Color profiling while Exposing To The Right

[deejjjaaaa]

You obviously can check RGB values of the patches against given values. They are objective. What you clearly can't do is to check chromaticity values of colors in a real world scene.
What apparently anyone understood but you, is that these latter tests must be evaluated by eye, and any serious photographer is able to do it.

why shall I trust your eye and not some tools like babelcolor patchtool ?

It doesn't take a genius to figure out which image has more realistic colors and, when the ratio is 9:1 (if not even 10:0) towards the custom DNG corrected, it's not that hard to write a thread like that.

well, first of all "realistic" is subjective... however if the test under a regular (but different) light using a totally different type of a target and proper tools to measure instead of your "eye" shows that multitude of dE's is better for your profile than for Adobe's standard then I have a question to Eric Chan (or whoever does profiles in Adobe) - how come ? we are talking about a generic dual illuminant profile here...

I even lost time to explain that on that very thread an still you keep on this provocateur tone. If you really like to test it's been shot at 1,3", ƒ/5, ISO 125 for the indoor and 1/20", ƒ/5.6 for the outdoor.

how about just posting these 2 raw files... you share the profile - then you shall not have any issues to post the source, right ? unless.......

[deejjjaaaa]

with the Neutral 5 patch raw green channels

why Patch N5 - ideally we shall be checking just one patch - and the one w/ the most reflectance to do both things at once - check how close we are to the clipping and use RGB values of that only patch to comparing with the value from Bruce Lindbloom calculator, as suggested by Jeff Schewe for a source of values from an ideally exposed target for the purpose of DNG PE profile building...

[Schewe]

Curiously, the numbers specified on the file don't match the actual measured values of the image.

Not disagreeing with any of your comments on this though.

You need to click on the option to Scale RGB in the calculator then they match the synthetic PP RGB file (the calculator doesn't round to full numbers as the synthetic target does). also, when you are using the color sampler in Photoshop what Sample Size is set in Photoshop? When I open the synthetic target in ACR, all the numbers are within a level or two with all of the luminance patches exactly correct. Since we are talking about picking the optimal exposure, those are the ones that are important to compare.

As I recall, in ACR, the Color Sampler does a sample of 5x5 screen pixels (Eric can correct me if I'm wrong). But the readouts for rgb really are pretty close with generally only one of the tree colors off by a level or two.

[bjanes]

why Patch N5 - ideally we shall be checking just one patch - and the one w/ the most reflectance to do both things at once - check how close we are to the clipping and use RGB values of that only patch to comparing with the value from Bruce Lindbloom calculator, as suggested by Jeff Schewe for a source of values from an ideally exposed target for the purpose of DNG PE profile building...

I chose a midgray patch for evaluation with ACR/LR, since the midgray values are less affected by the contrast curve as shown below for PV2012 with ARR 7.1. The graph is self-explanatory.



If you are looking at the raw file with Rawdigger, you could use any of the neutral patches, since the sensor response is linear; in this case, the white patch would serve the purposes you suggest. The problem with Mr. Schewe's method is what ACR/LR settings do you use? The results are markedly different with different parameters as shown in my previous post. PV2010 with a linear tone curve and correction for the BaselineExposure appears to give the best results. The BaselineExposure can be obtained from the DNG exif.

The "ideally exposed target" values are usually determined mathematically from the spectral reflectances of the target convolved with the spectrum of the illuminant and not from an actual exposure (see Danny Pascale). Rendered values from an actual target exposed in a camera can never be fully accurate because of metameric failure.

Regards,

Bill

[Schewe]

The problem with Mr. Schewe's method is what ACR/LR settings do you use? The results are markedly different with different parameters as shown in my previous post.

You are making this all to complicated (as is your nature). The OP simply wanted to know whether or not to ETTR to create DNG Profiles...a lot of back and forth ensued...since my name was mentioned, I answered the question regarding how to pick an optimal exposure of a photographed ColorChecker chart...when asked how to determine the optimal exposure, I answered that it's optimal to pick a "good" exposure. Eric chimed in that as long as the image isn't underexposed nor any color clipped it was "ok". I still think that an optimal exposure is useful for making accurate and smooth DNG profiles. Do you disagree? Using a synthetic ProPhoto RGB and a photographed ColorChecker target and ACR to check on what would be an optimal exposure works...its what I do. Am I wrong? All I can say is what I do works for me...YMMV.

Don't under or over expose a ColorChecker and the odds are you'll be able to make a good DNG profile. The further to move away from a "good exposure" the less good the profile is likely to be...agreed?

[deejjjaaaa]

If you are looking at the raw file with Rawdigger

no, the whole point was to stay in the realm of Adobe tools to evaluate which shot is the best for DNG PE editor... certainly w/ rawdigger I can find out very easy which shot is the most ETTR'd (but then that does not mean that the very most ETTR'd shot is the best one for DNG PE or is it)

, you could use any of the neutral patches, since the sensor response is linear;

and one of rawdigger co-author says that linearity of the camera (not a single sensel responce on die taken w/o consideration of the whole sensor, its toppings and the rest of the signal processing circuitry) is not exactly a given fact within 1/3-1/2 to raw channel clipping, if I am not mistaken, at least not for all cameras (even modern)...

The problem with Mr. Schewe's method is what ACR/LR settings do you use? The results are markedly different with different parameters as shown in my previous post. PV2010 with a linear tone curve and correction for the BaselineExposure appears to give the best results. The BaselineExposure can be obtained from the DNG exif.

it does not really matter which method for as long as an average Joe can stay just within ACR and DNG PE - because that the market where XRite and other vendors are peddling the product... digging for a value of BaselineExposure tag is not nice... Ideally the next version of Adobe's DNG PE shall offer some indication about quality of the source raw file in more certain terms... one can certainly argue that an average Joe can live w/ whatever profile happens, but that's not nice.

[bjanes]

no, the whole point was to stay in the realm of Adobe tools to evaluate which shot is the best for DNG PE editor... certainly w/ rawdigger I can find out very easy which shot is the most ETTR'd (but then that does not mean that the very most ETTR'd shot is the best one for DNG PE or is it)

Personally, I see no need to restrict oneself to Adobe tools. ACR/LR are for visual editing and rendering of raw files. If one desires quantitative photometric and colorimetric analysis of the raw file, it is often best to look elsewhere. As my previous post demonstrated, the most ETTRed shot is sometimes not the best for the DNG profile editor, since an ETTR shot without clipping in any channel was rejected by the editor with an error message that the yellow patch was overexposed.

and one of rawdigger co-author says that linearity of the camera (not a single sensel responce on die taken w/o consideration of the whole sensor, its toppings and the rest of the signal processing circuitry) is not exactly a given fact within 1/3-1/2 to raw channel clipping, if I am not mistaken, at least not for all cameras (even modern)...

It is often not documented, but I understand that some cameras do some preprocessing to linearize the data from the sensor that is written to the raw file. One must know one's camera. My previous work with the Nikon D3 does demonstrate that the sensor is reasonably linear right up to clipping as shown below. To the extent that the system response is not linear with higher luminance values, it is best to avoid excessive ETTR and work with the linear portion of the response.

it does not really matter which method for as long as an average Joe can stay just within ACR and DNG PE - because that the market where XRite and other vendors are peddling the product... digging for a value of BaselineExposure tag is not nice... Ideally the next version of Adobe's DNG PE shall offer some indication about quality of the source raw file in more certain terms... one can certainly argue that an average Joe can live w/ whatever profile happens, but that's not nice.

If you want quantitative data concerning the contents of the raw file, the BaselineExposure must be taken into account if you are using ACR/LR. It is a pain to look up the baseline exposure, and it is often easier to use Rawdigger, which also avoids nonlinearities introduced by the tone curve. Personally, I do not like the introduction of the BaselineExposure. It does enable one to get consistent results when using multiple cameras that allow differing amounts of highlight headroom, but if one is primarily using one camera it is a pain.

It would be nice to have some indication of the quality of the source raw file, but, as Eric has indicated, a reasonable exposure without clipping is sufficient and I think he has better things to do than work on such an indicator.

Regards,

Bill

[bjanes]

Don't under or over expose a ColorChecker and the odds are you'll be able to make a good DNG profile. The further to move away from a "good exposure" the less good the profile is likely to be...agreed?

I do agree. From what Eric has indicated, a "good exposure" is sufficient and striving for the most ETTRed exposure without clipping needlessly complicates things. Indeed, in my previous experiments such an exposure was rejected by the DNG profile editor.

BTW, I just received my copy of your Digital Negative book and am impressed. Very well done with some great photography for illustration. I have all the previous editions of the RealWorld Camera Raw series and had found that the incremental value of each edition was diminishing since much of the material was repeated. A fresh start is most welcome.

Regards,

Bill

[madmanchan]

Do you agree that getting an optimal exposure is best for making DNG profiles?

Yes. 

Do you disagree that using a known reference chart (synthetic PPRGB) is a good way of evaluating various exposure of a bracket in ACR?

I don't disagree.  I think it's also helpful just to check if the resulting color mapping generated by the table does what you expect.  The main limitation of a synthetic PPRGB chart is that it's usually calculated with one illuminant only (if your source doesn't specify, it's likely D50).  For the dual-illuminant profile, that means the color patches resulting from DNG PE's Chart Wizard probably won't match your synthetic ref chart because the DNG PE uses A/D65 (instead of D50).

[madmanchan]

The DNG PE needs to be somewhat conservative in its chart wizard with regards to clipping because color tables are always applied after white balance.  Your raw capture may not have clipped data in its native coordinate system (native RGB, without WB applied), but that data can become clipped after WB is applied.  That leads to problems with the color mapping.  DNG PE will generally detect this case and prevent you from proceeding.

My recommendation is simply to bracket exposures and then pick the brightest one that DNG PE will accept without giving you an error.

[mac_paolo]

My recommendation is simply to bracket exposures and then pick the brightest one that DNG PE will accept without giving you an error.

That's exactly what I always do

[Schewe]

BTW, I just received my copy of your Digital Negative book and am impressed. Very well done with some great photography for illustration. I have all the previous editions of the RealWorld Camera Raw series and had found that the incremental value of each edition was diminishing since much of the material was repeated. A fresh start is most welcome.

Thanks for the kind words...

[Bryan Conner]

Thanks for the kind words...

What?!!!!???  You have a new book and did not tell me (us)?  Shame shame shame.  Just kidding...ok, I have to go buy my kindle edition over at Amazon.de now.  Looking forward to reading and learning....

[bjanes]

I use Bruce Lindbloom's ColorChecker Calculator to determine what the target chromaticity should be in ProPhoto RGB and simply read the patches in ACR set to ProPhoto RGB...pretty simple really. Just set the calculator to ProPhoto RGB...

If you are lazy, you can download a synthetic ProPhoto RGB target file from Chris Murphy's web site. It gives you the RGB target readouts of ProPhoto RGB. Save it as a TIFF and you can load the photographed target brackets and the synthetic target in ACR and check the readouts for an optimal exposure.

Curiously, the numbers specified on the file don't match the actual measured values of the image.

Not disagreeing with any of your comments on this though.

It is interesting to note that there are two common methods to obtain the color checker values for a given illuminant and color space (which itself has its own white point). The most accurate method is to compute the values directly using the spectral reflectance properties of the target convolved with the spectrum of the light source. This is the method used by Bruce Lindbloom's ColorChecker calculator, and no chromatic adaption is needed as explained here by the author. This is the method that Jeff is recommending. Another method would be to download the synthetic chart thus derived (say in L*a*b or ProphotoRGB) and then convert to the desired space in Photoshop, which would use chromatic adaption (the Bradford algorithm by default in Photoshop, I think). This does introduce errors as Bruce explains in the link. I would think that Chris Murphy's synthetic chart in ProphotoRGB also uses the exact calculated values.

If one photographs the chart in D50 illumination and uses a D50 space such as ProphotoRGB, then no chromatic adaptions are needed. Unfortunately there is no standard D50 source, but D50 fluorescent lamps are available. If one photographs the chart under D50 and renders into a D65 space such as sRGB, a chromatic adaption is needed and one would likely obtain results slightly different than if the chart were photographed under D65. Illuminant E (which is not black body and has no color temperature but a CCT of 5455 K) is interesting since the XYZ color matching functions are normalized such that their integrals over the visible spectrum are the same.  You talk about measured values in the image, but I don't think that any measurements are made--the values are calculated.

It is all a bit confusing and I don't claim to understand the details and think this would be a good subject for another thread or tutorial.

Regards,

Bill

[Rhossydd]

You talk about measured values in the image, but I don't think that any measurements are made--the values are calculated.

If you view the image (in it's correct colourspace) the measurements reported by PS's eye dropper are different to the values defined in numbers on the image itself.

Maybe it's something to do with JPG compression, but it's a curious issue that suggests you shouldn't get too pedantic over matching exact numbers.

[samueljohnchia]

The main limitation of a synthetic PPRGB chart is that it's usually calculated with one illuminant only (if your source doesn't specify, it's likely D50).  For the dual-illuminant profile, that means the color patches resulting from DNG PE's Chart Wizard probably won't match your synthetic ref chart because the DNG PE uses A/D65 (instead of D50).

Hi Eric,

Could you help to verify the illuminant(s) used under the various circumstances when generating the color tables in the DNG PE's Chart Wizard?

Both color tables - Illuminant A for 2850K and D65 for 6500K (as per your post)
6500K table only - D65?
2850K table only - Illuminant A?
6500K first and then 2850K after (or vice versa) - Illuminant A for 2850K and D65 for 6500K?

I noted that the profile that I generated for both color tables at once is slightly different from the one where I generated the 6500k table first and then the 2850K after because the RGB readouts after applying the profiles in Camera Raw (7.1) is different. I did not move the control end points for the target when making the profiles. Is this down to some averaging error?

[Schewe]

What?!!!!???  You have a new book and did not tell me (us)?  Shame shame shame.  Just kidding...ok, I have to go buy my kindle edition over at Amazon.de now.  Looking forward to reading and learning....

Yeah, ya know...I do have a life outside of LuLa...sorry to sneak up on ya!

Yes, I did a new book called The Digital Negative...kinda tried to sneak it in through the back door. Didn't want to make a big deal about it. If ya wanna know it's shipping (in print and ebook form) and has it's own website The Digital Negative Book.com.

Sorry I didn't make a big deal about it...but now ya know.

Buy one for yourself and a few more for your friends...don't worry, we'll make more!

[Peter_DL]

My recommendation is simply to bracket exposures
and then pick the brightest one that DNG PE will accept without giving you an error.

When we create a series of DNG profiles by running the chart wizard on such bracketed exposures (not too bright/clipped or too dark/noisy),
which sort and magnitude of deviations would we have to expect with the resulting Hue/Sat.-corrections per patch ?

Peter

--

[madmanchan]

Hi Eric,

Could you help to verify the illuminant(s) used under the various circumstances when generating the color tables in the DNG PE's Chart Wizard?

Both color tables - Illuminant A for 2850K and D65 for 6500K (as per your post)
6500K table only - D65?
2850K table only - Illuminant A?
6500K first and then 2850K after (or vice versa) - Illuminant A for 2850K and D65 for 6500K?

Yes, if you do the 6500k table only, it's D65.  If you do the 2850K table only, it's Std illuminant A.  Regardless of the order you choose to do them in individually, the mapping is always the same.

I noted that the profile that I generated for both color tables at once is slightly different from the one where I generated the 6500k table first and then the 2850K after because the RGB readouts after applying the profiles in Camera Raw (7.1) is different. I did not move the control end points for the target when making the profiles. Is this down to some averaging error?

This sounds unexpected to me.  How different are the readouts? 

[samueljohnchia]

This sounds unexpected to me.  How different are the readouts? 

Thank you for the clarification. The readouts differ from 1 - 3 values. The image of the colorchecker that I used to generate the camera dng profiles is the same as the one that I used to evaluate the differences between the profiles. The readouts are in ProphotoRGB and 16bits/color channel. The patches second row, first on the right (warm yellow) and third row, second from the right (magenta) differ the most - 3 values in the blue channel. A fair number of the other color patches also differ by 1 - 2 points in the R, G or B channel, sometimes in more than one channel. Only the neutral patches remain identical.

The image of the colorchecker was white balanced by using the WB tool in camera raw on the second lightest neutral patch. However, I am not sure that this is ideal as we know that none of the 'neutral' patches are spectrally neutral. Only the large gray patch found on the colorchecker passport is spectrally neutral.

This document provides evidence of that claim.

There are some other things I noticed when building camera profiles with the DNG PE. I will collate the information and post it in a new thread, to not go off topic here.