Examples of ATI filtering failing?
- Thread starter kihon
- Start date
kihon said:
depends on what you mean with 'proof'......judging filtering iq is not easy it seems (look at the polls here)
then the difficulty is, as you mentioned, that you need to look at it in motion. the screenshot comparisons ive seen so far show little to no evidence of reduced iq to me. maybe someone comes up with some vids or a new utility one day.
the technical side of things can proof a lower filtering quality with these optimizations but fails to proof the relevance for actual gameplay imo so far. it casn help to get a better understanding though:
http://www.digit-life.com/articles2/gffx/nv40-rx800-3.html
But now that we know that the optimizations are there in the RV360 series, and there are lots of people with the cards, surely those who are better able to notice IQ issues with the filtering should be able to say - ok on my 9600Pro on game X I can see mip map boundries "crawling" on map Y when running along this corridor - or something like that?
Now that people are aware of the issue, given the number of RV360s out there we should be able to identify atleast one situation in a game where the filtering is failing?
Now that people are aware of the issue, given the number of RV360s out there we should be able to identify atleast one situation in a game where the filtering is failing?
kihon said:
me thinks you would have to play the same scene with another renderer providing 'full' trilinear first to notice any difference with rv360. many people may also use aniso in conjunction with trilinear which would make it harder to spot the possible weakness of the optimization....
it could also be the case that the iq difference is so small that you wont notice it at all
ERK said:
If there's no visible difference, do you think a CP checkbox option for "Reduce Game Performance by 15% Or More" would go over very well?...
Seems to me that's what an option for defeating non-IQ affecting performance optimizations would be likely to do. If an optimization degrades IQ that seems reason enough to provide a defeat switch, but if an optimization results in nothing more detectable than a performance increase, then what's the point of defeating it?jvd said:
I was hoping this thread would be specific to examples where the optimization fails. That way I (and others) can make an educated decision if the ATI cards are worth getting - and also if there are examples of when it fails, I would like to see how ATI handle it given they have said that they would correct it.
I can give you a opinion, but you cant aford a card anyways so why bother? But just form my usage of a AIW 9600pro for half a year in mostly shooter games i havnt seen a "failure" yet. and it (9600pro) looks better than my old 9700pro, but my 9800xt can run high res, and looks better. All three are FAR better looking than my buddys 5900. Im sure when sombody finds a "Failure' it will be posted all over, and ATI will respond. Glad your just trying to help out the folks who are going to buy a new card. Cant go wrong with a R300, to R420, and nv4x, all will be just great on the IQ. Just wish my X800s will come this week.kihon said:
kihon said:
The best education, by far, is practical. Buy yourself an x800 at a 30-day, satisfaction-guaranteed vendor like Best Buy--and take a look at it yourself for a month. If you aren't satisfied, return it for a full refund. That is by far the best way to evaluate it. The problem with Internet editorials on the subject is that they are filtered by the biases and prejudices the writer may have, but which he doesn't reveal within the body of his article. Such evaluations can be tilted either way and you'd never know it, and the experience of installing and using the card yourself is of course far superior.
I'm reminded of reading the original 3dfx V3 editorials on the IQ of the V3 by sites such as Anandtech, which cheefully put up screen shots that were inferior to the IQ provided by the V2 at the time, and which made all sorts of derogatory comments about the V3's IQ. Well, I followed my own advice above and discovered that AT was way, way wrong--the IQ of the V3 was excellent--better than any card I'd had, including the TNT I was using when I picked up the V3. It came out later that there was a serious problem with the screen-grab software AT used with its early V3, such that the IQ of the screenshots he grabbed were far below the IQ of what was on the screen. But AT played dumb, and pretended not to see the difference and posted those horrible screen shots anyway. It was among the most unprofessional, irresponsible things I've ever seen on the Internet since 1995. Trust me--if you want to know "the truth"--take my advice as you risk nothing and have potentially everything to gain.
Thanks for the feed back Karlotta - Im relieved to find the optimizations are not noticable for you on the 9600 - I suspect the x800 will have equal or better filtering.
I am unlikely to go for the X800 (atleast at its current price) but hopefully when they come out with a mid range card based on the 420 - ie not the ones based on the 3xx range, I can consider it without worrying too much about lower image quality than my 9500Pro.
Im not sure if any shops in the UK do a 30 day trial.
Thanks again for the responses.
EDIT: just to add - I have more confidence not simply because of Karlotta not noticing the difference, but also because no one seems to complain about it.
I am unlikely to go for the X800 (atleast at its current price) but hopefully when they come out with a mid range card based on the 420 - ie not the ones based on the 3xx range, I can consider it without worrying too much about lower image quality than my 9500Pro.
Im not sure if any shops in the UK do a 30 day trial.
Thanks again for the responses.
EDIT: just to add - I have more confidence not simply because of Karlotta not noticing the difference, but also because no one seems to complain about it.
Kihon: Just as a sidenote, I've got a 9600 pro to that I've been driving meself a little insane looking for faults with comparing it to me 9700 pro and I can't see a difference for the life of me.
I'm not saying there IS no difference, I'm just saying that in playing my games on both cards I can't tell a difference.
I'm not taking screenshots and comparing, or zooming or filtering anything...I just play one level for a bit on one and then turn around and play the same bit on the other card at the same settings and try and notice a difference....
...and I haven't been able to.
I'm not defending ATi, I'm just adding me .02.
I'm not saying there IS no difference, I'm just saying that in playing my games on both cards I can't tell a difference.
I'm not taking screenshots and comparing, or zooming or filtering anything...I just play one level for a bit on one and then turn around and play the same bit on the other card at the same settings and try and notice a difference....
...and I haven't been able to.
I'm not defending ATi, I'm just adding me .02.
Big Bertha EA
Newcomer
digitalwanderer said:
This is EXACTLY why I am getting sick of the whole witch hunt mentality of some 3D websites.....
It has already failed.
Their are two was you can talk about its failure.
1. Is mathmatically which I'm sure anyone who understands what trilinear is knows it has already failed this.
2. IQ I think the poll clear shows that more poeple think that trilinear looks better then trylinear.
Now you could try a third option but ATI has specifically said stuff making this third way not useful
3. The IQ degrading is small enough for the corresponding Preformance boost.
But ATI have stated explictly their is no IQ degradation at all they claim it lokos better in some case but never worse.
Truely if ATI's implementation was dynamic and didn't have any problems then ATI wouldn't have to disable ( which they do ) when colour mipmaps are used. Since it isn't able to handle these situation I once again give this as a reason for its failure.
Their are two was you can talk about its failure.
1. Is mathmatically which I'm sure anyone who understands what trilinear is knows it has already failed this.
2. IQ I think the poll clear shows that more poeple think that trilinear looks better then trylinear.
Now you could try a third option but ATI has specifically said stuff making this third way not useful
3. The IQ degrading is small enough for the corresponding Preformance boost.
But ATI have stated explictly their is no IQ degradation at all they claim it lokos better in some case but never worse.
Truely if ATI's implementation was dynamic and didn't have any problems then ATI wouldn't have to disable ( which they do ) when colour mipmaps are used. Since it isn't able to handle these situation I once again give this as a reason for its failure.
My advice is to just swing with it and let it all play out. People got a right to be miffed at the way it was implemented and the fact that there isn't an on/off toggle for it, if it doesn't affect image quality than no amount of anal retentive examination of it will show that.Big Bertha EA said:
To argue against investigating it will just bring down the accusations of fanboyism and bias.
Besides, if they have a point I wanna hear it and feel I gotta listen...I like to think I can get as good as I can give.
But I ain't seen any smoking guns either and until I do I'm just not worrying about it too much, life's too short. 8)
