Canon 1DsMklll - Camera Profiles in ACR for maximum accuracy?

[Dinarius]

I've shot a Gretag Macbeth colour checker under even illuminated flash and then applying a few basic edits: White balance, Exposure adjustment, Blacks and Brightness.

Then, in search of the most neutral looking result I apply the Faithful and Neutral camera profiles.

To my eye, on my calibrated monitor, the Faithful is slightly better. This seems to be confirmed if I run the colour picker over the red, green, blue and yellow patches and look at the readings.

Visually, the yellow patch in Neutral is a bit lemon in tone. and the blue is a bit washed out.

I have tried Adobe's DNG Profile editor and I find it unsatisfactory. After profiling, I get a near zero red value in the blue patch.

I would be interested in hearing from others who need very accurate (objective) colour.

Many thanks.

D

[Chris_Brown]

I suggest you try the DNG profile editor but with using a color meter and the two different kelvin temp. light sources.

[yannb]

Hello,

This is the procedure I normally follow for my Nikon D300 (Lightroom workflow). It should be fine for your camera too I guess.

- In Lightroom/Camera Raw, make a preset with following values at zero: exposure, black level, contrast, brightness, tone curve = linear
- Shoot the Colorchecker quite big, but do not fill the viewfinder entirely. ISO lowest value. Good quality light. I start with a spot metered exposure on the fourth patch, bottom row.
- Import in Lightroom/Camera Raw with the above preset, neutralize the gray balance on the fourth patch, bottom row
- Aim for a white patch that has a value of around 95-96% (see http://lagemaat.blogspot.com/2008/06/srgb-tone-curve-and-lightroom-color.html )
- Convert to DNG, open Adobe DNG Profile Editor, hit the last tab 'Chart' and open your image.
- Place the colored circles in the correct patches, hit 'Create color table'.
- Save the DNG profile
- Restart Lightroom/Photoshop and apply the new DNG profile to the image. Now the colors should look good, but the image will be quite dull.
- Get the Photoshop scripts from Rags Gardner and install them: http://www.rags-int-inc.com/PhotoTechStuff/ColorCalibration/ACR-V8.zip
- Set Photoshop's RGB color space to ProPhoto RGB, absolute or relative colorimetric
- Open the image in Photoshop through Camera Raw (set sharpening and noise reduction to zero in Camera Raw, convert to RGB with ProPhoto RGB)
- Run the script ACR_Calibrate_CC24 to find out how you can set your brightness, contrast and black level for the most accurate reproduction of the Colorchecker

Generally, I can an average Delta E 2000 of 2 on the 24 colors of the Colorchecker.

I hope this helps you further.

Regards,
Yann

[digitaldog]

If you own a MacBeth 24 patch target, you can build a very nice DNG profile using the free Passport software from X-Rite (no need for scripts, no messing with the calibrate tab, a true DNG profile).

http://www.xritephoto.com/ph_product_overview.aspx?id=1257&catid=28&action=overview

Click on the Support “tab” in on this page, click the Passport download link and pick the OS you wish. Works as a stand-alone or within Lightroom.

[John S C]

I can second the use of the passport software with the ColorChecker. Works a treat. You can make this a pre set and apply on import

[JRSmit]

If you own a MacBeth 24 patch target, you can build a very nice DNG profile using the free Passport software from X-Rite (no need for scripts, no messing with the calibrate tab, a true DNG profile).

http://www.xritephoto.com/ph_product_overview.aspx?id=1257&catid=28&action=overview

Click on the Support “tab” in on this page, click the Passport download link and pick the OS you wish. Works as a stand-alone or within Lightroom.

True, but compared to profiles created  with dpe, de blueish AND red/purplish colors are more saturated, i prefer THE dpe profiles.

[inur]

Excellent thread, I bookmarked it. There are  questions I want answers too and this is very helpful.

[Peter_DL]

If you own a MacBeth 24 patch target, you can build a very nice DNG profile using the free Passport software from X-Rite (no need for scripts, no messing with the calibrate tab, a true DNG profile).

Tried it,
but found the results totally inferior compared to the DNG Profile Editor.

If I remember correctly, the X-Rite profile goes beyond adjusting matrix primaries,
as with ACR calibration controls, but it is far away from correcting selective Hue&Sat.-ranges as with the DNG Profile Editor.
Please no (further) discussion about color accuracy which is certainly not the final goal for a pleasing rendition.

To be honest, I already deleted everything (including earlier discussions) about the X-Rite color passport.
Eric's DNG Profile Editor is the much more elaborated tool, imo.

Peter

--

[Peter_DL]

- In Lightroom/Camera Raw, make a preset with following values at zero: exposure, black level, contrast, brightness, tone curve = linear
- Shoot the Colorchecker quite big, but do not fill the viewfinder entirely. ISO lowest value. Good quality light. I start with a spot metered exposure on the fourth patch, bottom row.
- Import in Lightroom/Camera Raw with the above preset, neutralize the gray balance on the fourth patch, bottom row
- Aim for a white patch that has a value of around 95-96% (see http://lagemaat.blogspot.com/2008/06/srgb-tone-curve-and-lightroom-color.html )
- Convert to DNG, open Adobe DNG Profile Editor, hit the last tab 'Chart' and open your image.
- Place the colored circles in the correct patches, hit 'Create color table'.
- Save the DNG profile

Not wrong, but far too complicate.

The Chart Wizard of the DNG Profile Editor ignores all the tonal adjustments in ACR.
Even when these adjustments are previewed in DPE, it just builds a profile on the linear state.
No pre-linearization required.

The Chart Wizard does its own white-balance on the second gray.

The Chart Wizard automatically grabs the same Baseline Matrix from the Adobe Standard profile or any Camera Matching profile which ships with ACR for you camera, thus using it as a starting point for H&S-selective tweaks.

Peter

--

[Tim Lookingbill]

I'm in full agreement with Peter's take on this and from my twiddling/experimenting creating and custom editing several profiles with the DNG Profile Editor.

I did every sort of mangling of presets to ACR's and DPE's curves and it was just as if I'ld applied the edits in ACR so I didn't see any significance in changing all the presets as a starting point. I even built a profile tweaking the highlights section of DPE's base tone curve that normalizes the sensor's linear response to provide better highlight definition. No increased advantage over doing it in ACR. In fact it introduced some color cross-overs in the highlights of some images shot outdoors and under tungsten. The smooth response established by the underpinnings of the default matrix and curves is a very clean editing environment to build on top of that doesn't produce artifacts and noise.

One caveat I mentioned to Dinarius over at the LL's Color Management forum is when editing the Color Table section of the DNG Profile Editor tweaking the CCchart patches is it's best to have other objects of similar color to the edited patch in the scene viewed at 100% to make sure what is done to the patches doesn't kick up noise in those objects. This has happened to me and I had to see this at 100% or more view.


I haven't tried X-rite's Passport software mainly because it's a 189MB download and I'm just not up for it.

[Dinarius]

Please no (further) discussion about color accuracy which is certainly not the final goal for a pleasing rendition.

To be honest, I already deleted everything (including earlier discussions) about the X-Rite color passport.
Eric's DNG Profile Editor is the much more elaborated tool, imo.

Peter
--

Firstly, I beg to differ on your first comment above.  I do a lot of work for museums and galleries, so I'm looking for (let's call it) objective colour, not subjective colour. A "pleasing rendition" is something I reserve for other work and personal work, not gallery/musuem work.

Secondly, the high esteem in which the DNG Profile Editor is held baffles me! 

Unlike the HSL tabs in ACR and LR3, you *cannot* get an updated RGB reading after making edits in the Colour Tables tab of the DNG Editor. Correct? Therefore, unless I'm missing something, you can only make *visual* adjustments to the file, not adjustments using the eyedropper tool. So, it becomes a case of "It's right when it looks right", rather than, "It's right coz I KNOW it's right from the RGB readout" Yes?

For this reason alone, the "long-hand" Fraser method done in ACR (and not in LR3, which a. Doesn't allow you to work in a colour space you'll be using after saving out and b. Doesn't have a Colour Sampler Tool, which is pretty much essential for the Fraser method.) has to be better, surely?

One final point......

If you download Gretag CC files from this link>

http://www.babelcolor.com/main_level/ColorChecker.htm#ColorChecker_images

try the following experiment:

1. Open the Adobe RGB 1998 16bit file in CS5. The green patch should measure 101, 148, 78.

2. Now open the same file in ACR, with settings (bottom of the editing screen) set to Adobe RGB 1998 16bit, and take a measurement of the green patch. On my ACR 6.3 it reads 116, 146, 83.

Go figure, as they say.

My tuppence worth.........ACR offers more control in terms of colour than anything else. (CS5 offers everything, of course, but I want to do as much as possible at the RAW stage) You're not locked into the ridiculous Melissa RGB (why Adobe?) in LR3, nor into the % readouts. In an ideal world, LR3 would have the same ability to select working space as ACR and it would have a Colour Sampler Tool. As to the Adobe 98 files reading differently in two programs.........??

Finally, if I am missing something in my approach to the DNG Profile Editor, please let me know.  I'd like to use it because I know if I could get updated RGB readouts, it would be much faster than Fraser/ACR.

Thanks for all the feedback. Glad I started the thread.

D.

[Tim Lookingbill]

I was actually able to manipulate through edits using the DNG Profile Editor on a CCchart to get exact Lab numbers relying on Apple's Digital ColorMeter. And I'ld have to say each color patch when looked at individually including the B&W patches looked split hair exact except the overall image looked odd and unnatural.

Those Lab numbers were derived using a scientific instrument (spectrophotometer) that reads spectral radiance from each patch. Technically the spectro nailed all the Lab numbers of each individual patch.

There's just one problem.

A spectro doesn't function the same as our eyes when we humans view the overall appearance of any given scene under any given light source. Our eyes adapt to all levels of tonality when viewing a scene. This includes perceived brightness, contrast, hue, saturation and color temperature. That's one big moving target.

So a spectro is a static instrument that doesn't adapt and doesn't read moving targets.

I can guarantee you if you closely look at the original as you edit the image, you will get better and more precise accuracy than relying on the numbers. Sometimes the profiles work and sometimes I just use the default ACR profile. Regardless, I am amazed at how close I can get to the original scene next to my display under any given light source using any color tool that manipulates the matrices and/or color tables smoothly and gradually without relying on the numbers.

[Peter_DL]

... so I'm looking for (let's call it) objective colour, not subjective colour.

So why not use the Chart Wizard ?

Anyway.

What do the DNG Profile Editor's RGB/HSL/Lab readouts mean colorimetrically ?

The RGB/HSL/Lab numeric readouts reported in PE are computed based on scene-referred ProPhoto RGB linear coordinates.

Peter

--

[Schewe]

Clean up on isle 3...