Basic Guide to Color Calibration using a CMS (updated and enhanced) - Page 6

Looks like you got it - I think as you watch it more you will realize the secondary gamut is most important.

on the Hitachi vs810 LCD RPTV, the user menu has color decoding, and color management sub menus.

controls in the color decoding menu

Red
Green
Color
Tint

controls in the Color Management menu

Magenta -> Phase & Gain
Red -> Phase & Gain
Yellow -> Phase & Gain
Green -> Phase & Gain
Cyan -> Phase & Gain
Blue -> Phase & Gain

following the instructions here for color and tint will not involve the Red and Green controls in the decoding menu. The Color control has more effect on the Blue than Red or Green. Once all the primaries and secondaries are as close to the CIE chart as possible, skin tones have a clay hue to them.

Should I use the color decoding menu controls to set blue Y to 8% of white, red Y to 21%, and green Y to 71%, then follow through to the color management? or ignore the decoding menu and just use the color management settings?

thanks
vega

You are making the common mistake of using the color controls to sacrifice color brightness (the decoding) significantly enough that it impacts color saturation (the gamut). You have to measure xyY when you adjust - not just xy. It does not appear you have any color saturation control - you cannot shrink/enlarge your gamut properly.

The color decoding menu is for your video - the color management is for your display. You can isolate the display by avoiding the color decoder if you have a reference RGB source, unless yours is one of those that runs RGB thru the video codec.

thanks Tom and krasmuzik for your replies .......

Phase moves the color toward one of the other 2 adjacent colors. eg Phase for the blue, would allow adjustment toward cyan, or magenta. The gain controls, as far a sI can tell, adjust the intensity of the color. I'm not sure how this differs from the color decoding controls.

this set's CMS has a lot of adjustment. The green was way too high on the y axis out of the box, and I can bring it down, but the Y value seems to go well below where it should be. I believe that on this set the green cannot be reigned in without grossly under saturating it.

I read over the Calman instructions and remember seeing Bill mention that some digital sets have one primary over saturated, and there is not much to be done, other than to find a good compromise.

If I try to get the green to a compromise, which value (x, y, or Y) is the most important?

You should be using dE as your error measure rather than optimizing one thing over the other. dE gives the perceptual error in the xyY measures (if implemented properly).

Usually if you do not have proper controls you cannot fix out of gamut primaries - but turning down the color only a bit can make them appear less annoying.

Tom

adjusting the gain for any of the colors does move the position on the CIE chart. I tried using the decoding menu as follows

set color Y (blue) to 8% of white Y
set red Y to 21%
set green to 71%
adjust tint to get cyan as close as possible to it's CIE location

then in the CMS I adjusted each color to get the lowest dE. The color has improved a great deal. All colors with the exception of green have a dE less than 2, green is 10.8, and shows well above the mark (y value high) on the CIE chart. I can move the green to the location on the CIE chart, but dE goes to 37.

I repeated the process 2 more times, and color improved each time. I did notice the adjustments were getting smaller with each calibration.

Which green setting would be closer to correct, the lowest dE as krasmuzik suggests, or the closest location on the CIE chart?

dE increases because your Y value drops drastically when you moved the gain control to get it over to the right x,y. As noted earlier this gives watercolor video - you don't want to do this. Go with the lowest dE. Getting the hues corrected on primaries, and secondaries on target will make for a good picture once you have corrected the brightness decoding.

When I change the gain control in CMS, x,y, and Y all change. when the phase is adjusted, only x and y change. The green control in the color decoding changes Y the most, and seems to have a larger effect on Y than the CMS. Changing the color decoding values has a minor impact on grayscale, the CMS adjustments have less or no effect on grayscale (I'm guessing this is a good thing).

I have repeated the above steps three times, and have seen an improvement each time. I also see the adjustments are getting smaller each time. Blue Red and Magenta did not need any change on the last run.

I'll try your suggestion, and post the results tomorrow.

thanks
vega

Well it's a mixed bag, the CMS and the color decoding adjustments both change x,y, and Y values, but in different ways. The Y value is predictable, move the control to the + and it increases. The x, and y values differ, in CMS y changes are more pronounced, in the color decoding, x changes are more pronounced (this is also changed with the phase control). I managed to get dE down to 8.1, at the expense of cyan and yellow increasing to @ 2.8. Anything lower and the green looks washed out. It may be my eyes, as I have had this set for over 2 years, and to this point only calibrated the grayscale.

I did notice the color control will affect all colors, but blue by far the most. Should I try to move the color control higher and correct the levels for all P&S in the CMS? OR would I be spinning my wheels here? How would I determine the best trade-off at this point?

thanks
vega

Reason is Tint/Phase Color/Gain controls are simple math - adjusting the gain is a multiplication of a component - adjusting the phase multiplies two components . A CMS requires on the fly 3x3 matrix math - with a user interface that converts from perceptual adjustments back to matrix coefficients.

thanks for all the help, I've learned a lot, and have improved the color on my set while doing so. Looking at some of the other posts here, and other threads, I guess I'm lucky to get as close as I have.

vega

Hi,
I tried to calibrate my pio 5080 using HCFR and a spyderII with DVEHD. Not a good experience. Pretty much just getting garbage numbers for the primary and secondary color. negative x and ys. That sorta thing. Strangely enough with the filter off I can get what appears to be a very reasonable grayscale. Any thoughts???

Does the i1pro have stability problems when it starts to heat up? Any special problems using it for plasmas? I've read here that it is a pain to use. Some pro calibrator claimed in this forum that for most sets he'd just use the D2 even though he also had the I1pro.

OK, I used Tom's method for setting the RGB to the designated percentages of white. Why do we change to using the x,y dE as the guide for CMY? What happens if you use the percentage of white for the CMY? Some of the dE's I was seeing were in the 150-300 range. Changing hues did not seem to affect the dE as much as it affected Y. I ended up setting CMY to the proper percentage in spite of the dE and the picture looks great. I don't get it.

VM

I am using Sencore CP500 software to calculate dE. I guess my question is "How can the Y be on the mark with x,y so very far off?

Thank you, Tom. I'll check with Sencore.

OK, I did check with Sencore. They're reply is -

"True delta E does take into acount the full difference in the appearance of
two colors, including their luminance. This is usually calculated in a more
uniform color space than the CIE 1931 xy space.

For our purposes, we are concerned only with the chromaticity differences
between a reference white point, and the color of white achieved by a
particular display, so we ignore the luminance. We designate the metric as
dExy, which we calculate as sqrt((x1-x2)2 + (y1-y2)2). This tells us the
coordinate distance, on the xy plane, that the chromaticity of one color is
from another."

Tom, you didn't indicate the software package you use. What is it? Is there a work around for software with just a dExy?

You can use the spreadsheet provided by Krasmuzik found here , which will calculate deltaE in the more perceptually uniform CIELUV colorspace and includes the effect of errors in luma. Currently the colorHCFR package calculates deltaE using only chromiticity error which is fine for grayscale calibration as sencore states. The next version of this software includes the CIELUV based deltaE.

Hello gang, I have read here and on other topics that the CMS on the Pioneer elite's don't work. I would have believed that before but not now. I have calibrated my 1130 using the CMS controls and get what I consider a very good picture. But it only seems to work well in one of the many different picture modes. not sure why. Anyway my question for Tom is I have done the 75% white/red method for setting color. Tint is a little different because of the CMS it moves. Not a problem because you can move it back. Why can't you use a software program to set the secondary colors? Could you use equal parts of red and blue to get magenta, and equal parts of blue and green to get cyan? Or is it not a question of equal parts of the two colors but there location on the CIE chart. When I get a moment I will include my grayscale for the Pioneer.