UT2K3 DM-Icetomb performance feedback needed ASAP!

[fivefeet8]

Originally posted by ChrisRay
Oh, Need for Speed Hot persuit 2? What AF quality ect?

Here is a Fraps Timelog. Palm City Open track.


1280x1024 4xAA/8xAF-quality

2003-09-20 13:03:28 - NFSHP2
Frames: 8053 - Time: 133219ms - Avg: 60.449 - Min: 31 - Max: 110

[Behemoth]

Originally posted by fivefeet8
I took them in generally the same place.
they are quite very different places when it comes to comparing IQ.

The fps difference is considerable though. I was also trying to get shots of the light effects and see if there was a difference with them. Seems like those are haven't changed. Also the Pixel Shaded oil effects doesn't seem to be changed in any way as well. The shiny reflective metal in DM-Rustatorium also look the same as with the 45.23's. But of course it runs much faster now.
usually nvidia's IQ lowering optimization involves only a little bit of IQ reduction, its not something easy to spot if you dont put the same frames side by side to check.

I don't think nvidia can optimize every game, but from my testing of these 51.75's, most of my PC games seems to be running noticeably faster. Some have gained as much as 15 fps from the previous 45.23's. Most noticeable in Need For Speed Hp2.
they can globally apply "optimizations" if they want, like they force disable trilinear filtering in ut2k3 and forced lower fp precision in previous drivers.

[fivefeet8]

Originally posted by Behemoth

usually nvidia's IQ lowering optimization involves only a little bit of IQ reduction, its not something easy to spot if you dont put the same frames side by side to check.


That does appear to be the case with my own testing. But the FPS gained is very noticeable. Maybe I'll nit pick a little and retake a few screens to really compare DM rustatorium. I'll post them on that site when I have time.

[fivefeet8]

Posted some better comparison shots to look for differences at the site. DM-Rustatium. 1024x768 4xAA/8xAF-quality.

Here (http://www.geocities.com/pao_vang5f8/Screenshots/PC.htm)

Seems to be very little differences except for the FPS counter.

BTW ChrisRay. I used the ATi driver feedback form from your sig to ask for Digital Vibrance and SS AA. :p

[Behemoth]

yeah they are much better comparison shots. :) i still see a little more blurriness or a little missing something from the 51.75 shots tho but they are very minimal.

[The Baron]

Originally posted by Behemoth
yeah they are much better comparison shots. :) i still see a little more blurriness or a little missing something from the 51.75 shots tho but they are very minimal.
LOD adjustments, perhaps?

[Behemoth]

Originally posted by The Baron
LOD adjustments, perhaps?
i dunno, i have never understood and tweaked this setting :p
but do you get 2x performance just by tweaking LOD setting?

[The Baron]

Originally posted by Behemoth
i dunno, i have never understood and tweaked this setting :p
but do you get 2x performance just by tweaking LOD setting?
No, but that might be part of it.

[hovz]

in the det 50s, the objects in the backround are much blurrier, so they definatly sacraficed lod settings for performance

[fivefeet8]

Originally posted by hovz
in the det 50s, the objects in the backround are much blurrier, so they definatly sacraficed lod settings for performance

Much blurrier? Hmm. I don't think it looks much blurrier at all. In fact, a few of the shots show the 51.75's look less blurry. :confused:

[hovz]

if i had photoshop or something id draw red circles, but ill just point out the obvious one, in the rightmost picture of rust level, in the backround the textures are much blurrier in det 50, and actually i shouldnt have used much, its more of a minor sacrafice

[saulin]

Just wait till the official Dets 50s are out and they are modified for quality as it is the case with the StarStorm Detonators (http://www.scale.hjem.wanadoo.dk/Starstorm/).

Then Nvidia will have the speed and the IQ to handle those maps.

I'm using the StarStorm Dets 45.23

You can check the quality here. performance wise is not much different than the official detonators at all for UT 2003. However DM-Icetomb does run like crap for me since it is based on Dets 45.23.

Quility wise is well you can see the difference. Much sharper images. awesome AF.

Check these shots for quality.

http://www.ngemu.com/forums/showthread.php?t=41426&pagenumber=19

http://www.ngemu.com/forums/showthread.php?t=41426&pagenumber=20

[jAkUp]

Originally posted by saulin
Just wait till the official Dets 50s are out and they are modified for quality

o_0

[saulin]

Nice AF I wonder if ATI can do that.

http://www.ngemu.com/forums/attachment.php?attachmentid=48303

http://www.ngemu.com/forums/attachment.php?attachmentid=48299

[jAkUp]

Originally posted by saulin
Nice AF I wonder if ATI can do that.

http://www.ngemu.com/forums/attachment.php?attachmentid=48303

http://www.ngemu.com/forums/attachment.php?attachmentid=48299

o_0


/take it from personal experience. nvidia cannot touch ati's image quality.

[saulin]

actually Nvidia's AF seems to be better. On the other hand ATI does have better FSAA and less performance hit which is sweet.

If you compare shots of the regular dets against my shots above you will find that Nvidia makes the image blury.

That's why I said that when the Det's 50s are modified like this they will be sweet!!!

Seriously the AF on these drivers is pretty sweet. So you guys can't bring that talk about blury graphics.

[The Baron]

Ah yes, NVIDIA's awesome UT2003 graphics, now with even less Trilinear Filter^H^H^Hling. :rolleyes:

[Hellbinder]

I think its pretty obvious that Any major speed Nvidia is gaining from the DET50 betas is primarily from not rendering any fog. As well as other detail reducing tricks. I'd be willing to bet its the same reason that performance is up in NFS and any other games.

To think they went on tech TV claiming they did not understand why Valve refused to have them used. :rolleyes:

Not to mention they are still stating any HL2 benchmarks are "invalid" without them... :rolleyes:

Disturbing.

[saulin]

Ah yes, NVIDIA's awesome UT2003 graphics, now with even less Trilinear Filter^H^H^Hling.

anyways this article should explain it better.

So there you have it, in the tests we've shown you today, ATI tops NVIDIA in anti-aliasing quality while NVIDIA bests ATI in anisotropic filtering.

http://firingsquad.gamers.com/hardware/imagequalityshootout/page3.asp

Oh well I'm off to bed...

laters dudes.

[Hellbinder]

I have issues with that entire FS review. Not to mention their conclusions about AF are simply OFF. (actually I was thinking of their latest 5900U review)

Its funny how people will pick the one review out of 10 that *somehow* comes up with a conclusion they like to *prove* their point of view.

lets look at some of the other Quotes from this article.

EDIT 8/22/03: Taking a closer look at the image, you can see slight differences between the RADEON 9800 PRO and the GeForce FX 5900 Ultra, specifically if you look at the white rock on the left, just above the shoreline. Textures on the ATI card are slightly crisper than the GeForce FX, giving the 5900 Ultra a slightly blurrier look in comparison

and

There is a much more pronounced discrepancy with 4x antialiasing enabled. The RADEON 9800 Pro maintains sharp textures and smoother lines, while the GeForce FX 5900 Ultra applies a more washed-out effect to the image. To illustrate the difference, check out the rock in the blown-up image (200 percent):

and

The anisotropic filtering shots are not very conclusive. Both images are of comparable quality ; even blown up to 200 percent, the variance between them is too subtle to note.

and

It’s a close race with all of the “eye-candy” turned up. NVIDIA continues to maintain a clear filtering advantage, while ATI’s anti-aliasing implementation is undoubtedly better. This one would have to go to subjective preference, as there are trade-offs apparent in each architecture.

Which i have a HUGE problem with because these guys went out of their way to chose screenshots that happen to be at the Peak weakest point of Atis Adaptive method. If you were to look at other shots from a normal Field of view the Radeon image would undoubtedly be BETTER. Especially at 16x AF. Combine that with the AA and its a no brainer. All these shots used to compare are taken out of normal game experience.

The same issues that surfaced in NASCAR crop up once again in IL2. The RADEON 9800 Pro does a better job with anti-aliasing, while the GeForce FX demonstrates superior anisotropic filtering.

Again.. these screenshots are CLEARLY taken intentionally at the worst case sinario for ATi's Adaptive method. I have to wonder why they keep doing that. Most of the rest of the time the AF will look equal or Superior (16x).

Pretty Suspicious behavior if you ask me. their recent 5900U article is even more troubblesome.

[Hellbinder]

saulin,

Honestly. You dont stand a chance in this area. Especially choosing UT where Nvidia is not even offereing properly filtered AF. ATi also has some AF optimizations for UT, but they dont have anywhere near the impact on IQ. Further you can Turn them off. Which you cant do with an FX.

People will honestly never understand until they play with an ATi card. Once that happens (Generally) its all over and their eyes are opened.

[jAkUp]

please keep in mind this is a compressed jpg, and shrunk to keep it under 100k's

[Behemoth]

Originally posted by saulin
Ok I guess I'm still arround.

How about you show me some of yours UT 2K3 pics in the same map I have here.

http://www.ngemu.com/forums/attachment.php?attachmentid=48303

http://www.ngemu.com/forums/attachment.php?attachmentid=48299

that way we all can see how ****ty Nvidia's IQ is compared to the Radeon.

Or is it that guys try to put Nvidia's IQ much lower than what it really is?

Oh BTW I do agree that Nvidia's DX9 performance does suck.

Yawwwn!. Not I'm gone for sure. See ya tomorrow :)

noone is complaining the ordinary nvidia's IQ here. people are complaining the IQ from the nvidia's IQ lowering optimizations. people are complaining the IQ/speed ratio on fx cards, r3xx beat nv3x in this respect in many areas, not only in dx9. when nv3x renders everything in high quality, it lacks speed, when it keeps up to r3xx speed, it lacks IQ!!!!! you see the problems?

[saulin]

Not a fair comparison since there is sooo much compression. And yes I do understand that your image should look better than that as well. Yikes I had to do 65% image quality to get less than 100K.


However I wasn't blown away byt ATI's superb IQ. Belive me I know the Radeon 9800 Pro quality. My brother who is in the next room has one and yes I have tested it on my system.

[The Baron]

*sigh*

First, I am bothered by Saulin's ignorance. Second, I am bothered by the fact that no one has called him out on what is an utterly obvious conclusion.

He speaks of a modified set of Detonators that mysteriously have such great AF quality. Specifically, the textures are so much sharper and look much crisper.

HB, I am disappointed in you not noticing this. :)

The so-called "optimizations" present in this driver set is nothing more than an LOD bias change. You can do it with normal NVIDIA Detonators, you can do it with ATI Catalysts, you could do it with XGI Reactors, you could do it with ANYTHING. It doesn't matter. There's nothing special about it.

Situations where ATI's AF looks worse are rare, by the way. There are certain times in outdoor scenes where yes, ATI's does look noticeably worse than NVIDIA's AF, but you'd be hard pressed to find it. And NVIDIA'S *always* looks worse in UT2003 because of the lack of trilinear filtering.