ACRCalibrator deeply wrong results

[Pindy]

Using AcrCalibrator v1.2.

I just bought a ColorChecker and went for an outdoor daylight photo for calibration. The results came back:

Shadow Tint: -30
Red Hue: -34
Red Sat: 100
Green Hue: -29
Green Sat: -100
Blue Hue: 14
Blue Sat: -100

Now either I have one screwed up D700 or something else is wrong. ACR Calibrator hasn't been updated to "support" ACR 5.3, but it seemed to otherwise work. Any ideas if this is a support issue or if I've done something else wrong? I had ACR set to ProPhoto RGB/16bits per channel as per Mr Schewe in "Camera To Print" and did the correct pen tool path.

Thanks, I'm new to this type of calibration.

[Schewe]

Tom Fors script is no longer being updated since Adobe offered the DNG Profile Editor. Try that instead...available on labs.adobe.com

[Pindy]

uh oh—I see what I did wrong. I did indeed get the pen path order wrong. I was sure I did it like MR and JS said to in the video, but it's different on the chromoholics web site. I'm likely wrong about my memory of how MR & JS said to do it as well.

[Pindy]

Tom Fors script is no longer being updated since Adobe offered the DNG Profile Editor. Try that instead...available on labs.adobe.com

Jeff—didn't see your reply until after I posted #2. Many thanks for the referral. Calibration #2 in any event was much more like it. Onto DNGPE.

[Pindy]

Okay, did the DNG Profile Editor (that was so much better than the old script). I suppose the question is: if you think the new profile is making everything too yellow, and my monitor is calibrated, could the fact that I photographed the 6500K chart in 5pm daylight make the difference? The DNG PE's online manual says to use a "D65 simulator" (whatever that is) and doesn't talk about how forgiving the PE is of light that was probably more like 5000K.

I understand the profile is meant to be a starting point, but I want accuracy first, before I tweak for taste.

[digitaldog]

I understand the profile is meant to be a starting point, but I want accuracy first, before I tweak for taste.

Actually no, you want to tweak for taste because there is no accuracy to define here unless your goal is scene referred capture:
http://www.color.org/ICC_white_paper_20_Di...ment_basics.pdf

[JeffKohn]

Okay, did the DNG Profile Editor (that was so much better than the old script). I suppose the question is: if you think the new profile is making everything too yellow, and my monitor is calibrated, could the fact that I photographed the 6500K chart in 5pm daylight make the difference? The DNG PE's online manual says to use a "D65 simulator" (whatever that is) and doesn't talk about how forgiving the PE is of light that was probably more like 5000K.

The light temperature doesn't have to be exact, although you don't want it to be too far off. 5pm daylight should be fine. The calibration image should be white-balanced off one of the gray squares in the bottom row; I'm not sure if the chart wizard will do this for you automatically or not - I always set the WB when saving the DNG, and then run the chart-wizard.

If the chart was properly white balanced I'd be surprised if everything is too yellow. I do find that the chart wizard results in a profile that renders green foliage (grass, leaves, etc) more yellowish than I prefer though. So I shift the hue on those adjustment points a bit more towards bluish-green for more pleasing landscape images.

[Pindy]

The light temperature doesn't have to be exact, although you don't want it to be too far off. 5pm daylight should be fine. The calibration image should be white-balanced off one of the gray squares in the bottom row; I'm not sure if the chart wizard will do this for you automatically or not - I always set the WB when saving the DNG, and then run the chart-wizard.

Adobe says it automatically uses the second-whitest square to WB the chart when using the wizard.

If the chart was properly white balanced I'd be surprised if everything is too yellow. I do find that the chart wizard results in a profile that renders green foliage (grass, leaves, etc) more yellowish than I prefer though. So I shift the hue on those adjustment points a bit more towards bluish-green for more pleasing landscape images.

I suppose that's what I don't get about calibration: if it's calibrated properly, why should colors appear unnatural? I'm cool with making adjustments, I just know of no other recording medium, digital or analogue, where doing a calibration is expected to return something other than the intended result. I will try to take another couple of captures today in "closer" light and attempt to get my shit together in all other ways.

[Pindy]

Actually no, you want to tweak for taste because there is no accuracy to define here unless your goal is scene referred capture:
http://www.color.org/ICC_white_paper_20_Di...ment_basics.pdf

Downloading and reading...

Thanks Andrew—you're doubtless the right guy to ask about this.

[Pindy]

This is one reason why
producing "accurate" color from a digital camera can be difficult. Accurate is in
quotes since nearly every user has a different definition of what they mean or
want when they say accurate.

Okay fair enough.

Let's see if I'm getting this: would it be fair to say it's impossible to view a purely "scene-referred" image (in its "correct" colorimetry) because of the necessity of viewing it on some output medium, even if that medium is a profiled monitor?

There's a missing link, surely, in my understanding. Moreover, what's the plain way of stating why a profiled camera and profiled monitor will make an image display a too-yellow image?

[digitaldog]

Okay fair enough.

Let's see if I'm getting this: would it be fair to say it's impossible to view a purely "scene-referred" image (in its "correct" colorimetry) because of the necessity of viewing it on some output medium, even if that medium is a profiled monitor?

There's a missing link, surely, in my understanding. Moreover, what's the plain way of stating why a profiled camera and profiled monitor will make an image display a too-yellow image?

Right. That is, accurate is only non debatable when you've got measured data from something like a Spectrophotometer. Everything else is up to debate and subjective. Which is correct, a scene shot on Velvia or the same scene shot on Ekatchrome? Neither really. Which do you prefer? That's the correct answer. To view all this on a display, its got to be output referred even without the subjectivity as to which rendering you prefer.

If the image appears too yellow, well you have to tweak. Or to put it this way, profiling a device like a digital camera doesn't ensure and guarantee correct, preferred color. Its a starting point.

[ayaboh]

If the image appears too yellow, well you have to tweak. Or to put it this way, profiling a device like a digital camera doesn't ensure and guarantee correct, preferred color. Its a starting point.

Are we talking about the camera white balance here, or wrong colors? Variations in white balance will happen more or less all the time and can be adjusted in ACR or other converters. Pictures shot in clear daylight with way off colors are different. You can't tweak these back to neutral with ARC camera adjustments or the DNG Profile Editor (you want accuracy first, and then taste). Profiling a digital camera with a profile made for that specific camera should guarantee (close to) correct colors. This has nothing to do with your display or printer.

The only thing that gives (close to) neutral colors from my EOS1 is Magne Nilsen's ICC profiles for C1. Shame they can't be used in ARC or LR.

[Pindy]

Are we talking about the camera white balance here, or wrong colors? Variations in white balance will happen more or less all the time and can be adjusted in ACR or other converters. Pictures shot in clear daylight with way off colors are different. You can't tweak these back to neutral with ARC camera adjustments or the DNG Profile Editor (you want accuracy first, and then taste). Profiling a digital camera with a profile made for that specific camera should guarantee (close to) correct colors. This has nothing to do with your display or printer.

The only thing that gives (close to) neutral colors from my EOS1 is Magne Nilsen's ICC profiles for C1. Shame they can't be used in ARC or LR.

No, we're not talking about white balance issues. My test shots included a whibal card as a control.

[Pindy]

Right. That is, accurate is only non debatable when you've got measured data from something like a Spectrophotometer. Everything else is up to debate and subjective. Which is correct, a scene shot on Velvia or the same scene shot on Ekatchrome? Neither really. Which do you prefer? That's the correct answer. To view all this on a display, its got to be output referred even without the subjectivity as to which rendering you prefer.

If the image appears too yellow, well you have to tweak. Or to put it this way, profiling a device like a digital camera doesn't ensure and guarantee correct, preferred color. Its a starting point.

I shall tweak.