AA: which one looks better?

[Ante P]

^^ me agrees

At first I though people were blind when they didn't see that "the right pic looked obviously better" guess it was just my gamma

as Ichy I too don't really know how to config it properly, where' the noobie guide?

Essentially for each of the red green and blue areas in the test image adjust the gamma ramp for that component (using your hardware's control panel) until the solid colour appears to be exactly the same brightness as the dithered area around it.

When you have done this for all three components the white regions should also be correct, and your display is correctly adjusted.

aghh I did that on my monitor with Hercules built in proggy once
afterwards everything looked like the inside of a butthole (well actually I haven't seen the inside of a butthole but I would imagine it looking womewhat similair)

guess I'll trey it again =)

well now I've adjusted it, right image looks even better now

 

[Simon F]

^^ me agrees

At first I though people were blind when they didn't see that "the right pic looked obviously better" guess it was just my gamma

as Ichy I too don't really know how to config it properly, where' the noobie guide?

Essentially for each of the red green and blue areas in the test image adjust the gamma ramp for that component (using your hardware's control panel) until the solid colour appears to be exactly the same brightness as the dithered area around it.

When you have done this for all three components the white regions should also be correct, and your display is correctly adjusted.

Oh and DONT have your brightness on your monitor cranked up too high!

 

[Dio]

The way I set the brightness is to set it by the black floor - get a large region of black on the desktop, then crank the brightness right down. Increase it one step at a time until you see the changes in the black then back it down one step.

Then I adjust the contrast to set the gamma ramp (or, as an alternate way of looking at it, the max brightness of the 'white' bits) I want. Also on some monitors like the latest Iiyamas you may have to tweak the 'video mode' setting off or getting good control can be very difficult.

Mine was set to about 2.0. I found the best way to handle the problem with the checkerboard was to take my contact lenses out, that blurred it very nicely

Note that some monitors and video cards may be set differently in different resolutions. In particular I have seen monitors which have noticeably non-black black floors in fullscreen DOS prompts or other low resolutions.

 

[Ante P]

Well at anything but the lowest gamam availible in my settings the right pic looks much better (insanely much better with a high gamma).

I dunno, stuff looks pretty good as it is but I should familiarize emyself with this if I want to do some serious IQ comparisons of videocards/monitors in the future I guess.

I foudn that Digital Vibrance actually did my IQ some good when I had my gamam lowered. but just half way to the low setting, anything higher makes colors look like horrible neon signs from teh 80ies.

 

[demalion]

^^ me agrees

At first I though people were blind when they didn't see that "the right pic looked obviously better" guess it was just my gamma

as Ichy I too don't really know how to config it properly, where' the noobie guide?

Essentially for each of the red green and blue areas in the test image adjust the gamma ramp for that component (using your hardware's control panel) until the solid colour appears to be exactly the same brightness as the dithered area around it.

When you have done this for all three components the white regions should also be correct, and your display is correctly adjusted.

I assume you are talking about the color image Simon provided and not the selection of test images in the ATI control panel, while at the same time talking about adjusting the control point in the ATI control panel (since none of the test images there I see look like a "solid color" and a "dithered area")?

May I strongly recommend both a selection of solid/dithered adjustment images to the cycle list of the test images, and a "What is This?" explanation (if it can be changed on the fly it would be ideal) explaining what the heck to do with each of the test images? This would be a terrific boon. This suggestion has already been submitted through Catalyst Crew feedback form.

 

[Nite_Hawk]

Crusher/etal:

This is pretty interesting. At home on my Samsung 191T LCD display, the ATI 6X images looked much better than the Geforce FX 8X images. Here at work on a KDS VS-21E, the FX images look much better than the ATI images. I wonder what exactly is going on. I have a 9700pro at home, so I guess I'm lucky that the ATI images are better looking than the geforce ones and not the other way around, but that's pretty odd.

I mean, it's really pronounced. It's pretty much a clear reversal. On the LCD display, the ATI images are very smooth, and the FX images show starcasing. At work, it's the FX images that are smooth and the ATI images showing staircasing.

Odd....

Nite_Hawk

 

[Gollum]

That's really interesting, I also have a similar effect! Here at home the right one looks clearly superior to me, yet I could swear it was the other way around at work! <scratches head> Guess I gotta do some gamma adjustment on both monitors...

 

[Basic]

Here is the image I talked about earlier. Don't complain about my lacking artistic talent or dry fantasy regarding the contents iof the image. It's the idea that counts.

 

[Bambers]

However,

My feeling is that 8X is actually 4X MS + 2X SS. Why? Because the NV30's 2X is RGMS, and it's 4X is OGMS, and since we know that 8X is ordered grid, I can't see why they would use a different 2X MS sampling pattern for the hybrid mode than they would for the striaght 2X RGMS.

Of course, perhaps there's a perfectly logical reason why they would, and I'm just to ignorant to know any better.

I don't think nvidias definition of ordered grid is the same as everyone elses. The 8x mode on the gf3/4 is certainly 2xRGMS with 4xOGSS. If you look at the sample pattern over several pixels it all lines up as a contiuous grid. because its 2x2 supersampling the grid is not squashed and is square (although at 45degrees) and this is why I think it got called ordered grid. The 4xS and 6xS modes also have a continuous grid but it is squashed and I think this is why they call it skewed grid (and presumably where the S comes from).

Looking at the big pictures (right click and save target as apparently gets past geocities bandwidth restrictions , I've put them on my blueyonder webspace r300 and nv25) With gamma turned up quite a bit on my monitor the nv25s 8x is better on the more vertical lines, while more default gamma the r300 is better. On some of the horizontal bars though the nv25 is getting quite bad aliasing that makes it worse whatever my gamma is.

 

[McElvis]

Here is the image I talked about earlier. Don't complain about my lacking artistic talent or dry fantasy regarding the contents iof the image. It's the idea that counts.

Interesting!...

 

[McElvis]

Looking at the big pictures (right click and save target as apparently gets past geocities bandwidth restrictions , I've put them on my blueyonder webspace r300 and nv25) With gamma turned up quite a bit on my monitor the nv25s 8x is better on the more vertical lines, while more default gamma the r300 is better. On some of the horizontal bars though the nv25 is getting quite bad aliasing that makes it worse whatever my gamma is.

Looking at your images (thanks ), with my gamma corrected as in this thread, I am seeing the same results. The R300 picture looks a lot better. The white-top pipes in the left-lower-middle of the picture on the NV25 look awful.

 

[WaltC]

The images look pretty much the same to me. I'm not inspecting them closely, just looking at them superficially.

The thing about screen shot analysis that I hate, especially when it concerns a 3D scene, is that depending on the angles of light and all sorts of other things a frame grab of an object in one frame might appear very different from a grab 3 frames later--yet to the eye when the scene is in motion no distinction can be seen such is evident in the resulting screen shots.

I've used for years a little display calibration program called NTEST.EXE, a little program put out by Nokia years ago when the company made monitors (I had one--don't know if they still make them.) But it's a great little program that uses color bars and other things to allow you to easily and correctly adjust your gamma, brightness, contrast and color levels--and a whole lot more besides. And, the best thing of all, it doesn't do any of the stupid stuff those hokey (and *rotten*, IMO) color calibration programs do when they make you put a paper overlay on your screen, turn out the lights, and all of that goofy stuff... This is the best one I've ever seen, IMO.

BTW, not only are people's monitors and videocards different in subtle and sometimes not so subtle ways--people's *eyes and brains* do differing things with color and contrast information. I use the various sub programs within NTEST to calibrate for my eyes and brain on my screen, and afterwards trust that everything else I view looks as it should. And it does, far as I can see.

 

[Xmas]

Nick,
As for colour, this image is my version of your link tuned for a gamma of 2.2

To make colors match in this image, I have to set gamma to about (R/G/B) 0.92/0.84/0.82 in the driver panel.

That's with 'digital vibrance control' off. If I set it to 'low', I have to put down gamma even more to make the colors match, but grey won't fit any more in this case. What exactly does dvc do?

I have been using gamma 0.8 for a while now, with dvc low (I like that), but i set it to 1.8 in games like HL to see a bit more in dark areas. ClearType also looks quite good with gamma 0.8.


Oh, and I'm not sure which one of the small images I like more, the left one seems to have too big a jump from medium grey to black, the right one is exactly the other way round. Looking at the full screen images, I prefer the 8x shots.

For 8x sample pattern, have a look at this: http://www.3dcenter.org/artikel/2002/11-24_b.php

 

[Simon F]

SamX,
I suspect that all you are doing is resetting the DAC so that the output voltages are not a linear function of the RGB values.

AFAIU the default on PCs (workstations are often different) is to have a 1:1 relationship between RGB value and output voltage. The monitor's own non-linear behaviour then gives you an overall gamma of about 2.2

Perhaps you should first set the graphics card's internal values back to 1, then looking at the grey scale test image, set the monitor brightness to, say, about 1/2 and then try to adjust it so that the pattern matches a grey in the 2 to 2.5 region. (Stand well back when you make the comparison).

You should then be able to adjust the contrast to get the full range of brightness.

After that you can probably individually tweak the colours using the graphics card's control panel.

If I get a chance, I'll create R, G, and B monochrome versions of the gamma estimation image.

 

[Xmas]

SamX,
I suspect that all you are doing is resetting the DAC so that the output voltages are not a linear function of the RGB values.

Yes, of course that's what the driver panel does, write to the gamma table right before the DAC.

Perhaps you should first set the graphics card's internal values back to 1, then looking at the grey scale test image, set the monitor brightness to, say, about 1/2 and then try to adjust it so that the pattern matches a grey in the 2 to 2.5 region. (Stand well back when you make the comparison).

You mean that greyscale image: http://simonnihal.homestead.com/files/assorted3d/gamma_test.png ?

If I set everything in the 'color correction' panel to hardware defaults, I'm not able to get that pattern match a grey in the 2 to 2.5 region, no matter how I set brightness and contrast on my monitor (I should add that I'm using a Samsung 181T TFT display).

 

[Simon F]

You mean that greyscale image: http://simonnihal.homestead.com/files/assorted3d/gamma_test.png ?

Yes that's the one

If I set everything in the 'color correction' panel to hardware defaults, I'm not able to get that pattern match a grey in the 2 to 2.5 region, no matter how I set brightness and contrast on my monitor (I should add that I'm using a Samsung 181T TFT display).

I suppose the TFT has something to do with it, but my (HP) TFT at home seems to have a typical 2'ish gamma.

I would have assumed that the manufacturers would design them to have similar characteristics to CRTs. Curiouser and Curiouser (said Alice).

 

[Bigus Dickus]

[If I set everything in the 'color correction' panel to hardware defaults, I'm not able to get that pattern match a grey in the 2 to 2.5 region, no matter how I set brightness and contrast on my monitor (I should add that I'm using a Samsung 181T TFT display).

Shouldn't Gamma be set in the video driver's control panel, not by adjusting brightness and contrast on the monitor itself!?!

 

[Simon F]

[If I set everything in the 'color correction' panel to hardware defaults, I'm not able to get that pattern match a grey in the 2 to 2.5 region, no matter how I set brightness and contrast on my monitor (I should add that I'm using a Samsung 181T TFT display).

Shouldn't Gamma be set in the video driver's control panel, not by adjusting brightness and contrast on the monitor itself!?!

AFAIU, the gamma of CRTs is a function of the electronics, and turning up the 'brightness' control tends to flatten the voltage->intensity curve (i.e. lowering gamma). Since the 'standard' gamma of a CRT (~2.3ish) is reasonably close to the inverse of the eye's behaviour you might as well take advantage of it and get your monitor set up in the correct range (by adjusting the brightness).

Of course LCDs may differ, but I have found that mine at home has a gamma of 2ish.

Poynton gives details on setting up a CRT (it might also apply to some LCDs as well)


I suspect the problem with computer graphics is that many lighting calculations are done assuming a linear display. OTOH, I also suspect that the textures are probably stored in non-linear form (i.e. perceptually linear space). Since you get the best visual range with a non-linear setting, that would seem to be the better choice for setting up a display.