Image Quality Discussion

[Anarchist4000]

You are right and I'm sorry for that. But let me just say this to Anarchist4000:

Angle independent filtering means that, regardless of the angle, that a specific object is being rendered (and apart from the card's filtering methods), no "optimizations" will be used to filter that object.
"Optimizations" usually are blur effects or some form of edge smoothing, to save the GPU some work or simply covering for the lack of features of the card. This is extremely evident in G70/G71 and it also happens in R580. In the Tech-Report example I gave you @ 45º, R580 manages to blur out some of the edges and surfaces. In G80, no "optimizations" are used to achieve HQAF and the objects are rendered flawlessly. It applies its filtering methods in any angle, without these "optimizations", hence why it's angle independent.

And as the Tech-Report conclusion says, about performance when using this type of HQ filtering, G80 manages to crank twice the fill rate of the R580 at 16X AF, despite having to work harder to produce higher quality images.

If that truly is the case then why does high quality G80 AF differ by angle? All of the tests show that G80 clearly isn't performing angle independent AF with the high quality settings. While it's coming close it's still different slightly by angle. If it was angle independent the tests should show a perfect circle, which they aren't. And if the AF is so superior then why can't I tell the difference based on the screenshots provided?

 

[Silus]

If that truly is the case then why does high quality G80 AF differ by angle? All of the tests show that G80 clearly isn't performing angle independent AF with the high quality settings. While it's coming close it's still different slightly by angle. If it was angle independent the tests should show a perfect circle, which they aren't. And if the AF is so superior then why can't I tell the difference based on the screenshots provided?

Ah great, a new thread

First of all is, you're paying too much attention to the Mip Map transitions. Mip Map transistions are good synthetic benchmarks, for the card's image filtering methods. However, it's just that, a synthetic benchmark. Just like 3DMark can give you an indication of a card's power, by giving it points, after it executed a few tests. But it's still just a bunch of numbers, that may or may not mean anything in the real tests: Games.
That's the whole reason why I visit [H] more than any other site, for evalutions/reviews

Now looking at the HL2 screenshots in the Tech-Report example. Notice that G80, regardless of quality settings and/or angle, always filters the image properly. It does NOT apply any optimizations, to achieve High-Quality images. The R580 does. Sure that when applying HQAF, R580 does very well, but still you can notice a few edges smoothed out, but to a much, much lesser extent, making it harder to see the differences, between G80.
Point is, it's irrelevant for G80 if it's set to default or HQ, or if you're @ a certain angle. It will ALWAYS apply its filtering methods, without optimizations and even so, it has twice the fill rate of R580, which basically means no performance hit will come from it.

 

[Cliff Couser]

here's a less technical question.

Is the image quality of current high-end video cards the same?

I mean....remember the days where ATi had better IQ and overall image quality in games.....does this still hold true today?

Thanks!

 

[capable heart]

here's a less technical question.

Is the image quality of current high-end video cards the same?

I mean....remember the days where ATi had better IQ and overall image quality in games.....does this still hold true today?

Thanks!

It's not really true anymore. The 8800 series has fixed that.

 

[Silus]

here's a less technical question.

Is the image quality of current high-end video cards the same?

I mean....remember the days where ATi had better IQ and overall image quality in games.....does this still hold true today?

Thanks!

What capable heart said.

With G80, the table is turned towards NVIDIA, in what better Image Quality is concerned.

 

[HaMMerHeD]

If ATI dumps HQ AF for R600 and up, I will be eternally grateful. I'm tired of clicking the checkbox, I'm tired of spelling it out, and I'm tired of explaining what it does. Now that NVIDIA has beaten it with G80, HQ AF is meaningless.

Honestly, even though the synthetic tests show that G80 has better AF quality, it looks the same to me. Unless you take screenshots and look at it zoomed in a few hundred percent, there's no visible difference. Who plays a game looking for this stuff? People who don't like having fun, that's who.

 

[sam0t]

What capable heart said.

With G80, the table is turned towards NVIDIA, in what better Image Quality is concerned.

Technicly yes, in real worl, like Mark said, you cannot tell the difference if you arent looking with a 100% zoom. Finally we are beyond driver optimizations etc and its up to game makers to make their games look good. So a big YAY that finally 3D card makers are equal on IQ

Iam not tired of clicking the HQ AF tic box as all my profiles have it ON, but yes, ATI should make it default still. If not for any other reason, then for the one, that there wont be any more articles about IQ and ATI "default" comparisons. Bad english my bad.

 

[Cliff Couser]

hmmm...this is eye-opening for my case anyways....My last gaming card as a 64mb ATi. (many years ago).

so, i've been holding off until the R600 for my next gaming rig, on the basis of getting better IQ and image quality. (which I'm interpreting now was true UNTIL G80)

seeing as the R600XTX may not be all its cracked up to be, or if delayed again...I may just go over and get an 8800.

Thanks guys!