I assumed for some time that the user Tint control adjusted the relative intensity of the secondaries rather than the actual Hue of blue, but now I'm not so sure. I noticed that as you adjust Tint, the secondaries move laterally on a CIE chart quite responsively. This suggests that the Tint control directly adjusts the hue of the secondaries.

Yes?

"Hue" alters the phase of the subcarrier, relative to the reference burst.
This only applies to composite and s-video, which have colour subcarriers.

Look at the blue filter test - cyan&magenta are balanced if they contain the same intensity of blue. Magenta&yellow are balanced if they contain the same intensity of red as observed in the red filter. Therefore magenta is a balance of red & blue and the tint control will slide it along the gamut edge. Likewise for the other secondaries.

Since the historical analog use of tint for composite systems as a phase adjustment to extract the color difference signals no longer applies to component - tint is usually implemented for component to have the same result of adjusting the hue of secondaries by rebalancing the component levels. Not all component inputs have tint controls - as it is an extra manipulation not needed if the signal was just left alone to begin with.

Look at the blue filter test - cyan&magenta are balanced if they contain the same intensity of blue. Magenta&yellow are balanced if they contain the same intensity of red as observed in the red filter. Therefore magenta is a balance of red & blue and the tint control will slide it along the gamut edge. Likewise for the other secondaries.Kraz: You lost me here. If it is an intensity control, then like the Color control, small changes in Tint wouldn't reveal itself as a change in xy coordinates. But it does.

So how can changes in Tint be both an intensity control AND cause the secondary to move "along the gamut edge?" On the other hand, if it is a Hue control, then why is the instrument for seeing it a filter that reveals a balance of intensity between colors?

I am officially clueless on this.

Tom,

The tint is changing the relative blend of the primaries in each secondary. This will alter the intensity of the primary in the secondary sliding it along a line between the primaries in the xy coordinate system.

In the same manner that D65 is a blend of three primaries intensities to get a color of white - a secondary color is simply D65 white less one of the primaries to get a hue for the secondary. But we don't want to screw up greyscale to muck with tint - so we will do it earlier in the color decoder.

Lets look at Poynton's definition, which for analog composite is that given already (mucking with phase of subcarrier) - but he also points out that phase is a rotation of the UV diagram (vectorscoped color differences obtained from composite color - this is not the same as CIE u'v' colorspace) Remember matrix math - you effect a rotation thru magnitude manipulation? Now I refer you to component video color coding for SDTV/HDTV chapters - given YUV you can get the component color differences (YPbPr) through scaling and from there get to RGB through color decoder matrix - so you can essentially emulate this historical YUV rotation by mucking with the matrix coefficients. You really have to understand matrix math to follow these chapters!

You can also read the chapter on NTSC and PAL chroma modulation to understand how an analog circuit for tint needs to account for phase but PAL does not. He does not go into how to do a tint control for DTV - simply because there is no need for one (until some marketroid said where is the tint control!). Once you understand all the matrix math - you realize that is a lot of digital math to emulate a simple phase adjustment - and now you will realize why tint control is often missing in digital displays.

Kraz: You lost me here. If it is an intensity control, then like the Color control, small changes in Tint wouldn't reveal itself as a change in xy coordinates. But it does.

So how can changes in Tint be both an intensity control AND cause the secondary to move "along the gamut edge?" On the other hand, if it is a Hue control, then why is the instrument for seeing it a filter that reveals a balance of intensity between colors?

I am officially clueless on this.
Jeff and Kevin have it down. For me, I prefer to think of these as saturation and hue, since these have specific meanings when it comes to the CIE diagrams. Saturation ought to change the amount of the "off" primary that is present in a secondary. Adding more moves it closer to white (desaturating the color), and taking it out moves it further away (saturating it). Hue changes the mix of the "on" primaries that combine to form the secondary (think tug of war).

Bill

OK, I think I've got it. I guess I was right in my original assumption. I was just stuck on the fact that an adjustment of Y would have such an obvious effect on the xy of a secondary.

BTW, the reason I've been posting a lot about this stuff recently is that I ran across a display that did not have color isolation controls and for which a blue filter and SMPTE color bar yielded an obviously wrong result for Color and Tint. Hence, I sought some absolutely reliable methodology for setting these user controls other than filter/color bar.

Adjusting Color until a primary matched the prescribed % intensity of reference white seems to work (e.g., 21.3% for red) and getting good alignment of the secondaries on a CIE chart appears to get you Tint.

Thanks for all your input.

This is a very informative thread.

So when your typical TV is doing this decoding, they tend to offer only a single Saturation/Hue. My understanding is that these are the BLUE ones. As a result, with this Hue, you have the ability to 'tug' Cyan and Magenta away from or toward Green and Red respectively, correct? Saturation being the ability to push/pull yellow away from your D65 point?

If I have this right, then the ability to have these controls for all secondaries would allow you to completely 'fine tune' your decoder....basically the attempt to get your secondaries closer to your target....but then again.....target being.....

Which will lead me into one more question. To say a particular decoder is 'spot-on' seems to describe the ability to create a secondary that is both 'in-line' with its two primaries while being able to draw a straight line between this secondary and the 'off' primary while crossing your grayscale point (D65). It does NOT seem to be the condition that this secondary is at the 'Rec standard'.....only that they agree well with your primary-created gamut.....whether your primaries are accurate or not.

Any feedback is greatly appreciated....thanks guys.

Saturation being the ability to push/pull yellow away from your D65 point?The saturation control, usually called the Color control, generally adjusts the intensity or luminance of all of the colors. This has a secondary effect on saturation, pulling the colors towards white.

Look at the blue filter test - cyan&magenta are balanced if they contain the same intensity of blue. Magenta&yellow are balanced if they contain the same intensity of red as observed in the red filter. Therefore magenta is a balance of red & blue and the tint control will slide it along the gamut edge. Likewise for the other secondaries.

As I reread this, I believe it answers my last question. For your decoding to be correct, it basically fulfills these things. So the result is accurate according to your color gamut created by your primaries, whether they're correct or not.

Is modifying your decoding for the purpose aligning your secondaries to be inline with, say rec709, a 'no-no'? In other words, is it more important have a decoder balanced within your measured-primary-created gamut then to 'warp' them into being closer to NTSC/ATSC standards?

Warping the secondaries to be correct even if primaries/grayscale are not perfect is something the Infocus PJ's do - switch to 7500K or 9300K and the secondaries don't move. Since the secondary gamut is where most natural colors are - this can be very effective! Though I would suppose if you are using it in Japan for TV viewing that would be the wrong thing to do.

I think minimizing overall color error in the display is the goal - rather than probing the decoder to see if it is reference values.

Are you serious? Changing the gray scale does not affect the coordinates of the secondaries? Really?

Changing the grayscale preset does not - however changing the RGB greyscale calibration controls does.

Some on the forum have said that with Infocus they did not think 7500K was all that bad - and this is why. Only near whites are unbalanced - the secondaries are still on the SMPTE targets.

that's funny....several years ago, an infocus proj is exactly what i used....in 7500k. Not that I knew a single thing about video-calibration then....but I certainly thought it looked good.

Ok, last couple questions, I don't want to get too OT.

With my plasma, green is way off. So Cyan is 'left' and Magenta (after D65 calibration) is toward Red. Oddly, Yellow is perfect. So an option in my case would be to use a cooler white-point....thus placing my magenta more in-line. I don't see this as affecting my cyan/yellow too much at all. Does that sound right?

Alternatively, I could decrease my Cyan saturation (which puts more red in it), and play with blue Hue to creep my Magenta closer to spec....hopefully without compromising cyan.

Does this sound like possible directions to take? I don't think #2 is an option being I feed my inputs with RGB....thus saying bye-bye to decoder options (although I haven't tested the chroma settings in the service menu yet)

I use CIE LCH to make those decisions on which is the least tradeoff error to make when it can't be perfect - since it uses perceptual color measures. See the spreadsheet I linked in the calibration sticky thread if your software does not have that colorspace.

Since we are white adaptive if you can error greyscale to fix your secondaries that might be a good compromise.

Colin,

You can also use HCFR L'u'v' color space. Handy spreadsheet Kras, thanks!

Lu'v' is a step in deriving LCH - which is actually based on Lu*v* - the difference between Lu*v* and Lu'v' is that Lu'v' is capable of a CIE gamut diagram which gives color areas equal perceptual weight (unlike the CIE xy diagram). Lu*v* is perceptual color differences around a target and thus cannot have a CIE gamut diagram - the distance from the target combines saturation and lightness w.r.t. ref white which corresponds to how your eye perceives color - LCh is just the polar version of Lu*v* and LCH is the modified form that makes dE measures consistent across color spaces (your eye does not see degrees or radians of hue angle!)

In order to do LCH properly in calibration software would require having a calibration target rather than a gamut display - as well as retaking the white measure every time an adjustment is made. But LCH does properly indicate how your eye perceives color and is very useful when making color management system tradeoffs.