Basic Guide to Color Calibration using a CMS (updated and enhanced) - Page 2

The AE1000 has a functioning CMS. It has controls for Color, Tint and Brightness.

I missed this comment my first pass, but this is a point I tried to make a while ago here:
http://www.avsforum.com/avs-vb/showt...&&#post6637132

I have also posted a critique of the dE(u'v') method as going too far in the other direction, but that is for a different thread...

Bill

For EACH COLOR . (Sorry, I was a bit short there) One can store up to eight different colors per memory.

Tom,
The Optoma HD7100 has a full working saturation, tint and lightness adjustment per primary and secondary color. These adjustments are in the SM.
Despite all its adjustments, I do not seem to be able to get green to its theoretical x,y value in the gamut. Could this be a result of the grey screen I am using?
Any comments/suggestions?
Thanks a lot to you and all other contributors to this great thread.
Fermin

Hi Tom,

Excellent write-up!!

And since I finally had a couple of free hours last night I decided to calibrate my Sharp Z12 using your exact procedure outlined above. For the most part everything worked great. I was able to adjust color and tint as you outlined, and only had to make minor adjustments to color bias and gain in order to get the tracking under dE of 3 across the board (except for 0 IRE, which I know is below the effective range of the EyeOne Pro). Then I finished up the color decoding by adjusting green and blue...so far so good.

The only problem I ran into with the Sharp was when it came to adjusting the gamut - Green was a problem, and because of the green problem, cyan is not right either. I was able to move all of the other colors so that they are right on the CIE targets, but green is undersaturated (if I am interpreting the chart correctly - green was originally outside the triangle, but lower than the green coordinate and way to the right), so I have no way to correct it (and cyan is undersaturated as well). Despite that little problem, I double checked everything else a second time and the color decoding is very, very good, as is the grayscale tracking.

My subjective impression is that most colors look absolutely dead accurate, though the greens looks a bit "pale". This might be because I know that green is not quite right, or it just may be due to viewing oversaturated greens for so long - I really don't know, but either way I am EXTREMELY happy with the results and the CMS functions of the Sharp. It was a real pleasure to have that much control over color!

Many thanks again!

That sounds like a color replacement system to me. You have to run the gamut tests to prove the entire gamut actually changes - not just one color.

A true color management system will do a 3x3 matrix transform to warp the gamut based on the editing of RGB parameters. Having a lookup table to remap a few colors is not at all the same thing.

Thanks, Tom, I'll give that a try tomorrow night (my next opportunity).

Since I don't have a chart to post, I'll try my best to explain it. The green primary was initially located to the far right (like half way between green and red horizontally) and slightly lower (vertically) than the targeted primary. When I tried increasing saturation, there was no change in the reading at all, but when I lowered saturation the point dropped lower by quite a bit. When I tried adjusting hue in one direction I could move the point to the left, but the very left most I could get it was on a line between the green and red targets, which is where I left it. Cyan acted similarly in that it was inside the triangle initially, and no amount of change would allow me to place cyan on the line between green and blue - it was and is still inside of the triangle. These observations were confirmed visually as I watched the 75% green and cyan windows from the Accupel. Green looks "pale" for lack of a better term and I could not get it to look any "greener".

I started with 6500k as a starting point. Maybe I will try measuring a few other temperatures to see if there is a better starting point for calibration. I'll also double check to see if I did something stupid in feeding the signal from the Accupel to the 12K's DVI input, especially since you got better results than I did.

Thanks again for all of your help!

I have, and it's easy to move all colors into rec.709 coordinates.

Before CMS:



After CMS:

Tom are you talking about AVIA PRO's? There is a checkerboard display version of it that it is designed to be viewed on a PbPr component difference (or maybe was NTSC IQ?) vectorscope as grid. But knowing the target display is SMPTE-C and the DVD coding is REC601 you could do the matrix math to derive the xy from the geometry math for a regular stepped grid on a vectorscope (yes had they documented it would have been useful - one of these days I will write a MatLab program to figure the numbers out - Maybe Bill has some free time to R&D a CalMan gamut tester synced to AVIA PRO!). You are on the sequential version you step thru with window fields for making measures. But you can visually use the checkerboard version just to see what colors the adjustments impact. If you can't find it I will look up the menu path.

Certainly DisplayMate full gamut pattern http://www.displaymate.com/dwsct.html or any softwares full color CIE chart http://www.colorvision.com/images/Co...rumentCIE2.jpg is easier to see the results of color editing than making tedious measures. These smaller snaps may be hard to see - but those with these software should try it with the charts full screen.

anbjornk

Does editing red shift ALL colors with the red component - or like the Sanyo does it only shift a very narrow range of Red?. The Sanyo also has only enough framegrabber slots to get primaries, secondaries, and maybe a couple flesh tones. But look at the full color gamuts and you still see slivers of uncorrected colors.

I only see seven colors (RGBCMYW) on that CIE chart - and you have only x,y and thus missing Y data - which means it is very possible you compromised color brightness trying to get color gamut correct. But unless you look at a full color pattern - you cannot even prove that you fixed the gamut - you may have just fixed seven colors in it. The point is that while a CRS can remap any very limited set of colors - with a CMS you are editing the very locations of the RGB primaries the display produces to warp ALL of the colors - and some even further warp the secondaries.

Obviously it becomes crazy to measure orange, crimson, brick red, flesh tones, etc - so I am just saying just look at charts that have the full range of colors and play with the controls.

Actually, even though green supposedly has moved quite a lot in the CIE chart, I cant't actually see the difference in videomaterial. The shire in LOTR looks all the same. The only obvious change is in red/orange, where orange has been less orange. This can be seen easily on leloo's hair in the fifth element.

Btw. I used only the color and tint adjustments, and left brightness alone.

I think you guys are correct, it's a CRS rather than CMS.

This is why CIE L*u*v* is a better colorspace than xyY - green hues are not as easy to see as red and blue, even in video with green being brighter than red and blue! Whereas on your xyY chart you can barely even see the hue is off on red even though your eye could see it was too orange.

I can back up what Bob is saying about the greens on the 12k. I had the same exact experience.

Rob

Here's the HCFR file with the Spyder2. While the red is purposely off the mark, the green and the blue could not be adjusted exactly to the standards. No matter what I did, I could not get the green any further to the left, without moving it significantly towards white. I got very similar results from the Xrite DTP-94, although the green was slightly closer.

Rob

 

spydercust1lcd.zip 0.94921875k . file

Tom,
The attached gamuts are the before and after CMS. The default on saturation is 0 at its maximum. Green tint will move the green position into the triangle, under the green reference point. I don't know what can be done, if anything, to get the green color in its ideal place.
Readings made from the screen.
Any suggestions?
Anybody?
Thanks.
Fermin

Tom, first of all great thread and thanks for the tips! In the EMP-TW1000 thread you showed a post calibration chart. I am now going through and calibrating my Epson 1080 and if possible could you post or pm your settings to me to use as a reference so I can see how they compare. I known mine will be somewhat different.

Thanks !!!

Tom: thanks for your explanation and process.

One thing that I have been wondering - when altering the grayscale cuts/gains for each of the primaries, does changing each of the 6 controls have varying effects on the resulting x,y,Y numbers? And do these vary by display type/manufacturer/model?

I've been using the Spyder Platinum with the Ruby and CalMAN. Been a bit of shooting in the dark. I'm trying to get a better handle on what's happening as I apply changes, so I do this in a more intelligent manner. I've found that I can go down the wrong path with changing things (too much in one direction or another and impacting more distant IRE stimulus areas). Your approach of picking 30% and 90% works well. It's when I'm trying to move down to 20% -> 10% that I get into trouble (I won't even bother with 0% since I don't think the Spyder's up to it).

Any help would be appreciated.

Here's the gamut of the 12k with the DTP94. These are the same settings from the Spyder, that are shown in post #54 above. They are almost identical except for the green (and subsequently, the cyan), which measures worse. As with the Spyder, I could not get the green anymore to the left without moving in towards white. Considering these two graphs, I'd say the Spyder is a worthy contender considering how inexpensive it is. It's not perfect, but neither am I.

Rob

Sorry, here it is. See, I told you I wasn't perfect.

Rob

Robbyc30, assuming that the white lines in your above graph are the measured values, then your gamut looks almost exactly like mine. Mine just looks worse because the the graph is larger and easier to read...

So, is there something wrong with Bob's and my 12k? Is it possible my greyscale settings are affecting the gamut? My green gain is all the way down, at -30, twice as far as the red and blue. For the offsets, green and blue are in the -10 area, and red is -5. I can't see how I could raise the green gain much because if I raise the red gain, the high-mid IREs start getting pink. I have ordered another 12k because of a large, bright, sometimes flickering, hotspot just right of center screen. The new one should be here on Tuesday. I will report back if it has the same green problem.

Rob