PDA

View Full Version : Image quality analysis - more pics needed in reviews?


jimmyjames123
10-23-03, 04:55 PM
Regarding image quality analysis that is getting more prevalent (thankfully) these days...wouldn't it be nice to see the hardware reviewers look at several pictures, as opposed to the few pics that is typical in most reviews? Even in one game, if you capture one particular frame, one video card may look nicer than another card in one picture. However, in another picture in a different part of the game, the situation could easily be reversed, correct?

That's why IQ analysis is so tricky...not only do the pics need to be taken in roughly the exact same spot, but it is difficult to get a definitive conclusion when only using a relatively small sample of still images (and of course, there are also bandwith limits and time constraints which prevent some reviewers from doing anything really in-depth).

So what is a good solution? More pics used in the reviews, available for download? Thoughts from the reviewer regarding actual game playing experience from different cards with different games? Who knows.

digitalwanderer
10-23-03, 05:15 PM
PROBLEM: Bandwidth costs, and good .png screenshots chew up a good chunk of it. :( (I've been debating this issue recently for a review at EB. )

jimmyjames123
10-23-03, 05:19 PM
Yeah I know what you mean. But instead of posting all of the full size images, couldn't the reviewers at least do image quality tests with a variety of images and report the findings, even if they have to reduce the size of some of the images? Also, it would be nice for them to report their (hopefully honest) findings about how two competing cards actually looked in general during actual gameplay, just as a reality check.

LycosV
10-23-03, 06:20 PM
I'm not sure more pictures is the solution. Lots of reviews I've seen lately are bogged down with pages of pictures saying what you can see "both cards display the same image." Personally I only need to see when the images DON'T match. :) I think the best solution would be to have short video clips of the same thing with all cards. then people could watch the clips and decide for themselves. The disadvantage is, as someone said, that will wast a lot of webspace.

LycosV

The Baron
10-23-03, 06:22 PM
I wonder if someone could do this--using FRAPS or some other method of frame grabbing, go through a timedemo in Game X and capture random frames. Then, compare the results using some series of Photoshop actions (not sure exactly how, but somehow).

And do uncompressed PNGs compress if you RAR them or whatever?

Clay
10-24-03, 12:55 AM
Originally posted by The Baron
I wonder if someone could do this--using FRAPS or some other method of frame grabbing, go through a timedemo in Game X and capture random frames. Then, compare the results using some series of Photoshop actions (not sure exactly how, but somehow).

And do uncompressed PNGs compress if you RAR them or whatever?
I've tossed something similar around as well regarding some kind of script in Photoshop (or JASC Paint Shop Pro 8 for that matter) to do a pixel-to-pixel comparison. Kind of like a Beyond Compare for image files...

Good idea about RARing the uncompressed PNGs...heck I'd be willing to just make it known in the review that if you want to receive the high quality PNGs to just e-mail me and I'd send you the RAR of them...that would eliminate bandwidth costs. Of course, responding to the number of email requests could be a bit of a chore I suppose. :)

AnteP
10-24-03, 08:15 PM
RARing a PNG won't do you no good.
At best it'll make it 90 % of the original size or something like that.

Clay
10-24-03, 08:57 PM
Originally posted by AnteP
RARing a PNG won't do you no good.
At best it'll make it 90 % of the original size or something like that.
You're right, forgot about that.

Blacklash
10-25-03, 09:24 PM
I have been staring at two identical images from Max Payne II, one rendered on a 9600XT and the other a 5700U. The only difference I can see with the naked eye is the small white lines that are supposed to be lights in the bookcases are a tad brighter with the XT.

For most of us I think its now really more a matter of taste than of quality. Get what you appeals to you.

You can download the SS here if you like, and IQ compare yourself...

http://www.driverheaven.net/reviews/5700/index10.htm

The Baron
10-25-03, 09:40 PM
Mmm... that's true, to a point. AA and AF still have significant differences, though, and you can often say that one simply looks better than the other.

Blacklash
10-25-03, 10:27 PM
Ati AA is superior, at least in my opinion. I think its executed more efficiently and is "smoother".

I still find Nvidia aa acceptable at 4X. As far as AF goes, to me, its a toss up.

One can certainly assert X texture looks better than Y. From those linked screenshots, if one were to bother downloading them and reviewing them, they would see the antennae of the butterfly in "Mother Nature" clearer on the 5700U than on the 9600XT. My point? The differences are so small that they do descend to consumer taste.

Nvidia sorely offended me in the past with "stupid driver tricks" that did compromise some quality for performance. I am hoping they have learned their lesson.

The 5700U seems to be a step in the right direction. It is about time a real price war started and we the consumers benefit from it.

The Baron
10-25-03, 10:49 PM
At this point, ATI definitely has better AA. In most things, the differences between ATI and NVIDIA AF are all but impossible to see with the naked eye, but there are times when NVIDIA AF is clearly superior (outdoor environments, or environments with bizarre angles).

jAkUp
10-27-03, 11:50 AM
check out the new image quality article...

http://www.firingsquad.com/hardware/imagequality2/default.asp

fivefeet8
10-27-03, 01:05 PM
Originally posted by jAkUp
check out the new image quality article...

http://www.firingsquad.com/hardware/imagequality2/default.asp

Interesting. I guess the "nvidia IQ sucks" argument isn't as prevalent as it use to be.