I'm not Eric but many of us can probably answer most of the points.
2. No need to do any adjustments in ACR--except the color profile and possibly the WB although you can do that in the eyedropper in DNGPE. The ProPhotoRGB won't affect anything because it is an ACR-output profile and the DNG Profile Editor is working with the RAW data.
4. If you followed the process in Tutorial 6 then you did it right: Select Incandescent DNG, select 2850K color table (not both) and click Create Color Table. Select 6500K-ish DNG, select 6500K color-table, and click Create Color Table.
This Base Profile list does two things: It defines the base-profile that the adjustment-recipe tweaks are applied to when saving your new profile, and it lets you view your tweaks in terms of various applicable profiles for whatever your currently selected DNG is.
The ColorChecker profile added to the Base Profile list after using Compute Color Tables is a duplicate of whatever profile is embedded in the DNG when you open it in the DNGPE. It keeps you from having to remember what your embedded DNG Profile is, but it also obscures when your two dual-illuminant DNGs have different base profiles. I'd rather the list just had an asterisk next to the DNG's embedded profile so you could see if what you were saving was different.
Because the two dual-illuminant DNGs can have different base profiles, that's probably one thing to set in ACR before you get into the DNGPE, just to make them the same and one of the Adobe profiles, rather than something custom or from elsewhere, although that's certainly possible since the profile is actually embedded in the DNG not just the name of the profile like is in an XMP sidecar.
Thank you, and I used the same base profile for both dngs. The new information was that "ColorChecker" was a placeholder for the profile embedded in the dng. Where/how did you find/derive this bit of information?
Observation and experimentation. Check what profile is in the DNG before opening it. Once you build the color tables, flip back and forth between ColorChecker and the various other Base Profile items, observing the color-checker image at the left. You'll see no change when flipping between the ColorChecker profile and the embedded profile, and this lack of change is invariant no matter what you set the HSL sliders to.
>> So what did I create - a tweaked version of the D7000 Adobe Standard profile, or something based upon "ColorChecker", whatever that is? <<
"ColorChecker" showing up as the base profile indicates that the Chart Wizard uses the Baseline Matrix only to build its LUT on the top – independent from the starting profile. Any related "look-table" e.g. of the Adobe Standard profile is stripped off before.
>> Added by edit - Found a three year old thread in Luminous Landscape, with a member and Eric exchanging posts about returning to the Matrix profile for a given camera (forced by deleting all profiles), so I know a little more, I think. <<
Another way to access the Baseline Matrix profile for your camera, is to run the Chart Wizard, then Clear All Adjustments and finally export the profile. The resulting Baseline Matrix profile is right the same as the one obtained with the technique described in the LuLa thread.
Ok, I understand the theory, just looking for means to validate. Deja vu set in from three years ago, and I remembered an Adobe command line program, dngvalidate.exe, contained in their DNG SDK. With it one can "dump" a dng compliant file into a text format file and examine values.
So I used dngvalidate.exe on the D7000 Adobe Standard profile, and on the dual illuminant profile I created using Adobe Standard as the base. All of their matrices were the same, and the tonal curves appeared reasonably identical, although the ProfileToneCurve tag in my DNGPE created profile had 192 values, and the Adobe Standard ProfileToneCurve tag contained 128 values, so some interpolation was called for. Furthermore the ProfileToneCurve tags appeared to be two interleaved curves, i.e. every other value belonged to either the straight line curve or the base curve.
And it appears you are correct that the LUT is stripped off in Chart and a new one applied, and the created profile LUT was much larger than the Adobe Standard LUT. Apparently the Adobe Standard profiles are constructed differently (simpler, fewer values) than the DNGPE outputs, at least true for the D7000.
A third profile to compare with would be after computing the HSL values from the chart and exporting a new profile with ColorChart selected, go up an select the D7000 Adobe Standard base profile and export a new profile with the CC HSL values as the adjustment recipe to see if the differences you have observed are because ColorChart is different or whether the differences come from using the DNGPE to create a profile.
Ok, I performed the experiment, and it created a "different" profile with different visual results. However, I believe we may be into "garbage in, garbage out" land, in that changing the base profile from "ColorChecker" back to "Adobe Standard" changes the structure of the profile, as determined by using dngvalidate.exe, goes from 14 tags to 16.To recap, I went through the two illuminant Chart method as per before, except at the end just before exporting the profile I changed the base profile from "ColorChecker" to "AdobeStandard". The matrices remain the same, but the tonal values chart changes, and in addition there are now three (or maybe four) different hue/sat/value tables, two of the same structure but different values as the profile generated without changing "ColorChecker" and a third/fourth whichappears to be a copy of the table in the Adobe Standard. The profile works within ACR, but it gives a different visual result.I'm not sure this beta level software has much in the way of error checking and disallowing illegal combinations, so without knowing what's going on I would recommend one not change the base profile selection away from "ColorChecker" when using the Chart function.Here is a snippet of the text file created by using dngvalidate.exe on a DNGPE generated profile:-------------------------------------------------------------------Validating "dual.dcp"...Uses little-endian byte orderMagic number = 17234IFD 0: Offset = 8, Entries = 14UniqueCameraModel: "Nikon D7000"ColorMatrix1:0.8642 -0.3058 -0.0243-0.3999 1.1033 0.3422-0.0453 0.1099 0.7814ColorMatrix2:0.8198 -0.2239 -0.0724-0.4871 1.2389 0.2798-0.1043 0.2050 0.7181CalibrationIlluminant1: Standard light ACalibrationIlluminant2: D65ProfileCalibrationSignature: "com.adobe"ProfileName: "Nikon D7000 DNGPE 18-105VR Dual"ProfileHueSatMapDims: Hues = 90, Sats = 25, Vals = 1ProfileHueSatMapData1:h [ 0] s [ 0]: h= 0.0000 s=1.0000 v=1.0000h [ 0] s [ 1]: h= 0.0469 s=1.0006 v=1.0000h [ 0] s [ 2]: h= 0.2164 s=1.0026 v=1.0000h [ 0] s [ 3]: h= 0.5191 s=1.0063 v=1.0000h [ 0] s [ 4]: h= 0.9261 s=1.0112 v=1.0000h [ 0] s [ 5]: h= 1.3958 s=1.0170 v=1.0000h [ 0] s [ 6]: h= 1.8945 s=1.0232 v=1.0000h [ 0] s [ 7]: h= 2.3942 s=1.0297 v=1.0000h [ 0] s [ 8]: h= 2.8572 s=1.0357 v=1.0000h [ 0] s [ 9]: h= 3.2178 s=1.0403 v=1.0000h [ 0] s : h= 3.3788 s=1.0419 v=1.0000h [ 0] s : h= 3.2495 s=1.0388 v=1.0000h [ 0] s : h= 2.8264 s=1.0308 v=1.0000h [ 0] s : h= 2.2307 s=1.0195 v=1.0000h [ 0] s : h= 1.6373 s=1.0075 v=1.0000--------------------------------------------------------------------------This is a valuable tool for examining any dng based file, both profiles and images.Richard Southworth
The DNG Profile Editor is designed to allow the same adjustment recipe to be applied to different base profiles, to tweak the conversion for two different cameras to still be similar. This is what the first few tutorials are about.
Using a dual-illuminant color-checker profile adjustment recipe pair against another base profile does not produce useful results other than as an experiment to probe the operation of the DNG Profile Editor which has scant technical documentation.
I understand, but I believe the base profiles used for the Chart function have to be embedded in the DNG ColorChecker images, not chosen later using DNG Profile Editor. I believe all is "clear"; however, it would be nice if Adobe provided reasonable explantions for the total process, not just brief cookbook instructions on how to run the program.