camera profiling

[elliotjnewman]

I was reading up on the section on dpbestflow.org about creating custom camera profiles and have since started playing around with the DNG profile editor and also the xrite passport which I purchased recently.

Forgive my ignorance but I assumed, and perhaps misread, that all you need to do is profile the camera (and lens if you want to go there), only once. As long as you have evenly lit the chart, then for any custom lighting conditions thereafter you would just adjust the white balance to gain colour profiled results. Of course Im sure there is far more since to it than that, but after some tests in different lighting environments I see that each profile that the adobe profile editor/ccp creates is different... so I guess there is far more to it than just profiling a "base" for your camera and then setting custom WB for each lighting condition thereafter, it seems now I have to profile a chart for each lighting condition I shoot in, which is going to result in a very long drop down list in ACR!

Can anybody confirm thats the case?

Regards
Elliot.

[peterkrogh]

Eliott,
PRofile and White balance are fundamentally different. The Camera Profile describes the particular variations in your camera, as measured against the generic profile tht Adobe has made for your camera model.  It's more-or-les a one-time operation that creates a file to compensate for your camera. (You might want to make new profiles a couple times a year and see if they are substantially different, or you might want to profile if you are using some extreme light source, like very blue or red or a strange bulb. You also mght wnat to make  a custom profile if you are doing a catalog shoot and need absolute maximum color fidelity).

The Profile should be assigned as part of the defaults. 

In general, there is no need to make new profiles very often.

The White balance, by contrast, must be set for each image, and is totally independent of profile, as a control. (Of course, the profile affects how color is interpreted, so they are not totally separate in effect.)

Does that clear it up?
Peter

[elliotjnewman]

Peter

Thats precisely what I thought would be the case, and also how I read it on dpbestflow. However, I profiled my camera in two different light environments - indoors under tungsten and outside cloudy, and the results are different, quite different in fact.

Perhaps I need to test it further, under more rigorous test conditions, at least now I know that the aim is to create a better camera "default" rather than profiling every new lighting condition (which is what I was afraid of having to do!).

Elliot.

[peterkrogh]

Eliott,
A couple of questions:

1. What kind of camera are you using?
2. How are you comparing? Keeping everything the same and then swapping profiles only?  Can you describe what you're seeing?
3. Is there a correlation in appropriateness of the profiles - so, does a profile made under tungsten light do a better job with images under tungsten?

Peter

[R. Neil Haugen]

Just to drop in my two pennies USD, I've read a couple reviews that suggested using this gear either of two ways:


1) Either a sunny or light-cloudy "general" profile that was then used for everything; or
2) Profiling in BOTH sun AND tungsten lighting, and loading both in with LR using this to adjust by sliding between the two as the color temp changes.

Apparently there is someway within the profiler for one to enter that one has a pair of such profiles to use. Also, I have seen commercial photogs with particularly high color needs using it for a specific shoot. Such as needing to get a specific set of Pantone colors in fabrics or car paints, for example, to reproduce EXACTLY correctly.

Neil

[danaltick]

I commented on a feature of the profile editor called "Dual Illuminates" when it first came out here http://thedambook.com/smf/index.php?topic=3471.msg19066#msg19066.  This is a pretty cool feature that will interpolate for varying lighting conditions in a single profile.  What I'm not sure of though is if custom profiles that we create or the one's provided by Adobe (e.g . the color matching profiles) take into account our specific camera like the Adobe 4.6 camera profile does.  What I think would really be cool is if we could somehow seamlessly add our specific custom camera calibration profile (that we might create using a MacBeth color checker chart) as a delta to each of the provided Adobe color matching profiles (e.g. Camera Faithful, Camera Landscape, etc.) so that they are calibrated to our specific camera.  This may be already built in.  I don't know, but would really like to know.  Peter, if you know or can find out, please let me/us know.  Thanks.

Dan

[peterkrogh]

Dan,
>This may be already built in.  I don't know, but would really like to know.  Peter, if you know or can find out, please let me/us know.  Thanks.

My understanding is that this cannot be done (combine a specific camera calibration with a "look"). It's my understanding that the camera calibration profiles (2.5 D profiles) are applied after the basic color profile matrix (these are the two illuminants that make up the basic camera "profile" has been applied.

I think that the camera Looks are an implementation of this - the basic profile/matrix plus a 2.5 D profile. I don't think the pipeline is designed to apply 2.5 D profiles twice.

(I have an incomplete understanding of the color decoding pipeline in ACR - and anywhere else, really - but I think this is how it works.  I'll be speaking to the ACR team about some other issues soon, and see if time permits to bring this up.)

Peter

[Richard Anderson]

Hi Elliot,

Couple of quick questions. Do you create a custom camera white balance before you shoot the  the color checker in each lighting condition? Are you using the second grey patch to tweak the white balance to neutral before using the DNG editor? Is the target evenly lit with only one light source? Are your exposures good as judged by the histogram (meaning no clipping on either side)? If all of these conditions are met- then you should get reasonably similar profile results. We have tested dual-illuminate profiles, and they are only slightly different from ones made with a daylight source. We have also tested the x-rite passport, and find the profiles to be slightly punchier/saturated than the Adobe DNG editor, but not different in color. To double check your profiles, measure the color patches of the color checker and compare them to the actual color checker values. The actual values are printed on the back of the chart, or you can use Bruce Lindblooms info  http://www.brucelindbloom.com/index.html?ColorCheckerRGB.html   If your profiles don't give you an accurate rendition of the color checker, then there is something wrong. The main concept of the DNG profile editor is to normalize the sensor response to all colors which allows you to use one profile for all lighting conditions with the caveat that you normalize the white balance by clicking on a neutral patch in ACR. This is a prime advantage to DNG profiles.

Peter is correct that the profile editor cannot be used in conjunction with the "Looks" profiles, as the "looks" are themselves tweaks to the standard matrix + 2.5 D LUT profiles.

Regards,

Richard Anderson

[danaltick]

I have an incomplete understanding of this as well, actually this is the first I've heard of the +2.5 D LUT profiles.  If I'm understanding correctly it sounds like the 2.5D profiles are camera specific and get applied to a camera independent basic profile.  If so, I don't think this would be considered a dual illuminate with interpolation, but rather just two profiles added together to make one.  As Richard says, if the Look profiles are tweaks added to the 2.5 profiles, then they are camera specific; correct?  If I create my own color checker custom profile (or dual illuminate pair) and select it, does that replace both the basic profile as well as the 2.5D?  If so, then it sounds like the Looks profiles would need to be separated from the 2.5D profiles so they could be added directly to my custom profile; correct?  I hope I'm not speaking jibberish here ;-).

Dan  

[peterkrogh]

Dan,
Well, kind of.
The 2.5 D profiles are an addition to the camera raw rendering pipeline (which has always used a Matrix of profiles - one tungsten-ish, and one daylight-ish).  These LUT profiles allow a basic addition to the profile that comes with ACR. There are some preset values, such as the camera looks, and you can make your own, such as the calibration for a camera.  You can't add two 2.5 D profiles to the same rendering.

Peter

[danaltick]

Peter,

I was thinking a custom calibration profile replaced ACR's built-in matrix of profiles, but sounds like it's just another LUT profile; and like you said you can't add them together.

Dan

[elliotjnewman]

I dont have much time so I'll have to be quick...

here are the details you guys asked for:
Peter:
1. Canon 1Ds Mark II
2. Yes everything is the same, just swapping camera profiles
3. I have not tested to that extent

Richard:
No, I just set the camera to raw and then set it to match the white balance of the environment manually.
No, I think I loaded the dng file pre any white balance settings in ACR
I usually use the white patch or the grey patch next to it, Im very conscious of any colour clipping or exposure clipping, and the light source is even so thats not a factor.

I must amend a certain detail here - the light source indoors was fluorescent - not tungsten. and I used a gretag macbeth chart (A4 size) for the indoor shot and the colour checker passport for the outdoor shot, I hope this has not introduced inaccuracies, I would hope not, but this is what I was referring to when I said more tests are needed.

My camera is being fixed at the moment so I wont be able to do anymore tests for a few weeks, but I am happy to upload/ftp the two dng files for you guys to take a look.

[Richard Anderson]

Hi Elliot-

I would definitely cut down on the variables. Use the same chart if you change the lighting for instance. Also note that fluorescent light can be tricky since it can be an incomplete color spectrum and it also flickers- which has implications wrt shutter speed. We usually use strobe light through a softbox for the daylight lighting, and just turn off the strobe light and use the modeling light for the tungsten light source. This keeps the light quality the same- just changes the color temp.

My understanding of the way the profiles work is that the custom DNG profile sits on top of (tweaks or edits) the underlying standard Adobe 2.5 D profile for the specific camera. So yes, a custom DNG profile is not only specific to the camera model, but is made even more specific to a particular camera body. The Adobe DNG "looks" are specific to a camera model. A custom DNG profile would replace one of these specific "look" profiles.

Richard Anderson

[danaltick]

If you really want to do it right, I would say you probably should gel the softbox (or umbrella) strobe to something in the neighborhood of 3200K and keep the f-stop roughly the same.  I doubt most modeling lights are as clean as a strobe.

With regard to the DNG profiles: It would be kind of nice though if you could stack your camera body specific profile on top of the Look profiles to make them specific to your own camera body; however, I not sure how relevant that would be considering that the Look profiles are creative effects designed to deviate from one's own calibration anyway; unless of course your camera is way off calibration from its 2.5D, and if that's the case, you should probably have it repaired.  Hope I'm not talking jibberish again ;-).

Dan

[Richard Anderson]

Hi Daniel-

We use Dynalites. Dynalite describes the modeling light this way:  "Quartz Modeling Light: 3200˚K color temperature and 2000 hour rated. 150 watt in RH1050 flash head, 250 watt in MH2050, SH2000, SH4080 flash heads, positioned to produce illumination identical to flash. User replaceable."

Regards,

Richard Anderson