Make Camera ICC profiles, Mac and Windows versions

Started Jul 7, 2013 | Discussions
Iliah Borg Forum Pro • Posts: 24,106
Make Camera ICC profiles, Mac and Windows versions
5

Current versions (v.25 atm)
http://s3.amazonaws.com/IliahBorg/MakeInputICC/MakeInputICC_Win.zip
http://s3.amazonaws.com/IliahBorg/MakeInputICC/MakeInputICC_Mac.zip

"Make Input ICC" is a small wrapper around Argyll CMS http://www.argyllcms.com/ for creating input (camera) profiles and performing basic checks on those. It is intended for profiling over raw data using RawDigger to obtain information necessary to calculate profiles.

No installation is needed, just unzip it to a folder of your choice and run the application.

Workflow:
set the reference CIED file
set the device data DevD file
set the quality
set the algorithm*
set the white point mode **
press "Go"
check Neutrality report ***
check deltaE2000 report ****

CIED file (reference data) contains spectral, XYZ, or Lab data measured from the target you are shooting using a spectrophotometer. Some batch-measured CIED files (including spectral data for various colorcheckers) are in the ReferenceData folder. Choose ColorCheckerSG_spectral.cie if you will be running sample DevD files from the Work folder in the distribution.

DevD file (device data) contains RGB values obtained from the shot of the target. Two DevD files for ColorChecker SG card are in the Work folder. The tool I used to make DevD files is RawDigger http://www.rawdigger.com - it takes a raw file as input and allows to obtain DevD file through Selection Grid and/or manual patch sampling, see http://www.rawdigger.com/usermanual/selection-grid and http://www.dpreview.com/forums/post/51015916 on the use of flat field feature). You can also use Danny Pascale's PatchTool http://www.babelcolor.com/index.htm#PatchTool for the purpose of extracting Dev data from a tiff file.

You can use buttons to open CIED and DevD files, or drag them into respective fields.

If you will be running the sample DevD file with gamma 1, NIKON_D4_Z100_4prof_NEF_g1_max227.txt, you can start with setting algorithm to matrix only. For gamma 2.2 file, NIKON_D4_Z100_4prof_NEF_g22_max242.txt, a good start is gamma + matrix algorithm. You can try using auto scale white point and compare results.

Argyll utilities are in Argyll_bin_Mac or Argyll_bin_Mac depending on the platform. They are current, and you can replace them with a different version if needed. No installation is necessary.

The resulting profile, ti3 file, neutrality and deltaE2000 reports (and optionally, log) are going to Work folder by default. Generally files go to the folder where the DevD file resides. Reports are saved in csv format and you can open them with some spreadsheet software for analysis.

Profile neutrality report is displayed as a plot.

delatE2000 report is displayed as a table, you can sort it by pressing the column name. Obviously, the sorting by error (deltaE2000 column) is the most often case.

Resulting profile has ti3 data embedded, with some fields added to ti3 to know what parameters were used while profiling.

In my profiling experience it is convenient to put camera serial number and lens type into Description field.

Any complaints and suggestions are most welcome. If the wrapper emits an error please save the log and e-mail it to me.

(*) Contrary to common practice I suggest you start not from the default cLUT profile algorithm but with some matrix algorithm. Neutrality report will tell you one of the reasons why.

(**) Try all algorithms here and see the difference in your raw converter.

(***) The RGB curves on Neutrality report should closely follow each other and they need to be pretty smooth.

(****) If your max. deltaE2000 is below 6 you are in excellent shape.

OP Iliah Borg Forum Pro • Posts: 24,106
version 1.40
1
OP Iliah Borg Forum Pro • Posts: 24,106
v.43 is out
1

- added processing of 4-channel RawDigger output

- allow regional settings for decimal separator

Please let me know of any bugs, feature requests are welcome too.

Same links,

http://s3.amazonaws.com/IliahBorg/MakeInputICC/MakeInputICC_Win.zip

http://s3.amazonaws.com/IliahBorg/MakeInputICC/MakeInputICC_Mac.zip

OP Iliah Borg Forum Pro • Posts: 24,106
v.44 is out
smartphoto Regular Member • Posts: 172
Re: Make Camera ICC profiles, Mac and Windows versions

Hi, I'm trying to make 5dmkII camera profile for CaptureOne, but the MakeInputICC says unknown format or corrupted file. I saved CGATS txt file from RawDigger,but can't understand what's wrong with resulting txt file. Should I check some boxes in Save samples to file window?

I'm not good at it at all, though thought I should try.

-- hide signature --

The world is nice without me

 smartphoto's gear list:smartphoto's gear list
Canon EOS 5DS R Canon TS-E 90mm f/2.8 Canon EF 70-200mm f/2.8L IS II USM Zeiss Otus 1.4/55 Zeiss Milvus 21mm F2.8
OP Iliah Borg Forum Pro • Posts: 24,106
Re: Make Camera ICC profiles, Mac and Windows versions

Could you please have a look here http://www.rawdigger.com/howtouse/prepare-data-for-profiling

-- hide signature --
smartphoto Regular Member • Posts: 172
Re: Make Camera ICC profiles, Mac and Windows versions

Thank you a lot for your answer! I've now get another message saying  Number of patches doesn't match. I've shot a color checker passport photo, so selected it as a reference data.

-- hide signature --

The world is nice without me

 smartphoto's gear list:smartphoto's gear list
Canon EOS 5DS R Canon TS-E 90mm f/2.8 Canon EF 70-200mm f/2.8L IS II USM Zeiss Otus 1.4/55 Zeiss Milvus 21mm F2.8
OP Iliah Borg Forum Pro • Posts: 24,106
Re: Make Camera ICC profiles, Mac and Windows versions

smartphoto wrote:

Thank you a lot for your answer! I've now get another message saying Number of patches doesn't match. I've shot a color checker passport photo, so selected it as a reference data.

Was it shot in the portrait orientation? Could you please post a screenshot with the grid overlay?

-- hide signature --
smartphoto Regular Member • Posts: 172
Re: Make Camera ICC profiles, Mac and Windows versions

Somehow managed to get it started, managed to make a profile, which isn't so much different from C1 profile for 5dmkII. I've updated Rawdigger to the latest version, maybe that helped.

Is it normal, that a photo, which was taken in bright sunlight, seams so underexposed?

And once again - thank you for your immediate answers and your willingness to help me.

Edit: how much does it take to make a profile after choosing different algorithm type? And what algorithm would you recommend? Now I'm trying to make XYZ cLUT + matrix, pc is working for some 20 mins already (i7-2600, 16 GB RAM, Gtx970)

http://imgur.com/ErUK1Wz

-- hide signature --

The world is nice without me

 smartphoto's gear list:smartphoto's gear list
Canon EOS 5DS R Canon TS-E 90mm f/2.8 Canon EF 70-200mm f/2.8L IS II USM Zeiss Otus 1.4/55 Zeiss Milvus 21mm F2.8
OP Iliah Borg Forum Pro • Posts: 24,106
Re: Make Camera ICC profiles, Mac and Windows versions

Now I'm trying to make XYZ cLUT + matrix, pc is working for some 20 mins already (i7-2600, 16 GB RAM, Gtx970)

For C1, Lab LUT or matrix + single gamma. LUT profiles in high quality mode may take significant time. Matrix profiles are usually add less noise and colour is smoother.

http://imgur.com/ErUK1Wz

Right, you've adjusted the rotation of the grid to fit the position of the target. I would decrease the sampling region a little (ROI relative size in Grid Parameters dialogue) as you are coming dangerously close to the borders of the patches.

As to the underexposure, I shoot targets using bracketing. Cameras do tend to underexpose if one is looking at the raw data.

-- hide signature --
helltormentor Contributing Member • Posts: 758
Re: Make Camera ICC profiles, Mac and Windows versions

Iliah Borg wrote:

Now I'm trying to make XYZ cLUT + matrix, pc is working for some 20 mins already (i7-2600, 16 GB RAM, Gtx970)

For C1, Lab LUT or matrix + single gamma. LUT profiles in high quality mode may take significant time. Matrix profiles are usually add less noise and colour is smoother.

http://imgur.com/ErUK1Wz

Right, you've adjusted the rotation of the grid to fit the position of the target. I would decrease the sampling region a little (ROI relative size in Grid Parameters dialogue) as you are coming dangerously close to the borders of the patches.

As to the underexposure, I shoot targets using bracketing. Cameras do tend to underexpose if one is looking at the raw data.

-- hide signature --

So far, I have worked with DNG profiles and the results are very dependent on shooting conditions and lighting. I have no experience in ICC profiling. Could you please highlight advantages and disadvantageous of ICC profiling compared to that of DNG profiling? I might switch to C1 if ICC profiles produce more accurate results.

 helltormentor's gear list:helltormentor's gear list
Fujifilm X-E1 Nikon D610 Samyang 24mm f/1.4 ED AS UMC Carl Zeiss Apo Sonnar T* 2/135
OP Iliah Borg Forum Pro • Posts: 24,106
Re: Make Camera ICC profiles, Mac and Windows versions
1

So far, I have worked with DNG profiles and the results are very dependent on shooting conditions and lighting. I have no experience in ICC profiling. Could you please highlight advantages and disadvantageous of ICC profiling compared to that of DNG profiling? I might switch to C1 if ICC profiles produce more accurate results.

There is no simple answer to this. The first problem is - how do we evaluate accuracy. This seems stupid enough, we all have eyes to see - but what we see is not what we get. Monitor profiles, printer profiles, our preferences to saturated or bleached colours - all matters here.

To have an idea, as a first step, please download "Multi Target" and "Still Life" shots for your camera, for the base ISO, raw (those are available from Imaging Resource from pages like http://www.imaging-resource.com/PRODS/nikon-d4s/nikon-d4sA7.HTM - click on "Thumbnails" to get access to raw). Convert using Adobe converter and CaptureOne, and see which you like more.

Next, we can run numerical analysis over the converted TIFFs to see which one is in fact more accurate.

Next, you can make DNG and ICC profiles from the multi target and apply it to various raw samples, both shot in the same studio and your own. It is good to include ColorChecker Passport in your test shots to be able to run numerical analysis of colour and tone accuracy.

-- hide signature --
exdeejjjaaaa
exdeejjjaaaa Veteran Member • Posts: 8,263
Re: Make Camera ICC profiles, Mac and Windows versions
1

helltormentor wrote:

So far, I have worked with DNG profiles and the results are very dependent on shooting conditions and lighting. I have no experience in ICC profiling. Could you please highlight advantages and disadvantageous of ICC profiling compared to that of DNG profiling? I might switch to C1 if ICC profiles produce more accurate results.

may be visit topics @ LuLa :

http://forum.luminous-landscape.com/index.php?topic=100015.0

http://forum.luminous-landscape.com/index.php?topic=103561.0

so much good info shared, including by the OP

 exdeejjjaaaa's gear list:exdeejjjaaaa's gear list
Sony Alpha a7R II Sony FE 55mm F1.8 Phase One Capture One Pro +25 more
helltormentor Contributing Member • Posts: 758
Re: Make Camera ICC profiles, Mac and Windows versions

Iliah Borg wrote:

So far, I have worked with DNG profiles and the results are very dependent on shooting conditions and lighting. I have no experience in ICC profiling. Could you please highlight advantages and disadvantageous of ICC profiling compared to that of DNG profiling? I might switch to C1 if ICC profiles produce more accurate results.

There is no simple answer to this. The first problem is - how do we evaluate accuracy. This seems stupid enough, we all have eyes to see - but what we see is not what we get. Monitor profiles, printer profiles, our preferences to saturated or bleached colours - all matters here.

To have an idea, as a first step, please download "Multi Target" and "Still Life" shots for your camera, for the base ISO, raw (those are available from Imaging Resource from pages like http://www.imaging-resource.com/PRODS/nikon-d4s/nikon-d4sA7.HTM - click on "Thumbnails" to get access to raw). Convert using Adobe converter and CaptureOne, and see which you like more.

Next, we can run numerical analysis over the converted TIFFs to see which one is in fact more accurate.

Next, you can make DNG and ICC profiles from the multi target and apply it to various raw samples, both shot in the same studio and your own. It is good to include ColorChecker Passport in your test shots to be able to run numerical analysis of colour and tone accuracy.

-- hide signature --

By the way, is there any way to calibrate an existing color profile by having a spot color of the scene? Palette claims that its CUBE color digitizer can capture the true color of the surface that it is used on. It's main usage is for product photography, I guess but I thought it might be possible to adjust an existing color profile by having a spot color of the scene though I have no clue if it can be done in practice.

 helltormentor's gear list:helltormentor's gear list
Fujifilm X-E1 Nikon D610 Samyang 24mm f/1.4 ED AS UMC Carl Zeiss Apo Sonnar T* 2/135
OP Iliah Borg Forum Pro • Posts: 24,106
Re: Make Camera ICC profiles, Mac and Windows versions
2

By the way, is there any way to calibrate an existing color profile by having a spot color of the scene?

Yes, but user-oriented tools are missing, except ProfileEditir (X-Rite).

Palette claims that its CUBE color digitizer

Oh no, the error this device produces is way too large.

-- hide signature --
exdeejjjaaaa
exdeejjjaaaa Veteran Member • Posts: 8,263
Re: Make Camera ICC profiles, Mac and Windows versions
1

helltormentor wrote: Palette claims that its CUBE color digitizer can capture the true color of the surface that it is used on.

this was reviewed by Graeme Gill @ http://www.argyllcms.com/doc2/smcube/smcube1.html
you can buy some used (even very used) spectrophotometer for even less money and get spectral readings...

 exdeejjjaaaa's gear list:exdeejjjaaaa's gear list
Sony Alpha a7R II Sony FE 55mm F1.8 Phase One Capture One Pro +25 more
OP Iliah Borg Forum Pro • Posts: 24,106
Re: Make Camera ICC profiles, Mac and Windows versions
1

exdeejjjaaaa wrote:

helltormentor wrote: Palette claims that its CUBE color digitizer can capture the true color of the surface that it is used on.

this was reviewed by Graeme Gill @ http://www.argyllcms.com/doc2/smcube/smcube1.html
you can buy some used (even very used) spectrophotometer for even less money and get spectral readings...

If one wants new, I'm currently in love with Hamamatsu C12666MA http://www.hamamatsu.com/resources/pdf/ssd/c12666ma_kacc1216e.pdf for such things.

-- hide signature --
GeorgeHatzakis New Member • Posts: 2
Re: v.44 is out

Thank you for your effort

i am very happy with the very first results

Eos 5d mark iii

Capture one

One think don't now how to manage is the blacks (shadows) are more heavy than the phase profile

The tonality of the new profile with the curve" extra shadow" applied  is very close to the phaseone generic profile with standard curve

Can I use different settings for blacks in RawDigger to obtain better results in blacks?

I have the measurement spectral data from my shooting lights is there a way to include this information to the profile?

I try several targets ( cc24,SG,basiccolor Dcam) and I realize that the brightest white pach on target must be set for WB in Raw digger for best neutrality on MakeInput right?

matrix

single gamma matrix

gamma matrix

Wp scaling

this settings I thing work best for me

I also have good profiles (to my eyes-taste) from other profile creation packages but this combination (RawDigger+MakeInput )it rocks

congrats!!!

OP Iliah Borg Forum Pro • Posts: 24,106
Re: v.44 is out

Can I use different settings for blacks in RawDigger to obtain better results in blacks?

No! That will skew the white balance. But what you can do is to decrease flare in the shot. That helps immensely with profile accuracy and shadow contrast.

I have the measurement spectral data from my shooting lights is there a way to include this information to the profile?

Currently, it is not simple. You will need to calculate a new reference from the target reference and light spectrum.

I try several targets ( cc24,SG,basiccolor Dcam) and I realize that the brightest white pach on target must be set for WB in Raw digger for best neutrality on MakeInput right?

"Light grey" patch is usually more neutral and spectrally uniform. For white balance, I use G5 on SG target and B4 or C4 on CC24 target.

-- hide signature --
GeorgeHatzakis New Member • Posts: 2
Re: v.44 is out

Can you tell me how to numerically evaluate the accuracy of the profile?

Thank you in advance

Keyboard shortcuts:
FForum MMy threads