How to ICC profile a camera? (no DNG)

[Nino Loss]

Till now I did not have the need for more than a visual correction of the generic Phase One profiles available in Capture One.

I need something more accurate for the next studio jobs (and at the same time get rid of the Canon 5D2's red).

The only solution I could come up with was Argyll, but I did not manage to find the time to learn all that. (I keep going back there and do not even now where to start...)

BTW I find it interesting that X-rites new i1PhotoPro package does only include the colorchecker to work with Adobe products. So what do I do when I want to use C1?

many thanks in advance

nino

[tho_mas]

right hand side -> "Input"
http://www.basiccolor.de/english/index_E.htm

I once made a profile for my camera with BC Input and was actually shocked how usable it was out of the box for the exact same lighting conditions.
I also was shocked how unsable it was for different lighting conditions (even just slight differences).
Camera profiling is not trivial ... I think you need a lot of experience to create a profile that (1) looks good and (2) behaves good.
I'd rather tweak the generic 5D profile C1 provides.
C1 V6 features multiple color read outs... which should help to adjust a capture towards a regular gretag chart.

[Nino Loss]

right hand side -> "Input"
http://www.basiccolor.de/english/index_E.htm

I was hoping for a cheaper/free solution. If somebody would tell me that he tried Argyll, than I would maybe start all over again with that.

I once made a profile for my camera with BC Input and was actually shocked how usable it was out of the box for the exact same lighting conditions.
I also was shocked how unsable it was for different lighting conditions (even just slight differences).

That's exactly what I need. A profile for one precise shot under controlled conditions.

Camera profiling is not trivial ... I think you need a lot of experience to create a profile that (1) looks good and (2) behaves good.
I'd rather tweak the generic 5D profile C1 provides.

I'm aware of the difficulties involved with camera profiling. I could have gone to the x-rite/adobe solution with the colorchecker, but the result is not up to this studio task.

Tweaking the available profiles is what I did till now. For most everything that's enough. But to get something overall correct is impossible through visual assessment.

C1 V6 features multiple color read outs... which should help to adjust a capture towards a regular gretag chart.

regards
nino

How would you do that? The read out will be meaning full if it matches the target. I will have to set the film curve to linear and furthermore apply a curve to bring the black and the white patch into the range of the target values. After this I  can start to try to get the colors into sinc via the read outs. I did that. The result was not nice, because while you tweak here you untweak there. I guess you know what I mean. (BTW the read-outs from my c1 v6.0.1 copy are meaningless - I filed a case)

I need to do something. The colors are simply wrong. Especially the Canon red.

EDIT: The mentioned Color Editor issue has been fixed. It was due to a corrupt ProPhotoRGB profile.

[tho_mas]

How would you do that? The read out will be meaning full if it matches the target. I will have to set the film curve to linear and furthermore apply a curve to bring the black and the white patch into the range of the target values. After this I  can start to try to get the colors into sinc via the read outs. I did that. The result was not nice, because while you tweak here you untweak there. I guess you know what I mean. (BTW the read-outs from my c1 v6.0.1 copy are meaningless - I filed a case)

in this case (tweaking the generic profile of the 5D2) you should set the film curve to "film standard". The linear curve is only useful if you start to create a new profile from scratch.
As to the "how to"...
1.) I'd shoot a color chart
2.) load (or create) a gretag chart with the correct Lab values, convert the file to your desired working space in Photoshop
3.) read out the RGB values from the info palette in photohop
4.) set C1 to the same working space
5.) tweak the colors in C1 so that they match the respective RGB values of the gretag chart in Photoshop
... but, as you've said, it's very difficult if they all have to match dead on ... so if this is your target a decent profiling software is the way to go (that's what it is for after all).
Don't know about the workflow in Argyll, sorry.

edit: "(BTW the read-outs from my c1 v6.0.1 copy are meaningless - I filed a case)"
you are right. The read out values refer to sRGB (for some reason).
Oh man ...

[Nino Loss]

Thanks for these hints, thomas!

[...]
2.) load (or create) a gretag chart with the correct Lab values, convert the file to your desired working space in Photoshop

I am not sure if I understand the "load (or create)" part correctly. Do you mean an actual reference image, like the artificial one by Bruce Lindbloom?

[tho_mas]

Do you mean an actual reference image, like the artificial one by Bruce Lindbloom?

yes!

[Nino Loss]

yes!

thanks again, thomas!

Maybe you also know about the available software options for icc profiling of cameras?

[tho_mas]

Maybe you also know about the available software options for icc profiling of cameras?

no, I have no overview here... I'm sorry!

[Nino Loss]

Ok, so I'll try your suggested work flow with the Lindbloom image.

many thanks

regards
nino

[tho_mas]

edit: "(BTW the read-outs from my c1 v6.0.1 copy are meaningless - I filed a case)"
you are right. The read out values refer to sRGB (for some reason).

not quite right.
Only if you are working in the camera profile (i.e. "embed camera profile" set as recipe... that's what I do) the read outs show sRGB values.
If you set any other color space as output recipe the read out values are correct.

[digitaldog]

I once made a profile for my camera with BC Input and was actually shocked how usable it was out of the box for the exact same lighting conditions.
I also was shocked how unsable it was for different lighting conditions (even just slight differences).
Camera profiling is not trivial ... I think you need a lot of experience to create a profile that (1) looks good and (2) behaves good.

Yes indeed! Its not trivial and often results in failure. My experiences mimic yours. In the days when I actually built ICC camera profiles, they worked pretty well with consistent (studio like) conditions.

[Nino Loss]

[...] In the days when I actually built ICC camera profiles, they worked pretty well with consistent (studio like) conditions.

Andrew, that's exactly why I thought that in this case, which is a controlled studio shot, an ICC profile would be the way to go. And without ICC profiles there is no profiling in Capture One, as you know.

[Nino Loss]

not quite right.
Only if you are working in the camera profile (i.e. "embed camera profile" set as recipe... that's what I do) the read outs show sRGB values.
If you set any other color space as output recipe the read out values are correct.

Unfortunately, here Win7x64 with C1Pro 6.0.1, the values are really meaningless in the color editor, and as soon as you change something it really gets fancy like the Blue Patch being R0 G255 B46. Also, the values above the viewer do not change when I modify something in the editor! I really hope they gonna fix that soon.

regards
nino

[tho_mas]

Unfortunately, here Win7x64 with C1Pro 6.0.1, the values are really meaningless in the color editor, and as soon as you change something it really gets fancy like the Blue Patch being R0 G255 B46. Also, the values above the viewer do not change when I modify something in the editor! I really hope they gonna fix that soon.

so the rebirth of this error...? -> http://www.luminous-landscape.com/forum/index.php?topic=44342.0
okay, that's different on Mac. There is just the read out error when the output is set to "embed camera profile" (however the RGB values in the viewer are correct).

[Nino Loss]

so the rebirth of this error...? -> http://www.luminous-landscape.com/forum/index.php?topic=44342.0
okay, that's different on Mac. There is just the read out error when the output is set to "embed camera profile" (however the RGB values in the viewer are correct).

Yes, there are also two or three other old issues that where buried and now live again...

[Nino Loss]

I tried to build a camera ICC profile. The result gave very very good colors, far better than the generic profile! But the profile makes the image too dark, the blacks are crushed!? What am I doing wrong? Could anyone point me to a source where I could learn how to do build a camera profile properly?

I did the following with basICColor Input:

1.) I shot the chart in the exact same conditions as the shot for which I need a profile.
2.) I processed a Tiff file in C1 with "film standard curve" and "Embed Camera Profile"
3.) Create the profile in basiICColor Input with "colorchecker24 spectral.txt" (instead of "colometric custom.txt", Keep camera gray" and "Add absolute white" unchecked, no spectral conversion and observer left at 2 degrees.

regards
nino

[Nino Loss]

I tried again with basICColor. Same result crushed shadows. Too dark. What am I doing wrong?

 - I used a CC24, evenly illuminated (variation of one to two unites across the whole board). White balanced with BabelColor. I tried various exposures. Nothing helps.

 - With C1 I tried the standard curve as well as the linear curve. Both with or without applying additional curves in C1 in order to get to the "ideal" values for the gray patches on the CC (as indicated here http://www.pictocolor.com/UserGuides/inCamera40/CaptureChartCC.html )

- I processed with ICC Profile in base characteristics set to "No color correction" and output profile "embeded camera profile"

[Nino Loss]

WOW! I had a few minutes to try Argyll. I feel like having a new camera. Gorgeous sumptuous colors! Accuracy is not yet the point, but the sheer depth and richness of tones is wonderful.

No blocked up shadows anymore. No posterization. I have unleashed the beast in the 5D Mrk2.

I have still a long way to. Colors appear more accurate but the Argyll's profile check gives me a DeltaE peak at 17.6 and an average at 7.8! That feels like a lot. I tried a few recipes but don't know what I am doing wrong. The CC24 is evenly illuminated. Exposure is spot on, as is WB. I process with all color corrections off.




P.S.: Argyll is quick and easy, once I have set it up!

[Nino Loss]

Anybody?

where should I turn to to find some help and answers?

[FranciscoDisilvestro]

I don´t have personal experience, but I found this site (french) from a reference in the DXO site (since DXO also can use ICC Camera profiles)
The guy there offers a 570 color patches test chart for more precise profiles and offers to create the profile (paid). Maybe the 24 color chart you´re using is not enough.

http://www.christophe-metairie-photographie.com/eng digital target.html