Canon 5D Mk ii-Xrite Passport profile maxes blue saturation

[Charles Beasley]

I have been using two 5D "Classics" for many years with great satisfaction and a couple of years ago started playing with the Passport DNG profiles, and have created them from within Lightroom and from the stand alone ap. When applying the profile to export tif or jpeg of the CC file, always the dark blue patch (#13) of the CC will be 100% saturated.
This happens when rendering 16 bit tifs in either SRGB or Adobe RGB; going to ProPhoto brings the blue saturation down to 76%. 

I've done single illuminant studio (using PCB Einsteins,) dual illuminant (daylight, shade.) Covered my camera room walls and ceiling with black fabric. Almost perfectly evenly lit the passport. Everything I could think of. (I bought and used Coloreyes many years ago so am familiar with the vagaries of acquiring a good target; thank goodness the Passport is not so fussy!)

I figured the camera may have predated the software and been too old for the software to be accurate. I always thought that a newer model camera would be more accurate.

I finally got around to getting a 5D Mk II, and was a bit miffed to find out that it too leads to a profile that leads to 100% blue saturation.
I'm pretty sure the MK II was current and highly regarded at the time the Passport software was created.

I have always read that Canons tend to over saturate reds, so am a bit perplexed.

I have used the profile editor to bring it down close to the values from Bruce Lindbloom's reference files, but I have seen countless videos where profiles are created routinely and used right away without any checking to see whether any of the colors are oversaturated.

Has anyone else experienced this?
What would be the ramifications of using the profile "as is" (with the blue 100% Sat?)

EDIT: Just for grins, I converted the 16 bit ProPhoto export to Adobe RGB and the blue saturation  went from 76% up to 99%.
FWI, if I use the canned Adobe Standard LR profile and export to Adobe RGB, the blue sat is a more reasonable 73%.

TIA,
Charles

TIA

[Charles Beasley]

In light of another similar post, I am sending Passport images, one exported using the canned Adobe Std profile (blue sat @ 73%) and one using the X-Rite profile (blue sat 100%.)
In both cases, all settings in LR zeroed out, sharpening, exposure, etc.

PLEASE NOTE FOR SOME REASON THESE DID NOT UPLOAD WITH THE ADOBE RGB PROFILE INTACT. I DOULBE CHECKED AND THE VERSIONS ON MY COMPUTER ARE TAGGED ADOBE RGB. SO PLEASE ASSIGN ADOBE RGB BEFORE EVALUATING. SORRY-DON'T KNOW HOW OR WHY THIS HAPPENED.

[digitaldog]

In both cases, all settings in LR zeroed out, sharpening, exposure, etc.

Perhaps that's the problem. What about your settings?

[daicehawk]

In light of another similar post, I am sending Passport images, one exported using the canned Adobe Std profile (blue sat @ 73%) and one using the X-Rite profile (blue sat 100%.)
In both cases, all settings in LR zeroed out, sharpening, exposure, etc.

What light was used? Canon does not oversaturate reds, I would say, rather makes reds too cold\magentish.

[Charles Beasley]

Hi Andrew-thanks for the reply. Forgive my possible ignorance, but I would think that one would not want to make any adjustments to the raw file
before creating a profile from it. By zeroed out I mean that the raw file is straight out of the camera, WB  "as shot" and the default LR sharpening of Amt 25 changed to 0.
Before shooting the Passports, I CWB on a Macbeth gray card, then bracket the exposures, and pick the best exposure to use for profiling. 
With this file the lightest gray reads R=213 G=213 B=213, and the white patch is well under clipping at R=233 G=233 B=231, so I would think it would be a pretty good candidate for profiling.
Attached is a screenshot of the settings.

(I make sure it is well CWB'd and use "as shot" instead of doing a click balance to avoid the weird behaviour I discovered a couple of years ago:)
http://forum.luminous-landscape.com/index.php?topic=104883.msg862800#msg862800

It will do this whether I use the Passport or a standard full size CC. So I've used several different lighting scenarios, 3 camera bodies, and two different CCs and the one consistent thing is the 100% blue saturation.


daicehawk (thank you as well!): in this case one Einstein in a 4x6 softbox. It would do the same thing if I shot it in noon day sun. I've done this at least twenty times on the two old 5Ds and now 5 or 6 times on the "new" 5d MKII.

[daicehawk]

In light of another similar post, I am sending Passport images, one exported using the canned Adobe Std profile (blue sat @ 73%) and one using the X-Rite profile (blue sat 100%.)
In both cases, all settings in LR zeroed out, sharpening, exposure, etc.

Basically, where are your saturation level digits come from and why are you concerned provided you like what you see? To me the X-Rite blue looks closer to reality, although the skintone patch might be better with Adobe Standard.

[Tim Lookingbill]

Is there some reason why the 35L* gray patch (same luminance region of the blue patch) has the green channel reduced?

I entered the Lab numbers of your blue patch in Photoshop and it appears it is not a saturation increase but a hue shift towards a magenta-ish blue which reflects the non-neutral 35L* gray patch.

You seem to have a non-uniform neutrality issue with either the light being used or maybe the infrared/UV filters on your camera's sensor. My DSLR has weak UV/IR filters which of course makes reds and blues more pronounced.

You need to find out why the gray patches aren't as neutral across the entire luminance spectrum.

[Charles Beasley]

daicehawk-My saturation percentage comes from Photoshop Info Palette with the second readout set to HSB.
My eyedropper is set to 11x11 pixels.

I like the images I get when using the X-Rite profile much better than Adobe Standard. When I am working on one image at a time I have no problem accepting what I see on the monitor, but I have been wondering if it might bite me in the ass sometime when I sent  a big print to the lab, or if I am doing a batch of images and can't look at each one.
That's why I asked in the original post

"What would be the ramifications of using the profile "as is" (with the blue 100% Sat?)"

Plus I find it odd that it would be so far to the extreme; I could see it being a few point higher, but  100% as opposed to the low 70s?

[Charles Beasley]

Hi Tim-Thanks for joining in. PLEASE READ THIS FIRST:

EDIT:FOR SOME REASON THE TWO FILES I UPLOADED DID NOT UPLOAD WITH THE ADOBE RGB PROFILE INTACT. I DOULBE CHECKED AND THE VERSIONS ON MY COMPUTER ARE TAGGED ADOBE RGB. SO PLEASE ASSIGN ADOBE RGB BEFORE EVALUATING. SORRY-DON'T KNOW HOW OR WHY THIS HAPPENED. If you assumed srgb, then that probably threw all your hard work off. Again, so sorry. I usually work in srgb, but did these in Adobe so I wouldn't catch hell here.

>>>"I entered the Lab numbers of your blue patch in Photoshop and it appears it is not a saturation increase but a hue shift towards a magenta-ish blue which reflects the non-neutral 35L* gray patch."<<<

FWIW, I looked at the back of the CC package and it calls the blue patch "vivid  purplish blue!"

So we are on the same page,
1) is "35L* gray patch" the darkest gray (next to black?)
2) which image are you looking at X-Rite profile or Adobe Standard, or both?

I'm looking at both full size 16 bit Adobe RGB tifs, in Photoshop and I see no more than 2 points difference in any of the neutrals.
the darkest gray patch (next to black) reads 86,85,86 on the one using the X-Rite profile and the one using Adobe Std reads 86,85,87.
Would it be better if I attach the full sized tifs (don't know if there is a limit on file size here?) rather than the 50% downsized jpegs I posted?
I have always been under the impression that perfectly neutral neutrals across the whole tonal range was an unobtainable ideal.
I took a look at the RGB summaries on Bruce Lindbloom just now and was surprised to see that for Adobe RGB the two darkest patches are indeed
dead neutral, but all the others vary by 1-2 points.

Are you looking at them in Lab, and if so, should I be instead of RGB?

Also, please tell me what I am looking at with your image StandardCCBluevsMagentaBlue.jpg -the left side which I assume is Standard CC Blue looks closer to my images than the right side which I suppose is Magenta Blue? (I had to drag and drop your SRGB jpeg onto mine (ADobe) and got warnings about converting the color space, so may be some change there.)

[BobShaw]

I have had a 5D Mk2 since 2009 and it still works fine but has been replaced by a 5Ds. The 5D2 is a great camera but colour accuracy is not its long suit.

These are not "professional" (though for most things they are) cameras like the 1D. They are made for the masses and are all over the shop. You can not say they are consistently everything.

For any critical colour work I use a Hasselblad H3DII which is actually older and cost me less money, but absolutely nails colour.

[digitaldog]

Hi Andrew-thanks for the reply. Forgive my possible ignorance, but I would think that one would not want to make any adjustments to the raw file
before creating a profile from it. By zeroed out I mean that the raw file is straight out of the camera, WB  "as shot" and the default LR sharpening of Amt 25 changed to 0.

No not really, there's nothing special about the defaults or zero'ed out settings.

[daicehawk]

Hi Tim-Thanks for joining in. PLEASE READ THIS FIRST:

EDIT:FOR SOME REASON THE TWO FILES I UPLOADED DID NOT UPLOAD WITH THE ADOBE RGB PROFILE INTACT. I DOULBE CHECKED AND THE VERSIONS ON MY COMPUTER ARE TAGGED ADOBE RGB. SO PLEASE ASSIGN ADOBE RGB BEFORE EVALUATING. SORRY-DON'T KNOW HOW OR WHY THIS HAPPENED. If you assumed srgb, then that probably threw all your hard work off. Again, so sorry. I usually work in srgb, but did these in Adobe so I wouldn't catch hell here.

>>>"I entered the Lab numbers of your blue patch in Photoshop and it appears it is not a saturation increase but a hue shift towards a magenta-ish blue which reflects the non-neutral 35L* gray patch."<<<

FWIW, I looked at the back of the CC package and it calls the blue patch "vivid  purplish blue!"

So we are on the same page,
1) is "35L* gray patch" the darkest gray (next to black?)
2) which image are you looking at X-Rite profile or Adobe Standard, or both?

I'm looking at both full size 16 bit Adobe RGB tifs, in Photoshop and I see no more than 2 points difference in any of the neutrals.
the darkest gray patch (next to black) reads 86,85,86 on the one using the X-Rite profile and the one using Adobe Std reads 86,85,87.
Would it be better if I attach the full sized tifs (don't know if there is a limit on file size here?) rather than the 50% downsized jpegs I posted?
I have always been under the impression that perfectly neutral neutrals across the whole tonal range was an unobtainable ideal.
I took a look at the RGB summaries on Bruce Lindbloom just now and was surprised to see that for Adobe RGB the two darkest patches are indeed
dead neutral, but all the others vary by 1-2 points.

Are you looking at them in Lab, and if so, should I be instead of RGB?

Also, please tell me what I am looking at with your image StandardCCBluevsMagentaBlue.jpg -the left side which I assume is Standard CC Blue looks closer to my images than the right side which I suppose is Magenta Blue? (I had to drag and drop your SRGB jpeg onto mine (ADobe) and got warnings about converting the color space, so may be some change there.)

The neutral axis looks ok in both shots, and I do not pay much attention to what numbers are, provided you have your monitor calibrated and profiled. I checked them both with Adobe RGB assigned, there is really a minutia difference, so I would not bother. The thing with the saturation figures is just this - the blue primary in the X-Rite profile is just more saturated, that is why PS quotes the saturation figures that high. No problem unless there is some deep blue folded fabric in your shots you are afraid to become clipped and posterized on your monitor. I am of opinion that if if the hues and hue separation look right to you on your monitor (save for cyans on a generic sRGB monitor), then it IS RIGHT.   

[Tim Lookingbill]

I was going by the single LR screengrab that I didn't see was the AdobeStandard version. For some reason that file opened with the sRGB embedded profile and displayed correctly with the blue patch Lab numbers within range. The L*35 gray next to black does measure and appears reddish/warm which does happen with my camera using AdobeStandard on most of my images.

I use AdobeStandard mostly on mixed color temp scenes and wide dynamic range scenes such as sunsets mixed with cool shade. It also reduces clipping near full saturation so sunset lit clouds don't blow out looking like flat blobs of color.

I downloaded and opened the two full size versions linked previously to the screengrab version and had to assign AdobeRGB and see the issue with the custom profile blue patch which appears to suffer from contrast induced saturation seeing the luminance has dropped from the standard L*28 to L*22. That's what I get with my Pentax K100D X-rite CC custom profile. The blue and red are always pronounced which I was told in the past may be due to the UV/IR filtering on the sensor.

Who knows what, who, where and how these sensors are made and which camera model gets which sensor. It may appear you have a Sony sensor due to it behaving very similarly to my Pentax K100D.

[Charles Beasley]

Thanks everyone who has replied so far-it is truly appreciated. My takeaway from this is that I will do as daicehawk said
>>>"I am of opinion that if if the hues and hue separation look right to you on your monitor (save for cyans on a generic sRGB monitor), then it IS RIGHT.<<<"

I was just afraid that I might run into unforseen problems somewhere down the line on a big print from the lab-all my stuff is from a lab-no personal inkjet stuff.

I had in the past used the DNG Profile Editor to get the numbers down pretty close to Lindbloom's Color Checker Calculator, and was rewarded with images that look pretty much like ones using the Adobe Standard that I was trying to get away from!!

I will keep checking here in case anyone else wants to chime in.

[Charles Beasley]

Andrew-I am glad to see you here, seems like the last couple of times I looked for you you weren't here! I have always been appreciative of you unselfish help going all the way back to the Rob Galbraith forum days.

If you have a moment, please enlighten me on this:

>>>"No not really, there's nothing special about the defaults or zero'ed out settings. "<<<

I went back and rewatched your video http://www.digitaldog.net/files/DNG%20Camera%20profile%20video.mov
where you say DNG profiles (scene referred) use "rawer" data compared to an ICC profile, but to be clear, does this mean it ignores anything and everything, including white balance, that one changes in the develop module and just uses the raw data, as opposed to an ICC profile (output referred) that uses a rendered image?

The reason I am still not clear is that you said "my settings" might be the problem.

[digitaldog]

I went back and rewatched your video http://www.digitaldog.net/files/DNG%20Camera%20profile%20video.mov
where you say DNG profiles (scene referred) use "rawer" data compared to an ICC profile, but to be clear, does this mean it ignores anything and everything, including white balance, that one changes in the develop module and just uses the raw data, as opposed to an ICC profile (output referred) that uses a rendered image?

The reason I am still not clear is that you said "my settings" might be the problem.

ICC profiles are applied on rendered data. DNG (.dcp) on raw data. In the raw processing path of the raw converter. You can't do that with ICC profiles. They expect RGB (or CMYK) rendered data, not raw data. DNG camera profiles are white balance agnostic by design. WB isn't baked into the profiles (they are with ICC)!