a question about dithering

[emanresu]

Hello,

This may not be directly related to photography per se, but I have been wondering about dithering and equivalent color values.  When calibrating a monitor, we often see 4 color squares: white, red, green, and blue.  For each square, the outer rim is filled with the uniform color with channel value(s) somewhere in the middle between 0-255, and in the center of each square, it is usually a dithering of the channel at full intensity (255) interlaced with black.  The object of the calibration, or at least a part of it, is to adjust the monitor's (gamma? brightness?) so that the outer rim and the center look identical when viewed form afar.

I always thought the value of the uniform outer rim is just 127, which is half way between 0 intensity and full intensity.  Say I have a small 2x2 square dithered so two of the cells have full intensity for one color channel (and 0 for other two color channels), while the other two cells are pure black.

[255][000]
[000][255]

so my thought is if I blend the 4 cells into a single cell, I would take the arithmetic mean the values for each color channel, i.e.  (255 + 0 + 255 + 0)/4 = 127

But that doesn't seem to be the case after I created such an image.  The dithered center looks obviously much brighter than the sides.  Then I found a calibration image from the Internet, and on that, 186 is the value for the color channels.  I just cannot understand the reason behind this value, even though 186 looks almost identical to the dithered center.  Anyone care to enlighten me, please?

[Bob Atkins]

I believe you would be correct (uniform area value = 127 to match a black/white checkerboard pattern) for a gamma setting of 1.0

However PCs are usually setup with a gamma of around 2.2 so the "uniform area" has a value of 186

 (186/255)^2.2 = 0.73^2.2 = 0.5

Black and white or max/min values  (i.e. areas of 0 or 255) are not affected by gamma, so if you have an area of equal black and white squares or stripes that average out to a luminance of 0.5, you need a gray area of luminance 0.73 (186/255) raised to the power of 2.2 (gamma) to display at an equivalent brightness.

Hope that makes sense!

[emanresu]

Thanks Bob!  Mathematically this makes perfect sense.  However, when I look at my monitor's (or rather, the graphics card) setting, it says the current gamma level is 1.0; if I bump it to 2.2, it is very bright and looks washed out.  So is this a different gamma?

[KeithB]

Are you changing gamma for the graphics card or the system?  The OS may have an internal correction that happens before it gets to the graphic card.  Here is Apple's recommendations:
http://support.apple.com/kb/HT3712

[emanresu]

Hi Keith, quite frankly I dont know -- most likely it is for the graphics card because I was referring to the gamma setting on the nvidia configuration panel, but does it matter? Because the laptop LCD has no adjustment controls.

[KeithB]

If you are using windows seven, rightclick on the desktop and from the dialog that appears press advanced settings.  That will get you into the system gamma setting.

[emanresu]

There seem to be some missing steps, Keith.  Assuming you are referring to the "Advanced Settings" from the "Screen Resolution", windows 7 doesn't tell you anything about the system gamma setting, but usually the vendor's video card configuration panel can be accessed from there, and that is where it says gamma = 1.0

[KeithB]

Sorry, it is "what display settings should I choose" and it does not tell you what gamma is, just provides a way to set it for your monitor.  You could go in there, radically change the gamma and see if it changes the graphics card setting.

[Bob Atkins]

I very much doubt you are running a gamma of 1.0. By default the PC uses something close to 2.2. As Keith suggests it's possible the system is running 2.2 even though the graphics card is 1.0. i.e. they system is making the gamma correction and the 1.0 gamma of the card is essentially "no correction".

Without full details of your system (and maybe not even then!) it's hard to say what's going on, but if you use one of the standard gamma test patterns you will find out what your overall gamma setting is.

I'd suggest looking here - http://www.normankoren.com/makingfineprints1A.html. Norman's page is a bit technical, but there are test patterns which will indicate your current gamma setting. They are based on the same principle as I outlined above.

[emanresu]

Thanks Bob and Keith.  I looked around, and many of the monitors don't have gamma settings... but I did come across one dell monitor that has this "Mac mode" and 'PC Mode" for color settings, and one is slightly brighter than the other, so that must be its dumbed-down version of gamma settings