MacBeth Colorchecker Values

[Remo Nonaz]

I found this link, http://en.wikipedia.org/wiki/ColorChecker ,which was useful in understanding the colors and values in the MacBeth Colorchecker chart. However, what I am looking to find out is, what are the grayscale values of the five panels on the bottom of the chart in terms of percentages as shown in Ps curves panel. I can assume black is 100% and white is 0%. The third box looks neutral, but I'm not sure.

Is there a way to convert the RGB values, which are shown in the noted article, into grayscale percentages?

Any ideas?

[bjanes]

I found this link, http://en.wikipedia.org/wiki/ColorChecker ,which was useful in understanding the colors and values in the MacBeth Colorchecker chart. However, what I am looking to find out is, what are the grayscale values of the five panels on the bottom of the chart in terms of percentages as shown in Ps curves panel. I can assume black is 100% and white is 0%. The third box looks neutral, but I'm not sure.

Is there a way to convert the RGB values, which are shown in the noted article, into grayscale percentages?

Any ideas?

I am a bit confused by your request as your referenced article does not give RGB values for the ColorChecker but rather reports CIE xyY. Probably the best reference for color checker values are those by Danny Pascale here. The Photoshop Curves dialog does not use percentages but rather 8 bit pixel values. One can report normalized values as percentages and I think this is what Lightroom does. To normalize 8 bit pixel values, one divides by 255 to get values between 0.0 and 1.0

Regards,

Bill

[samueljohnchia]

I did a mode change in photoshop of a custom digital colorchecker. But you will also need to consider what gamma your grayscale is in.

For Gamma 1.8 I'm getting 6%, 26%, 44%, 60%, 74%, 86%
For Gamma 2.2 I'm getting 5%, 21%, 37%, 52%, 67%, 79%

If you go by dot gain percentages it will be different again. How on earth is this data useful to you, Remo? Maybe L* values might be a better perceptual match to the way we see, although some color authorities say that even the L* curve is a bit too dark.

The cyan in patch 18 is also beyond the gamut of sRGB. The table values of the wiki article is not in sRGB values, but in web colors. Wikipedia is rarely an accurate source of information.

[Remo Nonaz]

Well it's all a bit confusing to me.

Right now I'm trying to recover from a system re-build after some self-inflicted major OS issues. I re-loaded my Intel Graphics 3000 driver, which is stated to be up to date and the Pantone hueyPRO 1.5.1 software for my Huey. After calibrating my screens they looked like total crap - way too red. I re-loaded the software, ran some more calibrations and messed around with the base screen settings, eventually getting things to look pretty good. I got the notion into my head that somehow using a MacBeth chart, which I have a scanned copy of, might be useful for setting the brightness and contrast right. Actually, I found that a 21-step-scale worked better and the screens didn't need any adjustment. However, I'm still curious as to what the grayscale values of the bottom row of the MacBeth chart are.

But here is where it is really weird. When I view the scanned MacBeth chart in (I assume) a non color-managed application; like the desktop, it looks horrible. Viewed in Lr or CS4 it looks quite good (compared to a printed chart under good light). However, I downloaded a couple of the charts that are available from bablecolor.com and they look horrible even in Lr and CS4. They are all washed out, black is gray, all colors are muted.

I confess, I'm totally confused but I'm not trained in color management, so I'm sure I'm missing more than a few important points. Well, the objective is to be able to reliably make good quality prints without going mad or wasting reams of paper. I had that objective under control before the OS meltdown. Now I just need to repeat the process.

Thank you for your comments.

[sandymc]

In short, to reconcile GM24 value to what LR or ACR displays, you have to take both gamma and the tone curve that LR/ACR applies by default to raw images into account.

Back in 2008, I wrote a series of blog posts that go into all the detail, starting with this post: http://chromasoft.blogspot.com/2008/01/lightroom-aperture-and-capture-one-mini_24.html

Sandy

[samueljohnchia]

Remo, trying to scan in an color checker may not be the best way to have a digital rendering of the color checker. Is your scanner using a fluorescent light source? That alone can throw off the color rendering of the chart. Secondly, if your scanner is not profiled, you would have no idea what you are going to get. The tone response curve of the scanner would also throw off the greyscale steps. What color space is the scanned image finally in before you displayed it on your desktop, LR and PS? IIRC the desktop assumes the image is in sRGB, others can correct me if I'm wrong. In PS, it generally will honor the embedded color profile, if you have not adjusted Photoshop's default color settings. That would explain the difference you are seeing. If you are opening a baseline TIFF of the chart in ACR/LR, no adjustments are applied by default, and you will be looking at the colors converted into the current workflow settings' color space. So I would not worry about any internal gamma/tone curve.

In your first post you assumed that the colorchecker's white tile is 0% density and the black is 100%. But the white patch does not reflect 100% light (its not a mirror) and the black does not absorb 100% light (that would be a black hole!). Thecolor patches in the actual chart is not all that saturated also. The rather washed out rendition of the digital colorcheckers from the babelcolor website is probably much more accurate than your scanner's version, assuming you have your screen well calibrated. Bruce here has some reference images synthetically reproducing the colorchecker under different light sources:http://www.brucelindbloom.com/index.html?ColorCheckerRGB.html

All those colorchecker targets, including one that I custom made myself, appear pretty accurate on my display. Do note that not all displays can show the cyan patch 18; that is out of the color gamut of many standard displays.

All in all what you are doing, scanning in a colorchecker, viewing it on your display and printing the scanned version and comparing it to the original colorchecker may let you know if something is wrong with your color management workflow/pipeline, but does not tell you everything.

[Remo Nonaz]

I messed around with my screen settings and got the screen as close as I could to MacBeth chart. I then ran the Huey software again, which didn't make a big difference in the viewed image, but I still think the screen looks better in the unadjusted mode.

Last night I finally got a chance to do some printing. The images, adjusted using the Huey-corrected screen, print with a slight greenish cast. If I shut off the Huey correction and correct Lr to my unadjusted screen I get a print that is very close to my screen image.

I know that is not the way this is supposed to work. Several years ago I went through quite a trial getting the screen, Pantone Huey and the printer to all play nicely together. Eventually I prevailed and was getting excellent results with the Huey-corrected screen. Unfortunately, it was not a methodical process and nothing is documented. I'm just going to have to go through it again. 

[samueljohnchia]

Remo, I don't have a Huey and have never used one before. A quick google search shows many posts on other forums where people have issues with the Huey and display calibration & profiling. A greenish cast is often reported. I have an X-rite i1 Display Pro, and it works brilliantly. I and many others highly recommend it.