View Full Version : 3Dmark05 High and better IQ...
http://home.comcast.net/~ntxawghaam/14ati_aa_gt3_1831.jpg ati 4xaa 8xaf
http://home.comcast.net/~ntxawghaam/14nv_gt3_aa_1831.jpg
nv 4xaa 8xaf
http://home.comcast.net/~ntxawghaam/14ati_aa_gt3_0681.jpg
ati 4xaa 8xaf
http://home.comcast.net/~ntxawghaam/14nv_gt3_aa_0681.jpg
nv 4xaa 8xaf
uhm which one is doing the correct lighting/shadows etc? theres a big difference was just curious. Images are from ************
jbirney
10-02-04, 11:56 PM
Thanks for the link.
I was quite surprised by the following two sentences.
"It's a specification that falls outside of DirectX, but Futuremark use it anyway."
"So while DSTs improve performance on compatible hardware, to the tune of anywhere between 10% and 16% in our tests, they most certainly do not give good output in 3DMark05's case, compared to the case where they're turned off."
It looks like DST must turned off for a true apples to apples comparison and that would invoke quite the performance hit.
Yea I am kind of dissapointed as FM said last year that no vendor specific paths were going to be used. I guess they changed their mind
Am i the only one who thinks NVidia have the ADvantage in IQ this generation? any comments JB ?
Honestly its too early to tell. But given your and ruined past history and SLIGHT IHV preference...well... :)
Honestly its too early to tell. But given your and ruined past history and SLIGHT IHV preference...well... :)
ok.. its to early more comparisons are needed to declare a winner.. :D
but in the screenshots i posted.. which one have smoother lines ? and looks better to you ? :) if you want pick ramdonly any other set of screenshots from ATI and NVidia in 3dmark05 to compare their AA quality. im really sure that if 99 polls are done in either one most people will be surprised about which comes on top in IQ in the eye candy modes :)
Shamrock
10-03-04, 12:17 AM
Look at those pics ntxaug posted. The last 2 w/ the side of the ship.
Look at the inside of the ropes where the pullies are, MUCH more jaggies in the ATI shot.
Also, look at the curly wood on the left hand side. Inside that wood is a design, MUCH more jaggies there too.
Look inside the cannon, the circle doesnt look at smooth in the ATI shot.
Finally, Look at the sphere of planets and rings, ATI's shot tehre is HORRIBLE!
NV is doing a much better job of IQ here, it's OBVIOUS!
Look at those pics ntxaug posted. The last 2 w/ the side of the ship.
Look at the inside of the ropes where the pullies are, MUCH more jaggies in the ATI shot.
Also, look at the curly wood on the left hand side. Inside that wood is a design, MUCH more jaggies there too.
Look inside the cannon, the circle doesnt look at smooth in the ATI shot.
Finally, Look at the sphere of planets and rings, ATI's shot tehre is HORRIBLE!
NV is doing a much better job of IQ here, it's OBVIOUS!
in ATI defense :) at very high resolutions the jaggies will be less noticeable..for everyone .but still shows which one have the best algorithm. im sure most people will trade easily 1000 points in 3dmarks for much better IQ. i will love to see comparisons between ATI 6xaa and NV 4xaa at that resolution ->1024. to see how much closer NV AA looks in comparison with their best mode.. just for curiosity. but i dont have both cards to compare :(.
Quote:
Originally Posted by Ninjaman09
He was only speculating on what he noticed, and apologized in advance if he was wrong. There is absolutely no reason for this kind of hostility. Go take your fanboy attitude back to wherever you came from, jerk.
And by the way, synthetic benchmarks mean nothing in terms of real-world performance.
I agree here. No need to call him an nvidiot.
It looks liek ATI didn't decrease IQ at all though.
As you can see now, many comparison pics are posted ,i just wanna prove that something wrong with the CAt ai and i`m not nvidia fan boy i have Both of the Cards .. i just 7@%#ed with 3dmark05 IQ on my Ati xt-pe.
here is another set of screenshots..
http://www.************/photos/showphoto.php/photo/16/sort/1/cat/503/page/2
http://www.************/photos/showphoto.php/photo/29/sort/1/cat/503/page/1
shadows diferences are obvious.. for the diferent shadow techniques. one is noisy and the other is sharp.. it comes about which one you like more. :) and flipping back and forth you will notice more jaggies in one screeshot.
here is another set of screenshots..
http://www.************/photos/showphoto.php/photo/16/sort/1/cat/503/page/2
http://www.************/photos/showphoto.php/photo/29/sort/1/cat/503/page/1
shadows diferences are obvious.. for the diferent shadow techniques.
and flipping back and forth you will notice more jaggies in one screeshot.
What I noticed right off the bat was the thickness of the smoke. In the ATI shot more of the images is being hidden by the smoke which should give it a performance boost. Still nVidia should not be allowed to use DST for default benchmarking. IMO it's cheating.
Look at the inside of the ropes where the pullies are, MUCH more jaggies in the ATI shot.
Come on guys, the Ati screenshot doesn't show 4xaa, but 2xAA. :lol2:
Here I provide the same screen with real 4xAA:
http://members.chello.nl/h.appel1/4.jpg
And if you're interested; here with 2xAA and 6xAA:
http://members.chello.nl/h.appel1/2.jpg
http://members.chello.nl/h.appel1/6.jpg
nice moire pattern... AI AI AI?
The moire appears when forcing AF through the 3DMark05 CP (pro version). There is no moire in the default game test or when forcing AF through the Ati cp.
Here the same screen on my Ati X800XT without moire:
http://members.chello.nl/h.appel1/X800.jpg
Come on guys, the Ati screenshot doesn't show 4xaa, but 2xAA. :lol2:
Here I provide the same screen with real 4xAA:
http://members.chello.nl/h.appel1/4.jpg
And if you're interested; here with 2xAA and 6xAA:
http://members.chello.nl/h.appel1/2.jpg
http://members.chello.nl/h.appel1/6.jpg
you seems that have shrinked the images... from higher resolutions :rolleyes: the size doesnt match with the posted 4x aa at 1024x768 screenshots... to compare with yours at higher resolutions is pointless..
Screenshots are taken in-game at 1024x768 and are shown at 1024x768, no re-sizing here.
Screenshots are taken in-game at 1024x768 and are shown at 1024x768, no re-sizing here.
if you look at the size of Bjorn sceenshots.. they dont match the size of yours .. its looks here you have rezized them to a lower custom ratio in photoshop..
Bjorn:
http://home.comcast.net/~ntxawghaam/14ati_aa_gt3_1831.jpg
Me:
http://members.chello.nl/h.appel1/4.jpg
Bjorn:
http://www.skenegroup.net/tertsi/3dmark05/x800moire.jpg
Me:
http://members.chello.nl/h.appel1/X800.jpg
mustrum
10-03-04, 04:48 AM
How lame can it get? People start to fake SS to find IQ differences?
:lame:
perhaps its my Opera browser.. but looks more bigger in their site. that any screenshot posted here.. im going to give you the benefit of the doubt about the AA settings used ,but will be nice to see NV comparisons there.. i can notice slighty more jaggies even in your shots... but they are not that drastic as the earlier shots.
if you later have time.. post how it looks at same settings 4x/8af 1024x. this shot.. in the same way of your previous post.
http://www.************/photos/showphoto.php/photo/15/sort/1/cat/503/page/2
Look at those pics ntxaug posted. The last 2 w/ the side of the ship.
Look at the inside of the ropes where the pullies are, MUCH more jaggies in the ATI shot.
Also, look at the curly wood on the left hand side. Inside that wood is a design, MUCH more jaggies there too.
Look inside the cannon, the circle doesnt look at smooth in the ATI shot.
Finally, Look at the sphere of planets and rings, ATI's shot tehre is HORRIBLE!
NV is doing a much better job of IQ here, it's OBVIOUS!
what about the lighting between the first 2 shots? the nv seems to be missing some light source or the ati one seems to be missing shadows on the aft of the ship and the railings where theres a set of small post
How lame can it get? People start to fake SS to find IQ differences?
:lame:
Where?
what about the lighting between the first 2 shots? the nv seems to be missing some light source or the ati one seems to be missing shadows on the aft of the ship and the railings where theres a set of small post
it looks to me that ATI and Nvidia effects (even when the frames are the same ) are out of syncronization by 1 second or so.. that might explain why obvious diference in position of the fire or smoke at times .. in some shots..
just like here...
this time using ATI screenshot provided by [APPLE] at ...6xaa/8xaf (http://members.chello.nl/h.appel1/6.jpg )
Nvidia screeshot by Bjorn3d 4xaa/8xAF (http://www.************/photos/showphoto.php/photo/33/sort/1/cat/503/page/1)
you need to play with the sliders of your browser until you get both pictures more or less aligned in the same place for better comparison. if you switch between both pictures looks like in the NVidia picture the explotion happens 1second earlier and in the ati shot is like 1 second later. also the shadows looks diferent for the diferent algorithms used.. one tecnique is noiser and the other sharper.. neither one is "correct". both are the simplest shadows of each technique. and each technique have disadvantages.
about the Antialiasing.. i notice very slighty more jaggies also in the ATI 6xaa screesnhot than in the Geforce shot. just follow the circular shapes. and the ropes at the top.
I'm seeing a few other differences as well. When trying to grab a shot for IQ comparisons I thought the nvidia shot was overbright (kind of like the issue they had a while back with NFSU). It could still be that way I'm not sure. But, when watching the scene again I noticed that the mist from the creature was much more prevalent on the nvidia side. On the ATI side, the mist during the time in question was barely noticeable. So, I think the mist caused the difference but I'm not sure which one is correct. If anyone has both cards, this was around the 1325 frame area of the Canyon Flight game test. 1325 was the frame I actually captured to compare and noticed the big difference in lighting (oh, using the 8.07 vs the 66.81s). It basically looks like the mist on the nvidia side almost overwhelms the scene briefly (like a cloud of mist passing through whereas on the ATI side, it's more like a couple of whisps of mist that are barely noticeable.
ATI....................................Nvidia..... ....................ref image
http://img56.exs.cx/img56/3577/ati_807_gt3_1325.th.jpg (http://img56.exs.cx/my.php?loc=img56&image=ati_807_gt3_1325.jpg) http://img85.exs.cx/img85/3601/nv_6681_gt3_1325.th.jpg (http://img85.exs.cx/my.php?loc=img85&image=nv_6681_gt3_1325.jpg) http://img34.exs.cx/img34/3681/refimage_gt3_1325.th.jpg (http://img34.exs.cx/my.php?loc=img34&image=refimage_gt3_1325.jpg)
Edit: It's not a result of the 66.81s. I tried the 61.77s and the screencap was identical to the one from 66.81.
schuey74
10-03-04, 01:06 PM
also the shadows looks diferent for the diferent algorithms used.. one tecnique is noiser and the other sharper.. neither one is "correct". both are the simplest shadows of each technique. and each technique have disadvantages.
No, one technique is not noisier. ATI is doing the proper soft shadows and Nvidia is using DST so the results are much sharper and therefor incorrect. DST does not adhere to DX9 specs and the outputted image is incorrect, there is no arguement there. At the moment Nvidia is gaining 10% - 15% in performance because of this.
On the flipside, it's obvious that when AF is not being forced thru the CP, some of ATI's AI optimizations are becoming apparent. But as Apple has already showed, these optimizations are eliminated when forcing AF thru the CP. The only thing up in the air is how much performance is gained by this, which can easily be checked by running one bench with AF set thru the CP and the other set thru the app.
Could someone check my results above (the ATI vs Nvidia vs Ref Image) shots to see if you get the same thing. To me, it looks like ATI isn't showing most of the mist from the scene but I'm not totally confident in what I'm doing. :)
I was worried that generating the reference image would take quite a while (since they warn it could take up to hours) but it only took a few minutes.
Riptide
10-03-04, 01:51 PM
Mist? It just looks washed out to me.
It's b/c there is a cloud of mist that goes through the scene. If you notice, the reference image looks almost exactly like the nvidia one. When I first saw it, I thought something was wrong on the nvidia side but when watching the scene in motion instead of just grabbing the IQ screenshot, I saw that I could barely see any mist in the ATI scene whereas in the Nvidia scene you can see it pretty easily. It quickly overwhelms the scene and then clears up somewhat. On the ATI side, it's barely detectable and just a couple of poofs of mist mainly to the side of the scene. Of course, the ATI looks better looking at the screencap but it reminds me of the bridge comparison a long time ago between ATI and nvidia. The nvidia picture looked so nice and clear while the ATI pic wasn't in the distance since the fog was being rendered properly on the ATI side (it wasn't on the nvidia side). So even though the nvidia pic looked better in that case, it wasn't correct. To me, that seems a lot like what might be happening here. The ATI pic does look better but it doesn't match the reference image very well.
Edit: You kind of have to see the mist/cloud in motion to understand it. Yes, it's hard to see what is going on in the actual screencap but just look at the reference image.
Edit 2: Whatever is going on, I don't think it's the 8.07s. I went ahead and reinstalled the 4.9s and got the same image on ATI. So ATI and Nvidia both seem to be very consistent between their drivers. Maybe I should redo the reference image to make sure I didn't mess something up with that.
As a double-check, I generated the reference image on the ATI system. Here is that image. Also included the 4.9s just to show it (the strange difference) stays consistent betwen the 4.9s and 8.07s.
ref image....................ati 4.9
http://img80.exs.cx/img80/1638/refimage_ati_gt3_1325.th.jpg (http://img80.exs.cx/my.php?loc=img80&image=refimage_ati_gt3_1325.jpg) http://img35.exs.cx/img35/8918/ati_49_gt3_1325.th.jpg (http://img35.exs.cx/my.php?loc=img35&image=ati_49_gt3_1325.jpg)
so without the mist performance is increased? i know one thing be it fog or mist theres always a performance hit, maybe ati left the mist out to increase performance.
vBulletin® v3.7.1, Copyright ©2000-2012, Jelsoft Enterprises Ltd.