degree angels, and if so, how much lower and why it is lower. Why I want to know this should be obvious, but I'll repeat: Because this is a technical board, which means I'm here because I want to know how things work, not because I want to nitpick.
This bickering is so frustrating, it's like asking a group of mathematicians how two different fractals are made, but in stead of telling you the algorithm for making them, they start to discuss which one is prettier. I can decide for myself which one is prettier, I just want to know how it works!
Have I repeated myself enough now? I think so
[/i]
Am I the only one interested in this for technical reasons?
I am not interested in pro ATI ppl telling me it's not noticeable so it's not important! Also, I'm not interested in anti-ATI ppl bashing ATI for what appears to be a perfectly fine anisotropic filtering implementation.
But even if it looks great I'm still interested in knowing if a lower level of anisotropic filtering is used on 45degree angels, and if so, how much lower and why it is lower. Why I want to know this should be obvious, but I'll repeat: Because this is a technical board, which means I'm here because I want to know how things work, not because I want to nitpick.
This bickering is so frustrating, it's like asking a group of mathematicians how two different fractals are made, but in stead of telling you the algorithm for making them, they start to discuss which one is prettier. I can decide for myself which one is prettier, I just want to know how it works!
Have I repeated myself enough now? I think so
[/i]
Joe DeFuria
Legend
Personally, I think the most accurate word is "limitation."
The rotation thing is a quality limitation of the "ATI" implementation of aniso, just as non AA of alpha textures is a quality limitation of most multisample AA implementations.
Can we all agree to go with "limitation?"
Joe DeFuria
Legend
No, I think many of us are. In fact, I started a new thread to "request" that B3D do a tech article on anisotropic filtering that would hopefully satisfy our interest.
http://www.beyond3d.com/forum/viewtopic.php?t=2417
Please add to it if there's other aspects you'd like to see explored.
Joe DeFuria said:
Personally, I think the most accurate word is "limitation."
The rotation thing is a quality limitation of the "ATI" implementation of aniso, just as non AA of alpha textures is a quality limitation of most multisample AA implementations.
Can we all agree to go with "limitation?"
I agree.
One limitation of the R9700's Anisotropic Filtering implementation is that it can, under very specific circumstances, look almost, but not quite, as bad as the Ti4600. Generally, however, it looks markedly better.
Mize
Joe DeFuria said:Personally, I think the most accurate word is "limitation."
The rotation thing is a quality limitation of the "ATI" implementation of aniso, just as non AA of alpha textures is a quality limitation of most multisample AA implementations.
Can we all agree to go with "limitation?"
I'll just refer it as "that contentious aniso thing"
I'll quote myself here.
I think the pictures answered this question to my satisfaction. The 9700 aniso implementation is not noticably different from card X on planes rotated around the Z axis based on evidence in the Serious Sam shots. Closer mip-maps are due to the POV being closer to the floor.
There was clearly an issue/bug/limitation on the 8500 since aniso levels were calculated based on rotations through the X and Y axis but not the Z axis. There has been no indication that anything similar happens on the 9700. I mean, a problem like that, you wouldn't have half-way. Clearly Z axis rotations are taken into account when calculating aniso levels on the 9700.
antlers4 said:
I think the pictures answered this question to my satisfaction. The 9700 aniso implementation is not noticably different from card X on planes rotated around the Z axis based on evidence in the Serious Sam shots. Closer mip-maps are due to the POV being closer to the floor.
There was clearly an issue/bug/limitation on the 8500 since aniso levels were calculated based on rotations through the X and Y axis but not the Z axis. There has been no indication that anything similar happens on the 9700. I mean, a problem like that, you wouldn't have half-way. Clearly Z axis rotations are taken into account when calculating aniso levels on the 9700.
Bigus Dickus
Regular
Wow, silly me for thinking that some GF4 comparison screenshots would easily and quickly settle the matter.
Of course, the reason they didn't is because there are questions about the settings used in each case. Naturally, you'd want those to be the same (other than AF setting, and I'm undecided as to whether 8x:8x or 8x:16x would be better).
A couple of comments... since so many others have addressed Chalnoths post, I see no need to do so. I was shocked when I read it, for obvious reasons. Second... just what were the settings in the GF4 shot? Either the LOD is reduced, or Quincunx was used, becasue it really does seem to be a "blurry mess" as someone else described it. I'm assuming this is due to some particular setting used, so I'm interested in knowing which was responsible.
Of course, the reason they didn't is because there are questions about the settings used in each case. Naturally, you'd want those to be the same (other than AF setting, and I'm undecided as to whether 8x:8x or 8x:16x would be better).
A couple of comments... since so many others have addressed Chalnoths post, I see no need to do so. I was shocked when I read it, for obvious reasons. Second... just what were the settings in the GF4 shot? Either the LOD is reduced, or Quincunx was used, becasue it really does seem to be a "blurry mess" as someone else described it. I'm assuming this is due to some particular setting used, so I'm interested in knowing which was responsible.
Similar threads
