[argyllcms] Re: Profiling between two images
- From: Pascal de Bruijn <pmjdebruijn@xxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Sat, 31 Jan 2009 14:09:48 +0100
On Sat, Jan 31, 2009 at 1:16 PM, Klaus Karcher <lists@xxxxxxxxxxxxxxxxx> wrote:
> Pascal de Bruijn wrote:
>>
>> I've more or less asked about this in the past, but I gave up on it,
>> back then...
>>
>> I have two images of an IT8 chart...
>>
>> 1. A digital camera generated JPG (white-balanced, and gamma corrected)
>> 2. A RAW file, converted to JPG (white-balanced, but not gamma
>> corrected, so it's linear RGB)
>>
>> Both images, are in fact, the same image, just stored in a different
>> way...
>>
>> I'd like to generate a profile, which would turn my nr 2 image into
>> something very close to my nr 1 image.
>>
>> Is this possible? Can anybody help me with accomplishing this?
>
> Ok, I think finally I got what you try to achieve: your "image 1" is no
> faithful reproduction, but you like the "look" you get when using workflow
> 1 and want to obtain the same biased result with workflow 2.
>
> Provided that the workflows are deterministic, you can create a profile for
> workflow 2 that largely replicates the "look" (or errors from a colorimetric
> point of view) of workflow 1.
>
> What you need is a profile built from a "faked" ti3 file. This file has to
> link the device values of image 2 with cie values calculated from image 1
> with profile 1.
>
> There are several ways to achieve this. I want to outline one of them using
> scanin and fakeread.
>
> Klaus
>
> # image1.tif and image2.tif are images of a ColorCheckerSG
> # processed with workflow 1 and workflow 2
>
> # read the device values from image1:
> scanin -dipn image1.tif ColorCheckerSG.cht ColorCheckerSG.cie diag.tif
> # image1.ti3 contains device values from image 1
> # and cie values from ColorCheckerSG.cie
> # the cie values in ColorCheckerSG.cie are trivial
> # as we replace them later anyway
>
> # create ti1 for fakeread
> cp image1.ti3 fake.ti1
> # change line 1 of file fake.ti1 from "CTI3" to "CTI1" and
> # line 9 from
> # COLOR_REP "XYZ_RGB"
> # to
> # COLOR_REP "RGB"
>
> # create ti2 for fakeread
> cp fake.ti1 fake.ti2
>
> # change line 1 of file fake.ti2 from
> # CTI1
> # to
> # CTI2
>
> # add column "SAMPLE_LOC" to file fake.ti2
> # (contents depends on your cht file)
> # example for ColorCheckerSG:
>
> # before:
> # ...
> # NUMBER_OF_FIELDS 10
> # BEGIN_DATA_FORMAT
> # SAMPLE_ID XYZ_X XYZ_Y XYZ_Z RGB_R RGB_G RGB_B STDEV_R STDEV_G STDEV_B
> # END_DATA_FORMAT
> #
> # NUMBER_OF_SETS 140
> # BEGIN_DATA
> # A01 89.130 92.990 75.770 98.873 99.556 99.984 0.23857 0.15282 0.19374
> # A02 0.80000 0.84000 0.70000 7.3287 6.6960 7.3602 1.3017 1.3735 1.2807
> # A03 18.560 19.270 15.790 44.461 48.695 55.216 0.41040 0.28607 0.27125
> # ...
>
> # after:
> # ...
> # KEYWORD "SAMPLE_LOC"
> # NUMBER_OF_FIELDS 11
> # BEGIN_DATA_FORMAT
> # SAMPLE_ID SAMPLE_LOC XYZ_X XYZ_Y XYZ_Z RGB_R RGB_G RGB_B STDEV_R STDEV_G
> STDEV_B
> # END_DATA_FORMAT
> #
> # NUMBER_OF_SETS 140
> # BEGIN_DATA
> # A01 A1 89.130 92.990 75.770 98.873 99.556 99.984 0.23857 0.15282 0.19374
> # A02 A2 0.80000 0.84000 0.70000 7.3287 6.6960 7.3602 1.3017 1.3735 1.2807
> # A03 A3 18.560 19.270 15.790 44.461 48.695 55.216 0.41040 0.28607 0.27125
> # ...
>
> # use fakeread to read the cie values for image 1 as "seen" by profile1
> fakeread profile1.icc fake
> # fake.ti3 now contains device values from image 1
> # and corresponding cie values calculated with profile1.icc
>
> # use scanin to replace device values in fake.ti3
> # with device values from image2
> scanin -r -dipn image2.tif ColorCheckerSG.cht fake diag.tif
>
> # modify file fake.ti3
> # change DEVICE_CLASS "DISPLAY"
> # to DEVICE_CLASS "INPUT"
> # and COLOR_REP "RGB_XYZ"
> # to COLOR_REP "XYZ_RGB"
>
> # create your fake profile
> colprof fake
>
> # when you assing fake.icc to image2.tif,
> # it shold look largely like image1 with profile1
This seems to be very close... However I don't have profile1.icc...
Workflow 1 -> JPG produced by my camera (of which I don't have a
proper ICC file)
Workflow 2 -> TIFF produced with UFRaw, as linear rgb...
The difference we are left with is gamma/shaper, and some color correction...
Of course, I'm assuming the transformation is static, and not
dependant on the image data...
Is there any way to make this work without profile1.icc?
Regards,
Pascal de Bruijn
Other related posts: