Re: Camera Profiling
Re: Camera Profiling
- Subject: Re: Camera Profiling
- From: Simon ONeill <email@hidden>
- Date: Sat, 31 Jul 2010 12:36:41 +0100
Hi Lindsay,
I do not use Capture One, but suggest strongly that any custom profile (using ProfileMaker) should be produced using the same CameraRAW conversion that you want in your standard workflow, so using Canon DPP might (if you hit a lucky coincidence) create an accurate profile for a CaptureOne workflow, but should not be relied upon.
I was unaware that Profilemaker specifically demands a linear gamma - is this documented? I too have found that ProfileMaker is VERY particular about exposure and evenness of illumination, but having profiled Canon5D and Canon10D (using ACR as my preferred workflow using both Lightroom -Mac and Photoshop -Windows) I have found (after much trouble in getting even and correct illumination) that the profile is visually acceptable for (my) reprographic purposes. I confess that although I have measured colour temperature variation with output level in my Elinchrom Style600s I usually ignore it (partly because variation from 5000K for the levels I usually set is small), but perhaps this is a potential error source for me (and I should go away and check it more carefully!).
I use measured data for my ColorChecker SG (obtained with my Eye-One and PM5 Measure tool) However I would be very surprised if generic data would give such poor results as you describe.
One point perhaps, the ColorChecker 24 has a "light skin" and "dark skin" simulation patches - and these are embedded in the ColorCheckerSG set - do these also show as wild results when you check the test image using the troublesome profile(s)?
Wild colour shifts are common when a profile is applied twice (typically when Printer and Photoshop both try to colour manage), is it possible that something in your workflow is applying two profiles sequentially to your input data?
One word of caution with comparing the gamut shapes for input profiles: I think that the gamut shapes for an input profile give a rather extreme (and almost imaginary) view of the profile, showing a and b values which the profile predicts, but do not represent anything that the input device is likely to capture from a real world source. The gamut shape displayed for a printer is another thing entirely and represents something much closer to real measurements, whereas I suspect that the gamut shape for an input device represents an extrapolation of the data ( = not real measurements but estimates of something not ever present ... for example, representing R=0, G=0, B=255 or some other extreme case which the sensor may be incapable of ever recording due to the spectral width of the Bayer colour filters).
However, if you took your ColorCheckerSG image and used CTPro ColorSmarts to compare the two (or even three) input profiles for the test chart image, you might get more meaningful results than the gamut volume comparison.
It might be worth running this problem past the Chromix forum as I suspect that there is not a 100% overlap in readership with this one.
One further comment: As the distinctive element in this issue seems to be CaptureOne, perhaps the solution should best come from them?
Simon
On 29 Jul 2010, at 05:05, Lindsay Merritt wrote:
> Hi folks.
>
> OK, I know the general opinion on camera profiles and their worth but please, bear with me.
>
> I have a new Canon 7D and I use the latest version of Capture One V5 to process my raw files. I love the output of C1, have used it for many years and don't want to change.
>
> The problem is, the skin-tones I get from the 7D are just awful when processed in C1. The skin-tones could only be described as luminous orange. I've satisfied myself that this is a C1 profile issue for this camera. I did this by using a 24 patch ColorChecker card and by using both the DNG profile editor and ColorChecker Passport for ACR. The camera produces excellent results either with the standard profile or my custom one in ACR.
>
> We can shoot anything up to 15 portrait sessions a day in our studio and I need speed and reliability, hence Capture One. I don't like using ACR and LightRoom not only will nor launch on my brand new Mac Pro, I don't like the interface and dislike the external database. Trust me, Capture One and I are friends...
>
> My camera room is painted grey so as not to add color casts. My Elinchroms are gelled to match at 5250K as measured with my Minolta CT meter. We use a WhiBal reference in the first frame of every shoot. This is to allow as perfect a color rendition as possible with the least effort required in post production. This system has worked perfectly for years until now.
>
> I decided to build a custom profile for the 7D. I'd done them before when I photographed artworks with a Kodak DCS 760. The results were outstanding (and published). This process required a linear gamma tiff file. I can't remember the software I used to create the profile but I think it was made by Profile City (now defunct).
>
> I set the lighting to perfection and using a Kodak Grey Card, established the exposure with the Histogram, one line dead in the middle of the screen, double checked the exposure with Minolta Falsh Meter 1V and it matched to the tenth of an f-stop. Lighting was even (and proven so when later measured with the Eyedropper Tool in CS4).
>
> I photographed both the GMB ColorChecker DC card and the GMB ColorChecker SG card to see which would produce the best results.
>
> At this point, this is where I ran into trouble. I believed that the profile should have been created from a linear gamma, tiff file with no embedded profile and ProfileMaker was telling me that the image was too dark. It wanted a processed ProPhoto RGB tiff file. I installed Canon Digital Photo Professional in the hope that this would help but not only did I have to embed a proffile, the largest available was Wide Gamut RGB.
>
> When processed, obviously a tone curve gas been applied and the file looked normal, and what ProfileMaker was expecting but I knew that the result would not be satisfactory and was proven right. I checked the gamut volume against the Capture One standard CanonEOS7d-Generic.icm profile in ColorThink 3.0 Pro. I knew then I'd not succeeded but tried the profile out anyway. No deal.
>
> I have ColorEyes 20/20 and reinstalled that but if anyone has used that, it gives you the name of an animal and you require the match-phrase. I did that, received my match phrase and when I relaunched the application the following day, my bloody animal had changed and the match phrase no longer matched. Who knew?
>
> I've e-mailed for another match-phrase with the new animal but have not been answered. I'm sure they believe I'm attempting multiple installs...
>
> So finally, my question. What do I need to do here? I can't see that using an exported Tiff file complete with an embedded profile and tone curve applied can possibly allow me to build a profile worth a damn.
>
> Am I missing something? Is there software that will build an accurate profile from an untouched converted linear gamma TIFF file? What software (if not ProfileMaker) would Capture One be using to build *their* profiles?
>
> I can't test ColorEyes 20/20 (no match phrase) but the manual leads me to believe that I'm heading to the same dead end that ProfileMaker took me.
>
> I'd accept any advice, am happy to provide the raw files I took of the test charts etc.
>
> Anyone, please?
>
> Cheers,
>
> Lindsay Merritt M.Photog.
> About Faces Photography
> 318 Princes Highway
> Bulli. NSW.
> Australia 2516
> _______________________________________________
> Do not post admin requests to the list. They will be ignored.
> Colorsync-users mailing list (email@hidden)
> Help/Unsubscribe/Update your Subscription:
>
> This email sent to email@hidden
email@hidden
+44 7824 664505
+44 1344 421762
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden