Radeon HD5xxx filtering issue

[SirPauly]

The thing that bothered me was the perception of filtering quality based by the flower in a static environment. So many linked that to me with blanket statements, when filtering is about a moving environment to me.

 

[KARpott]

The problem nowadays is that everyone judges filter quality by the AF-flower, and ATI knows that!!! Hell, just take a look at the first few posts in here. Please don't take this personally, but we really need some kind of common awareness for this problem, and claiming screenshots inappropriate because they don't show exactly the same scene (but instead with one or two degrees orientation difference) does not help at all!

 

[BFG10K]

I was talking about their trilinear filtering/mip transitions.

Understood; the point I was making was that the issue largely manifests itself only when AF is in operation, not with regular trilinear.

 

[Xarick]

Wow.. the more I read the forums the more sorry I am I bought a 5850 4 months ago. The price may have been good, but all this is not worth it. I think I am gonna craigs list this thing and move back green.

 

[Flipped Gazelle]

Wow.. the more I read the forums the more sorry I am I bought a 5850 4 months ago. The price may have been good, but all this is not worth it. I think I am gonna craigs list this thing and move back green.

What is it about your 5850 that is making you unhappy?

 

[BFG10K]

Wow.. the more I read the forums the more sorry I am I bought a 5850 4 months ago. The price may have been good, but all this is not worth it. I think I am gonna craigs list this thing and move back green.

If you’re happy with the card then there’s no need to dump it.

 

[KARpott]

I wouldn't sell my hd5850, but as long as this is not solved, I won't buy another ATI again.

 

[Avalon]

I've noticed it a bit on my HD4850, but I didn't think much of it. I would definitely love to see this improved for the HD6000 series. I'm a bit surprised this isn't getting more attention. Has anyone tried to make Anand aware of the problem?

 

[distinctively]

The problem nowadays is that everyone judges filter quality by the AF-flower, and ATI knows that!!! Hell, just take a look at the first few posts in here. Please don't take this personally, but we really need some kind of common awareness for this problem, and claiming screenshots inappropriate because they don't show exactly the same scene (but instead with one or two degrees orientation difference) does not help at all!

I don't have TrackMania so I cant do comparisons myself between two cards otherwise I would.
So, do you really think that a change in resolution and screen ratio has no bearing on the results? I definitely see what you are demonstrating, I just have no means to make a comparison. I've seen screen shots change everything dramatically with a single degree. I've seen terrible stretching with changed of screen ratio. Resolution changes have an overall affect on just about every aspect of rendering, including AF. There seems to be absolutely not reason why identical settings couldn't be used. A lot of these pics are done as jpg files with varying file sizes. Even the compression could result in small changes.

In the end, I'm not saying that any one of these points disproves you. What I am saying is that your argument isn't very convincing. I get to see comparative screen shots all the time where I am given:
PNG pictures
System specifications
Identical screen shots at identical times of day in game relativity.
Identical resolution and ratio.
Accurate identification of driver versions and driver settings used, not just AF settings.

There is good reason why being this thorough is important. For all we know, the problems perceived here only exist in a small amount of games under very specific circumstances. We hear all the time that AA is broken here or blurring occurs in this game under these conditions.

I am not saying you are wrong. What I am asking you to do is convince me. Yes I'm skeptical. I have been witnessing your posts for quite some time on different sites. I haven't seen a good comparison yet. All the while, I have been scouring my own games trying to duplicate the described problem with my 5770. I have asked my oldest son to keep an eye out for it too. We have not found a single incident of this yet. The games being observed were (if I remember correctly) Mass Effect 1+2, Bioshock 2, Metro 2033, Oblivion, UT3, Wolfenstein and Serious Sam HD. Now, you may be perceiving a problem but my experience is a pretty awesome track record here. Of course I'm skeptical.

Tell ya what. Give me what I ask for and convince me. Find me some better screen shots. If this is a hardware problem, there should be a multitude of witness to it. Better screen shots should be readily available. Do that and I will support your efforts.

 

[KARpott]

TrackMania Nations Forever is for free:
http://www.gamershell.com/download_24766.shtml
Play the track Nations/White/A04-Acrobatic and look at the texture on the platform. Use the settings from my first post. There you have your comparison. Alternatively, it is quite visible on some grates in MirrorsEdge (compare 2xAF with 16xAF). I don't bother posting any pics because they would not convince you anyway. (unless you explicitly ask for it, of course)

My shots are in .png, the first post includes settings which hold for all the shots from me. But I'm not a rich man with two different graphics cards so I can't give you identical NV shots apart from the AFTester thingy. Who can, actually, apart from testers?. Don't the videos work for you, either? They show it quite perfectly, and they display exactly the same scene, thanks to the TM replay function.
I can actually understand your point. To really prove this, we need clean comparison shots. But if all signs say that it's going to rain within the next ten minutes, I don't need a proof.
Besides, you present it as just my point and you claim you have witnessed my posts on different sites. I only posted about this here and at 3DCenter, and if you can understand German, you can read that there, almost everyone agrees on that issue, although argueing about how much it disturbs while gaming. And there's more media than the shots from me!
It's not even my find, those guys from pcgameshardware (a good German Hardware magazine) stumbled upon it (along with BFG of course!), but probably, they're not convincing enough for you. Quite honestly, I like that you're skeptical, but to me you seem to be a bit paranoid. Why should I even bother starting this here? Because I'm from Nvidia and try to infiltrate the internet with subversive ideas?
And in case you can witness it with TM, please spare me with "But it's only in TM". It happens in extreme cases only. Full stop.

EDIT: Oh, and you claimed you have not witnessed anything in all of your games. How is that more convincing? You don't even show anything! Who tells me that you're not an ATI fanboy trying to undermine my credibility? But I believe you, because I know the issue can be hard to spot when you don't know what you're looking for.

EDIT2: Btw, just coming back from a flight through UT3. I bet you have not touched the map "Defiance", did you? I can perfectly reconstruct the UT shot in my first post. But I agree with you that 99% of all textures in UT do not have visible transitions, although I actually found another texture with the same problem (different to the one from the shot above).
Hint: Look for heavily shaded textures with VERY VERY high repetition frequencies. It's very likely that you can witness the "banding" on VERY VERY VERY fine grates, with a lot of contrast. As I said before, the visibility not only depends on the texture itself, it also depends on how it is shaded or lit.

 

[Xarick]

If you’re happy with the card then there’s no need to dump it.

No offense but how can anyone be happy with ati these days. When I bought mine everyone said drivers were great. ATI then released the 10.6s, .7s, .8s, and .9s. The funny part is I owned a 9800pro and hated it for the drivers.

I have had to tweak shadows on stalker, tweak many things on gothic 3. As well as several other games. Now I read that ati can't even do AF correctly.

At this point I am unsure how anyone who games on the pc can consider or recommend ATI hardware.

 

[Aristotelian]

At this point I am unsure how anyone who games on the pc can consider or recommend ATI hardware.

:whiste:

Here we go again.

 

[Xarick]

sorry, but I get frustrated with this stuff. I just finished starcraft 2. Works fine except for the fact that the mouse cursor corrupts periodically and then magically fixes itself. ATI driver issue that has not been fixed and I doubt will. I just miss having no issues and not reading forums to learn about how my card can't do x or y correctly.

 

[Skurge]

How much does the issue bother you?

Does it affect you enjoyment of the games you play or is it just something you noticed and would like to be fixed.

Im just curious.

 

[KARpott]

In case you refer to the topic of the thread:

As I explained, it affects my enjoyment of TrackMania, for obvious reasons. Now that I know where to look, I observe it in several games, but in most cases, it's not that much of a deal. That's why I wouldn't sell my card right now. But I pretty much want this to be fixed and I want the reviewers to take such things into account when they review the filtering quality.

EDIT: Here's the only thing I can do to give you better ingame shots:





Settings: Win764Bit - HD5850/Cat 10.9 - 1680x1050 - 8xMSAA via application, box filter - af via application - MipMap HQ - A.I off - ingame maximum, no postprocessing

It's done with these replays:
http://www.multiupload.com/YF1QHYGHPG
In each replay, the car breaks to a complete stop at one point. When it stands absolutely still, you can capture the screen via print-screen button and save it uncompressed. The second ?should? work with NationsForever as well. Here's the raw material I created:
http://www.multiupload.com/K0LSMOKN7O
USE THE FOLLOWING SETTINGS:
1680x1050 - 8xMSAA via application - af via application - ingame maximum, no postprocessing

Would be great if someone could upload some NV screens!

 

[Genx87]

I thought this was common knowledge? I know we have had this discussion before in other threads? The question I have is if AMD\ATI were to do the same texture filtering as Nvidia how would their cards perform?

 

[Avalon]

sorry, but I get frustrated with this stuff. I just finished starcraft 2. Works fine except for the fact that the mouse cursor corrupts periodically and then magically fixes itself. ATI driver issue that has not been fixed and I doubt will. I just miss having no issues and not reading forums to learn about how my card can't do x or y correctly.

Which card and which driver and OS? I'm running an HD4850 with 10.5 on win7 64, and do not have any issues with SC2, Civ5, MW2, and BC2, the latest games I currently play.

 

[bryanW1995]

The current theory is that it’s a limitation in their texture cache design. Unfortunately ATi isn’t responding to the issue (I emailed Catalyst Maker directly with my findings but I never got an answer), so it’s unlikely we’ll ever know the true reason.

It'll be interesting to see if the 6xxx series fixes the issue.

you should acquire a cayman xt asap, I'm sure that many of us are interested to see if amd has cleaned up their act sufficiently to convince you to switch.

 

[bryanW1995]

sorry, but I get frustrated with this stuff. I just finished starcraft 2. Works fine except for the fact that the mouse cursor corrupts periodically and then magically fixes itself. ATI driver issue that has not been fixed and I doubt will. I just miss having no issues and not reading forums to learn about how my card can't do x or y correctly.

I've had a few graphical glitches in civ 5. you know what, screw this, I'm ready to go back to the red team. I'll trade you my gtx 460 768 for your 5850. heck, i'll even pay shipping

 

[Flipped Gazelle]

I thought this was common knowledge? I know we have had this discussion before in other threads? The question I have is if AMD\ATI were to do the same texture filtering as Nvidia how would their cards perform?

I think if this were common knowledge, decent tech-sites like AT, TR, etc would have have IQ comparisons in their reviews - well, I would hope so.

It's easy enough to find "graphics irregularities" in both brands that, for me, it doesn't matter. But I haven't encountered an issue that affected gameplay; if I did, I would obviously be annoyed.

 

[KARpott]

Here are some mirror-shots as well, if these are more convenient. I guess, I'll have some NV-ATI-comparison shots some day this week.

 

[MrK6]

So play with AF at 2x if the transition bothers you? I mean, it's a racing game, all the textures should be blurred anyway, so what does the level AF even matter?

 

[KARpott]

It bothers me, period. I'm playing with 2xAF, but the blur is visible while driving, which I don't like. But that doesn't matter anyway.
ONCE AGAIN: Primarily, this is about making you aware of the existence of the transitions, not just in synthetic tests and benchmarks, but also in games. Whether you can live with this or not is up to you. I never said to anyone that HE/SHE HAS TO HATE the ATI filters, I was just criticising the AT verdict on IQ. But I'm constantly confronted with "who cares", "why do you even bother". Please, please, stop it, guys! Don't take it personally, but if you don't bother anyway, then why do you even contribute to this discussion? Not that you're not allowed to express your opinion, but leaving this thread behind is equal to "I don't care", at least for me.
Let me quote Carsten from PCGH on that matter (from beyond3d forums):
„But here's what I am afraid of: Where's the absolute limit of optimization? I mean, you can see more shimmering on current radeons for example, there are differences in NFS Shift with FP16 and R11G12B11. But they are hard to quantify and people just need to be aware of this.

We've had this situation already, when no one took really notice of - yes I call it that - cheats. Until they were delivering really frightening results on driver defaults at some point.

WE - and that is people who are aware of how powerful modern GPUs are - must make sure, that all companies competing in that market invest their ressource not in who can get closer to the line without getting caught, but to really improve PC gaming for all of us. Stable drivers, useful features and good performance when AAA-games (and other titles) are released. Working together with developers to disarm performance pitfalls in their code beforce shipping. But NOT altering the render output to what THEY deem to be an acceptable image quality level - that's OUR part to decide with sliders and checkboxes within the drivers!

Thanks!”

 

[Genx87]

I think if this were common knowledge, decent tech-sites like AT, TR, etc would have have IQ comparisons in their reviews - well, I would hope so.

It's easy enough to find "graphics irregularities" in both brands that, for me, it doesn't matter. But I haven't encountered an issue that affected gameplay; if I did, I would obviously be annoyed.

I have no idea why they wouldnt bring this up. I think it was either Scali or Benskywalker who was complaining about some benchmarks a few month ago where the reviewer was applying max filtering for Nvidia while the ATI filtering was on equal terms of a lower Nvidia setting. It may be it is common knowledge but no sites know which settings match up?

 

[BFG10K]

The thing that bothered me was the perception of filtering quality based by the flower in a static environment. So many linked that to me with blanket statements, when filtering is about a moving environment to me.

Exactly; the same applies to AA too. Screenshots are mostly useless for showing differences in AF and AA, though in this case the tester app does an outstanding job providing it’s configured properly. But to really see differences, you have to see in-game motion.

I discovered the AF issue by myself after trying a few games, and I didn’t have any heads-up from anyone online. After gaming on nVidia hardware for many months previously, I spotted the regression in AF quality immediately.

In certain games with high-frequency/repeating textures I’d see bands of detail following me around wherever I moved. Then I fired up the tester app and lo and behold, it confirmed what I was seeing in-game.

Some people can’t see the problem and/or aren’t bothered by it, but it’s definitely an issue that exists.