View Full Version : GF3 vs 8500 IQ
Violator
11-21-01, 06:23 AM
Apologies if this has been posted before (don't think so).
Makes an interesting read tho' :D
http://www.geocities.com/legion88/
Kingslayer
11-21-01, 07:47 AM
Maybe this is why the image quality of the benchmark has been totally ignored.
People have different opinions on what video quality is.
Is the card with the better video quality the one that renders the graphic closer to what the designer designed to to be?
Or is the card with the better video quality the one that appealls to the users eyes more?
While these test clearly showed that the nVidia card is redering the graphics closer to design, it is also showing that the Radeon is a whole lot more appealing to the naked eye, regardless of what the original image looked like.
Look at the original truck picture. The original's license plate is totally blurry, yet the Radeon rendered graphic is almost readable. All but the second letter. While that might not be the way it was supposed to be rendered, thats phenomenal.
That was very enlightening. I am surprised by those findings. I still feel Radeon cards are superior in rendering 2D imagining, and the 3dMark 2001 visuals also appeared "better". Maybe ATI takes some liberties when rendering a scene instead of recreating the 'bland' original...
I'm also wondering if this is limited to the Radeon 8500, since there were blatant improvements when I went form a GF3 to my Radeon VIVO in all image quality (gaming and 2D).
]-[itman
11-21-01, 07:49 PM
I see it this way:
You have a painting. You ask two different painters to copy the painting. One does so very close to perfectly. The other makes the picture look better than the original with different "techniques." So yes this is really good now but it may cause problems in the future. What happens if a new "painting" or game comes out that has better "techniques" than the second painter while the first painter still copies it. In this case the first painter's picture looks better. So either the second painter will be stuck will a worse picture.
Just my opinion on the matter.
Default
11-21-01, 08:15 PM
Originally posted by ]-[itman
I see it this way:
You have a painting. You ask two different painters to copy the painting. One does so very close to perfectly. The other makes the picture look better than the original with different "techniques." So yes this is really good now but it may cause problems in the future. What happens if a new "painting" or game comes out that has better "techniques" than the second painter while the first painter still copies it. In this case the first painter's picture looks better. So either the second painter will be stuck will a worse picture.
Just my opinion on the matter.
radeon 8500 has been proven in many benchmarks to be superior in future games that take more advantage of the new dx 8.1. for instance compare the performance gap between the geforce TI500 and radeon 8500 in q3 then take the same engine apply it in newer ways (Castle Wolfenstine) and some how the radeon closes the gap. This has been a trend in most newer games that have been hitting the market now, radeon just creams TI500 in most of the new games. I think in the end nobody is going to notice if 1 or 2 pixels are different from card to card. The fact is that radeon 8500 was designed for the next generation of games, which is why ATi strongly permotes the support for 8.1 on the card. Unfortunatly this is also why 8500 does so poorly in present day games. seams like they were thinking too far ahead and forgot about the present.;)
]-[itman
11-21-01, 10:01 PM
I see your point but I was looking farther into the future like how the gf3 is now the geforce2(not pro or ti or any of the new versions just the original geforce2). When the price drops and their not considered "top of the line" cards. And this may not happen I'm just saying it could cause problems if something new is created in future games that conflicts with the 8500.
thestramel
11-21-01, 11:58 PM
Thanks. That was a good read. However, I'm curious about the reference images. How do you choose a universal reference image? Seems like it would be hard not to be biased in some way. I mean if the reference image was produced with a specific hardware architecture in mind how can you compare image quality with a card using a different architecture? Seems like the author preferred the Geforce3 image rendering over the ATI because it more closely matched the reference image (perhaps the reference image was produced using a Geforce3?). Kind of a tough call to make. All of his points regarding the two cards seemed pretty subjective to me even though the point was to be completely objective. Also, he mentioned many times the main goal was to render images as the original developer/artist/programmer intended. How do you prove this was accomplished? I guess it all boils down to what the user likes. If you like how it looks who cares who else does.
funnyperson1
11-22-01, 12:02 AM
its whatever loks better to the readers...
]-[itman
11-22-01, 09:39 AM
To touch one of your points, 3dMark has the reference picture in it's memory. Then the highlighted images you saw was 3dMark comparing the two images not the reader/writer comparing them. I see the author's point however if you can make a picture look better than the original isn't that saying a whole lot about your card? The only problem i can see with it is what i stated above.
Default
11-22-01, 12:30 PM
Originally posted by ]-[itman
To touch one of your points, 3dMark has the reference picture in it's memory. Then the highlighted images you saw was 3dMark comparing the two images not the reader/writer comparing them. I see the author's point however if you can make a picture look better than the original isn't that saying a whole lot about your card? The only problem i can see with it is what i stated above.
it would more than likely be fixed with driver updates. Or the game will be made to support it.:eek:
]-[itman
11-22-01, 01:11 PM
Originally posted by Default
it would more than likely be fixed with driver updates. Or the game will be made to support it.:eek:
If it's a new game with new technology they're not going to make the game with old technology just so it can support a "second generation" card(thinking in the future when the 8500 will be 2nd gen). And it may be fixed by drivers but then again it might not and judging how ATI does their "older card" drivers I wouldn't bet on it unless they turn around here real quick which is very possible as well.
Cisco Kid
11-22-01, 01:39 PM
Originally posted by ]-[itman
If it's a new game with new technology they're not going to make the game with old technology just so it can support a "second generation" card(thinking in the future when the 8500 will be 2nd gen). And it may be fixed by drivers but then again it might not and judging how ATI does their "older card" drivers I wouldn't bet on it unless they turn around here real quick which is very possible as well.
But the 8500 performance has really increased quite a bit with the latest driver from ATI. They are now on a cycle of pumping out a driver hopefully once a month. I think you will see a big imp in its performance. It is the card for future games and it will last the average gamer quite a while...
cisco kid :)
living in Vancouver BC I am tempted to drive down to Bellingham and grab up the visiontek Ti200 for $99 US damn thats a good deal , I think all would agree andf I am an ATI fan bigtime
]-[itman
11-22-01, 01:53 PM
I agree you CAN'T beet that deal. I do hope that ATI keeps on going with their drivers because i plan on getting an 8500 for christmas. But I'm not going to get it if they haven't gotten to the point where you can atleast play the games you want to play without conflicts.
thestramel
11-22-01, 11:45 PM
Yeah. There are definitely conflicts with some games. What's funny is that the retail version ships with CS and the card has problems with opengl. I've played CS and sometimes it just crashes to the desktop and the display settings are changed to 640x480. I do hope they continue the driver improvements.
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.