The IQ Gaming Thread

[EastCoasthandle]

Is there much difference in IQ when using AA vs no AA other then jaggies, lets find out. Lets use Crysis Benchmark tool (http://downloads.guru3d.com/download.php?det=1791) and set the "Screenshot each frame" to 1900. and post the results. You can click on the view button to find the ss. Below you will find a SS of the in game settings:
http://i11.photobucket.com/albums/a154/eastcoasthandle/5f315ca6.jpg
As you can see:
1 demo loop
1280x1024 with no AA
1280x1024 with 4xAA
You will post 1 screen shot with 4xaa and one with no aa at frame rate 1900.

When other benchmarking tools become available to allow use to ss certain frames within other games they will be added to this OP. Both ATI and Nvidia users are encourage to post your results. Please keep in mind this is not about how high frame rates are but the IQ of the SS.

Edit: Disable motion blur, as this allows for a clearer picture.

[Jizzler]

Looks like it only captures jpeg (med quality jpegs at that, ~130KB @ 1280x1024).

Probably not the best candidates for IQ comparison.

Nice program otherwise, just need lossless screen caps.

[Dr. Spankenstein]

Well, there is the option to change the screenshot capture format to .tga. You could then convert to .png, if you were so inclined.

I don't have a problem with .jpeg if you are comparing for AA effectiveness. Should be pretty obvious.

Thanks for taking it upon yourself to try this! Waiting patiently for results!



p.s. to change format you might want to create/add to your autoexec.cfg
con_restricted=0
e_screenshot_format .tga

[Jizzler]

Is it a CVar setting? I looked around the readme, but didn't see anything...

The jpeg artifacting in these shots is pretty apparent, so for image comparison I'd prefer lossless caps.

[Jizzler]

Yeah, just look at the (jpg) shots taken. :D

Fine for posting a sweet scene or score, but but pretty bad when the topic is image quality (whether it be DX9 v. DX10 or No AA v. 4xAA, etc).

@Dr. Spankenstein - I see your edit and I'm going to try that now. Will post caps.

[Jizzler]

Are our caps different somehow?

Here's one of mine: http://jack.abyssmal.com/temp/ScreenShot0000.jpg

Artifacting - Around the text, background items, and most importantly since this is a discussion about AA, the palm trees.

[Jizzler]

Couldn't get those settings to work, so I Frapped it at nearly the same frame.

http://jack.abyssmal.com/temp/Fraps-Crysis-Frame-1902.png

Now that's a screenshot - no artifacting.

These two shots above are at the No AA setting.

-edit-

Here it is at 4x AA:
http://jack.abyssmal.com/temp/Fraps-Crysis-Frame-1902-4xAA.png

The decking is noticeably improved. So is the house, some of the bushes. Rock edges and the items in the background are cleaner.

All done on 8600GTS.

[Jizzler]

Did you look at mine? It's absolutely horrid compared the PNG. I can see your PNG, but the link for your JPG isn't coming up...

Sorry that I'm anal about these things, perhaps it's because I work at photography studio ;)

Anyhoo, I edited my above post with a 4xAA shot and my findings.

[FR@NK]

I can see your PNG, but the link for your JPG isn't coming up....

I cant tell any difference between the two.

[EastCoasthandle]

Did you look at mine? It's absolutely horrid compared the PNG. I can see your PNG, but the link for your JPG isn't coming up...

Sorry that I'm anal about these things, perhaps it's because I work at photography studio ;)

Anyhoo, I edited my above post with a 4xAA shot and my findings.
You are obviously making a mountain out of a mole hill, the differences between jpg and png is very subtle for this excerise. No, it's not about being anal, you say that the IQ between jpg/png is apparent and it's not apparent. :shadedshu

I cant tell any difference between the two.
Thank you...

[Jizzler]

You are obviously making a mountain out of a mole hill, the differences between jpg and png is very subtle for this excerise. No, it's not about being anal, if you were you would be able to show minuscule differences. You say that the IQ between jpg/png is apparent and it's not apparent. :shadedshu

Thank you...

Are you kidding?

This is night and day to me: http://jack.abyssmal.com/temp/JPG-PNG-Compare.png

A "100" jpeg @ 1280x1024 should be in the 600-900KB area (depending on the scene). These jpegs being produced are no more than 140KB. There is so much loss data that a comparison is meaningless. It's not motion blur as shown by the blow-up, those are jpeg artifacts.

[ghost101]

Im afraid the artifacting in the jpeg pictures are obvious. And yes, it looks like motion blur is disabled in those screenshots.

edit: or it could be due to the different frames that explain the nonexistence of motion blur, one is where the camera is still/moving

[EastCoasthandle]

The difference I've obtained between the motion blur with jpg and png are nearly none existent to the human eye for the exercise found in the OP. The pics posted by jizzler are different from what I've seen. I've already tried this myself didn't find that much of a difference. The only difference I notice between jpg and png using frame 1900 with blur enabled is that green lettering is more brighter using png.

[ghost101]

The difference between the motion blur with jpg and png are nearly none existent to the human eye for the exercise found in the OP. The pics posted by jizzler is completely bogus and have been altered. I've already tried this myself and png doesn't disable motion blur, the pic is altered. The only difference between jpg and png using frame 1900 with blur enabled is that green lettering is more brighter using png. Palm trees do not become more clearer just because you use png.

What do you mean? No one is saying there will be a difference in motion blur. What he is saying is that artifacting significantly affects your analysis of AA. Which should really be done in a frame which does not have motion blur i.e. where the camera is still.

[Dr. Spankenstein]

Couldn't get those settings to work, so I Frapped it at nearly the same frame.

http://jack.abyssmal.com/temp/Fraps-Crysis-Frame-1902.png

Now that's a screenshot - no artifacting.

These two shots above are at the No AA setting.

-edit-

Here it is at 4x AA:
http://jack.abyssmal.com/temp/Fraps-Crysis-Frame-1902-4xAA.png

The decking is noticeably improved. So is the house, some of the bushes. Rock edges and the items in the background are cleaner.

All done on 8600GTS.

I find the difference between the .jpg and .png astounding! Plus there is a appreciable gain in IQ with 4XAA.

Thanks for those!

Why can't I see ECH's screenies?

There is also a way to cheat this. Create an Uber-res screenshot (http://www.incrysis.com/forums/viewtopic.php?id=11286)

Looks like this...
http://www.freewebs.com/rsplayerkillers/00002.jpg

Hrmm...needs more AA methinks!:rolleyes:

[ghost101]

Palm trees do not become more clearer just because you use png.

Yes they do, less artifacting.

[EastCoasthandle]

What do you mean? No one is saying there will be a difference in motion blur. What he is saying is that artifacting significantly affects your analysis of AA. Which should really be done in a frame which does not have motion blur i.e. where the camera is still.

-one image is using motion blur the other is not
-one image has been altered while the other is not

If you are going to show results you either use motion blur or disable motion blur. Not use motion blur in one pic to say jpg is showing artifacts.

[ghost101]

that is incorrect

Take a screenshot of very fine text with jpg and png output. Tell me which is better defined.

[ghost101]

-one image is using motion blur the other is not
-one image has been altered while the other is not

If you are going to show results you either use motion blur or disable motion blur. Not use motion blur in one pic to say jpg is showing artifacts.

Its probably because the frames are not the same. Like ive said many times, camera is moving or still.

edit: FOCUS CAREFULLY ON THE POSITIONING OF TREES

edit2: just to help you

http://jack.abyssmal.com/temp/Fraps-Crysis-Frame-1902.pnghttp://jack.abyssmal.com/temp/ScreenShot0000.jpg

[EastCoasthandle]

Its probably because the frames are not the same. Like ive said many times, camera is moving or still.
That makes no difference from the dramatic differences that you praise. The photo is altered.

[ghost101]

That makes no difference from the dramatic differences that you praise. The photo is altered.

You are so stubborn its a joke.

Please just google jpeg artifacting. I cant be bothered.

[EastCoasthandle]

Just to help you
http://i11.photobucket.com/albums/a154/eastcoasthandle/th_Crysis_1901.png (http://i11.photobucket.com/albums/a154/eastcoasthandle/Crysis_1901.png)
http://i11.photobucket.com/albums/a154/eastcoasthandle/th_Crysis_1900jpg-1.jpg (http://i11.photobucket.com/albums/a154/eastcoasthandle/Crysis_1900jpg-1.jpg)
motion blur turn off.
As you can see, no disernable difference like the photos posted by jizzler.

[Jizzler]

Crysis from my estimation take jpeg screenshots at level 50. If it took them at 100 (minimal data loss), then I wouldn't have the problem that I'm having right now with the shots.

You can take my blow-up out of the equation, but the fact is I have a jpeg screen capture using the Crysis benchmarking tool set for frame 1900 (though it takes 1899 every time). It has artifacts. I didn't lie or alter the photo as anyone here in the forum can replicate the test and post it as well.

[Jizzler]

Just to help you
http://i11.photobucket.com/albums/a154/eastcoasthandle/th_Crysis_1901.png (http://i11.photobucket.com/albums/a154/eastcoasthandle/Crysis_1901.png)
http://i11.photobucket.com/albums/a154/eastcoasthandle/th_Crysis_1900jpg-1.jpg (http://i11.photobucket.com/albums/a154/eastcoasthandle/Crysis_1900jpg-1.jpg)
motion blur turn off.
As you can see, no disernable difference like the photos posted by jizzler.

Yeah, and that's a ~750KB jpeg you posted.

[EastCoasthandle]

Crysis from my estimation take jpeg screenshots at level 50. If it took them at 100 (minimal data loss), then I wouldn't have the problem that I'm having right now with the shots.

You can take my blow-up out of the equation, but the fact is I have a jpeg screen capture using the Crysis benchmarking tool set for frame 1900 (though it takes 1899 every time). It has artifacts. I didn't lie or alter the photo as anyone here in the forum can replicate the test and post it as well.

You have taken this issue and blown it so far out in left field it isn't funny. I have already posted 2 screen shots. One using jpg the other png which clearly show how similar they are.

[EastCoasthandle]

Yeah, and that's a ~750KB jpeg you posted.

So what? It's what Crysis Benchmark Tool used for jpg. :rolleyes:
Something I already explained in my OP (instructions wise)
Why is it, if you follow the instructions found in the OP our photos are so drastically different?

[Jizzler]

So what? It's what Crysis Benchmark Tool used for jpg. :rolleyes:

So if somehow our Crysis have different defaults for jpeg quality this whole thing could have been dealt with back at post #9 when I posted my actual jpeg screenshot or even post #2 after I ran the benchmark and saw that it created medium-quality jpegs.

[ghost101]

There are still artifacts, soom into a single leaf and compare with the png picture.

Notice how when anandtech or [H] compare AA quality they zoom in, AA differences are subtle. I dont have any image software and so cant do it.

So next time dont accuse someone of foul play just because you are stubborn.

[EastCoasthandle]

There are still artifacts, soom into a single leaf and compare with the png picture.

Notice how when anandtech or [H] compare AA quality they zoom in. I dont have any image software and so cant do it.

This is completely off topic to the fact that jizzler's pics are completely different. I never said nor implied that jpg never show compression artificats. What I said was that I didn't find any discernible difference for the exercise of this thread. :D

[Jizzler]

Crysis from my estimation take jpeg screenshots at level 50. If it took them at 100 (minimal data loss), then I wouldn't have the problem that I'm having right now with the shots.

You can take my blow-up out of the equation, but the fact is I have a jpeg screen capture using the Crysis benchmarking tool set for frame 1900 (though it takes 1899 every time). It has artifacts. I didn't lie or alter the photo as anyone here in the forum can replicate the test and post it as well.

Yes, like I said, I wouldn't have much of a problem using high quality jpegs. The ones the Crysis Tool is producing for me are crap in comparison.

[EastCoasthandle]

Yes, like I said, I wouldn't have much of a problem using high quality jpegs. The ones the Crysis Tool is producing for me are crap in comparison.

I don't have this problem, that is how I was able to show you the 2 screen shots
-png using fraps
-jpg using Crysis Benchmarking Tool

[Jizzler]

Well, my two Frap PNG's, one No AA, one 4xAA are there back at post #10 to represent the 8600GTS sect for anyone who wants to do some nitty gritty down to the pixel AA comparisons.

I need to get to sleep :)