Ad
Ad
Ad
Pages: « 1 [2] 3 »   Bottom of Page
Print
Author Topic: ColorChecker Passport Profiles - vs White Balance  (Read 10154 times)
Schewe
Sr. Member
****
Offline Offline

Posts: 5474


WWW
« Reply #20 on: February 17, 2012, 11:27:06 PM »
ReplyReply

My conclusion is that Jeff is right.

Folks...again, I'm really happy that X-Rite did the passport system–on a lot of reasons. But I think their "marketing" over sold the usefulness in shot by shot profiles...I just don't believe in the hype. It's an unneeded complication that many have bought into.

Yes, do profiles on commonly used light sources. Yes do a standard dual illuminate profile for YOUR camera...but doing a shot by shot set of profiles will drive you nutz...don't fall into that trap. Really, you won't see the bennies...
Logged
Mike Guilbault
Sr. Member
****
Offline Offline

Posts: 817



WWW
« Reply #21 on: February 18, 2012, 07:02:03 AM »
ReplyReply

So, for example, if I create a profile for my studio lighting - and then take that lighting with me on location, the same profile should work just fine. 

How about for daylight, and daylight meaning all the variations of daylight from morning to night, overcast, shadow, etc.  Will the dual illuminate profile be sufficient for all those?

What conditions would make you create a new profile?
Logged

Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1154



WWW
« Reply #22 on: February 18, 2012, 10:58:41 AM »
ReplyReply

From my experience I have to agree with Jeff. I have maybe four custom DNG profiles I created for different conditions.

A dual illuminant, a single D65 illuminant for an old legacy lens, another dual illuminant just selecting "Both Color Tables" using one image of a sunlit CCchart in Adobe Profile Editor Wizard (puts extra cyan into greens, deeper blues), another D65 single illuminant for a 100 watt 5500K HD Ottlight CFL and a single illuminant D65 for my kit lens.

Switching between them on any number of images mostly gives subtle shifts to certain colors and some more pronounced depending on the subject.

Some folks I offer digital imaging advice asking how to get a certain look don't realize the complicated optical affects going on with WB and the image's color table that affect color constancy. They can't get a specific color to look vibrant as they see it in the actual scene without wrecking the rest of the image with R=G=B WB shooting under daylight balanced fluorescent lights without using any reference targets.

They don't know what is causing this even though the image looks OK. They adjust the saturation slider and it looks off. They don't know a camera profile will adjust the relationship of surrounding colors including neutrals so that one specific color will look as intended without having to punch up saturation.

The camera profile will not make big color corrections/adjustments. It only shifts HSL appearance relationships according to what that given illuminant does to the appearance of a certain scope of colors. The actual real illuminant (such as tungsten or flash) have additional hues (spectral spikes?) imbued by the manufacturer that a camera profile can't completely compensate for because D65 and A illuminants are not real physical lights. It's just code.

However, they do a surprising amount of compensating of the color table for certain memory colors like adding more cyan into greens shooting under red orange tungsten. Not too much, but often just enough to where an adjustment to HSL hue or luminance slider will get you the rest of the way. Users tend to focus on matching one color in an image to the actual scene and not how it looks in relation to the rest of the image which is what a profile can only fix. The HSL panel and other color sliders can fix the rest.

The image below shows all that a custom profile can fix even for shooting under oddball lights. The Pantone swatch #306 (underlined in black) goes from sky blue to turquoise, the actual color, using a profile but only to an extent depending on the light used. The lower right image is the most accurate. All of them are quite acceptable on their own, but once matching to scene is desired then more work is required a camera profile can't fix on its own.
« Last Edit: February 18, 2012, 11:00:57 AM by tlooknbill » Logged
opgr
Sr. Member
****
Offline Offline

Posts: 1125


WWW
« Reply #23 on: February 18, 2012, 12:02:44 PM »
ReplyReply

So, for example, if I create a profile for my studio lighting - and then take that lighting with me on location, the same profile should work just fine.  

Yes, except of course for mixed additional light sources with extreme spectral characteristics…
(modern LED lighting on Asian weddings anyone?)

How about for daylight, and daylight meaning all the variations of daylight from morning to night, overcast, shadow, etc.  Will the dual illuminate profile be sufficient for all those?

Yes, because you would generally want the normal daylight response from the sensor so you can reproduce some of the original colorcasts normally associated with such lighting conditions. The one thing to watch our for is this:
At sunset, the color temperature can get so low, the dual profile setup may start to use the tungsten response even though it is not appropriate for the true lightresponse of the sensor… (it's still a daylight spectrum, albeit very warm). If you truly experience errors in colorresponse, simply create a dual daylight DNG profile.


What conditions would make you create a new profile?

When confronted with extreme lighting spectra. An important example would be those modern non-led energy saving lightbulbs. I think there may be great amounts of infrared content which impacts sensor response, at least for older camera's. Which brings me to one other primarily theoretical situation where a lens may actually make a difference:

The lens coatings may filter infrared, which in turn may influence sensor response. But I have no idea whether this would ever have happened in any practical situation. Although: I do recall a slight Leica mishap and some proposed solutions…


EDIT: I also don't know whether modern LED lighting can be corrected consistently with profiles. Anyone have any experience with this?
« Last Edit: February 18, 2012, 12:05:15 PM by opgr » Logged

Regards,
Oscar Rysdyk
theimagingfactory
Schewe
Sr. Member
****
Offline Offline

Posts: 5474


WWW
« Reply #24 on: February 18, 2012, 12:25:33 PM »
ReplyReply

How about for daylight, and daylight meaning all the variations of daylight from morning to night, overcast, shadow, etc.  Will the dual illuminate profile be sufficient for all those?

Those are all white balance changes, not spectral illumination changes. All forms of "daylight" come from the same source, the sun. So, on one hand you have the sun, in all it's various white balances and on the other hand you have tungsten–which while all over the board regarding degrees Kelvin are again light created by heating a filament of metal. Yes they are all different in white balance but not all that different in terms of spectral output.

So, that's what a dual illuminate profile is designed to do–compensate for the sensor's response to substantially different spectral output. And yes...a good dual illuminate profile will be sufficient for all those...

The main times you would want to do additional custom DNG profiles is where the illumination is somehow different than sun or tungsten. LED lights for example would need their own DNG profile. Or HMI lights. Calumet sells fluorescent light panels for stills or video. I would definitely make custom DNG profiles for these lights...if you shoot tungsten with dichroic filters, I would suggest making a custom DNG profile because taking tungsten and filtering out the yellow won't really be perfect so it would fall into a hybrid lighting. Agree with the above post that mercury vapor lighting is prolly so deficient in color bands as to fall outside of what a DNG profile could correct...but I don't know that from experience...never made one under that sort of lighting.

As far as separate daylight vs strobe...with all the DNG profiles I've made I've used strobes to light the daylight target and Mole-Richardson tungsten lights to make the tungsten targets. These profiles work perfectly fine for shots from Antarctica to the America South West...I'm not saying you shouldn't do a single illuminate DNG profile of your studio strobes...but I think the expectation would be very, very minor differences.

There's another whole area of usefulness for DNG profiles and that's to help nail client's PMS trademark colors. So, say you were shooting John Deere tractors, a normal custom DNG profile may not render the John Deere "green" optimally....this would be a case where you could take the DNG profile, bring it into DNG Profile Editor and actually edit the way the profile would render greens. Yes, once you edit a color range it will impact other colors but John Deere prolly couldn't care less since they are primarily concerned with their "green" being right. The Passport software can't (at this time) actually edit DNG profiles, just make them.

The only other time I could see making a custom DNG profile based upon a shot would be in a mixed light situation where you need accurate color of something in a place in the shot that you could put a Passport. The mixed light at that place would need to be the same as what was falling on the object and it would prolly be a one off custom profile. The DNG profile won't be able to correct for color casts and white balance so much as correcting for the sensor's response in that mixed light situation.
Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5474


WWW
« Reply #25 on: February 18, 2012, 12:29:38 PM »
ReplyReply

EDIT: I also don't know whether modern LED lighting can be corrected consistently with profiles. Anyone have any experience with this?

If you are talking about LED lighting rigs for photography, they definitely fall under the weird spectral output...I've seen some RGB LEDs that allow you to dial in a color output–those are pretty nifty. But white light LEDs have a unique spectral output. They are not really full spectrum.
« Last Edit: February 18, 2012, 12:34:00 PM by Schewe » Logged
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1154



WWW
« Reply #26 on: February 18, 2012, 04:45:11 PM »
ReplyReply

Kirk Tuck just came out with an LED lighting book...

http://www.amazon.com/LED-Lighting-Professional-Techniques-Photographers/dp/1608954471

Not crazy about the look of skin tones lit by these lights and shown in that book if it's accurate looking. I keep seeing a slight green/cyan patina permeating throughout, but it's subtle.

In all these type of odd spectra lighting an increase in contrast usually amplifies this patina which is probably why Tuck shows his images with a somewhat flatter dynamic range.
« Last Edit: February 18, 2012, 04:51:25 PM by tlooknbill » Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5474


WWW
« Reply #27 on: February 18, 2012, 06:05:04 PM »
ReplyReply

Not crazy about the look of skin tones lit by these lights and shown in that book if it's accurate looking. I keep seeing a slight green/cyan patina permeating throughout, but it's subtle.

Yep...that's one of the issues of white light LEDs...however, if he had done a custom DNG profile, I suspect the results would be much better.
Logged
Mike Guilbault
Sr. Member
****
Offline Offline

Posts: 817



WWW
« Reply #28 on: February 18, 2012, 10:26:41 PM »
ReplyReply

Thanks Jeff & Oscar... those replies pretty much cleared it up for me (I hope!). 
Logged

hjulenissen
Sr. Member
****
Offline Offline

Posts: 1679


« Reply #29 on: February 19, 2012, 01:13:23 AM »
ReplyReply

Those are all white balance changes, not spectral illumination changes. All forms of "daylight" come from the same source, the sun.
I am sure that you are right in that WB is the right tool to compensate for different natural light during the day if one has a profile built from daylight (or a proper similar source). I dont think it really matters if the source stays constant and the filtering change, or the other way around, though.

I think that the reason why WB "works" must be that certain classes of illumination have variation that is very "smooth" and can be modelled by a single parameter (the spectrum of a black body heated to different temperatures).

If the sun or the atmosphere contains large peaks and dips, or large "chunks" that have very low energy, then I cant see how a simple model could model this accurately. Then a full color characterization should be needed.

-h
Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5474


WWW
« Reply #30 on: February 19, 2012, 01:22:27 AM »
ReplyReply

If the sun or the atmosphere contains large peaks and dips, or large "chunks" that have very low energy, then I cant see how a simple model could model this accurately. Then a full color characterization should be needed.

Correct, more or less...I seriously doubt the SUN will change much regardless of the atmospherics involved. But weird or non-standard spectral output will make a difference. That's when a custom DNG profile will be useful.
Logged
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7407


WWW
« Reply #31 on: February 19, 2012, 02:18:47 AM »
ReplyReply

Hi,

But the dyes on the Color Checker may have very different properties than the surface of your subjects. So if you have a spiky illumination spectrum the CC may be of little help.

Best regards
Erik


Correct, more or less...I seriously doubt the SUN will change much regardless of the atmospherics involved. But weird or non-standard spectral output will make a difference. That's when a custom DNG profile will be useful.
Logged

madmanchan
Sr. Member
****
Offline Offline

Posts: 2108


« Reply #32 on: February 22, 2012, 02:36:18 PM »
ReplyReply

As Jeff and others have noted, what really matters to the relevance of a custom DNG profile is the spectral radiances that you are photographing:  that means the spectral characteristics of the illumination, and also of the materials you're photographing.  The DNG Profile Editor and X-Rite ColorChecker Passport software are aimed at helping you with the former.  (The fact that they use a fixed chart means that the process will work well for many materials, but not all.)

It's unlikely that a custom profile will help you with one flavor of daylight more than another.  This is because natural daylight has a pretty full and reasonably smooth spectrum whose changes can be well approximated by per-channel scaling (i.e., simple white balance).  You'll see the benefit of additional custom profiles once you start working with light sources that are spectrally quite different (e.g., compact fluorescents).
Logged

sgs8r
Newbie
*
Offline Offline

Posts: 3


« Reply #33 on: March 18, 2012, 03:52:41 PM »
ReplyReply

Hope I'm not too late too jump into this thread...because I'm still a bit confused.

I've always understood building a profile using the ColorChecker Passport to be kind of a more sophisticated version of white balancing. The software knows what each patch in the CC is supposed to be (presumably under some standard illumination), compares with what the device (camera-sensor) has recorded, and uses the resulting differences to create a "correction function" covering the entire color space. This "correction function" is basically the profile. Applying the profile would adjust all the recorded values in the image (in the Lightroom preview) using the correction function. Image colors matching the CC values would be corrected perfectly (or close to it) and other colors would be close subject to the limitations of using only the 24 CC values to span the entire color space.

So since white (actually 6 shades of grey) are among the CC values, white balance would be taken care of automatically as part of the overall profile correction.

Clearly this isn't the case because using the LR eyedropper has a noticeable effect (generally a good one) even after the profile is applied. So maybe the profile is only a relative correction, i.e. it adjusts the recorded values only to restore the relative differences among them (working from some kind of "center-of-gravity" among the CC colors)?

This might explain something else that has bothered me...using profiles without removing ambient light characteristics you want to preserve  (e.g. magic-hour light, or nightclub lighting). The sensor sees the combination of an objects color characteristics and its illumination---can the 2 components ever be separated?

Finally, the whole notion of "color temperature" seems confusingly limited. How can a 1-dimensional measure capture the characteristics of an illuminant when human color perception is already three dimensional and actual illuminants are effectively infinite dimensional. Clearly I'm over (or under?) thinking this.

Thoughts?

-Steve  (heading off to Wikipedia...)
Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5474


WWW
« Reply #34 on: March 18, 2012, 11:01:21 PM »
ReplyReply

I've always understood building a profile using the ColorChecker Passport to be kind of a more sophisticated version of white balancing.

Completely wrong...sorry. Making a DNG profile has nothing to do with white balancing (other than you do need have a raw capture that is a relatively well exposed raw image and a consistent, repeatable spectral source). Making a DNG profile is all about profiling the spectral response of a sensor under different spectral illumination. Two really divergent things. You really need to ignore what you think you know and understand the basics. A sensor has a certain response to certain spectral illumination. The purpose of the DNG profile is to finger print that response. You'll need one profile for daylight, but it will be less ideal under tungsten, so it's useful to do one under tungsten as well. Te best of both worlds is to do a profile under dual illuminations...

Reread what has been written...it's all about profiling how a sensor responds to light and how it renders color. White balance is a color correction adjustment not really a color rendering adjustment...you need to clear your head of that misimpression.
Logged
dnmiller4
Newbie
*
Offline Offline

Posts: 17


« Reply #35 on: April 19, 2012, 08:04:35 PM »
ReplyReply

First, I need the definition of what a DNG profile is.  The phrase is used so much, and don't know where or when to put it in my CMS thinking.

Equipment is: Canon EOS 1Ds MK II, MacBook Pro 4.1, OS 10.6.8.  Lighting is Paul C. Buff White Lighting 1800's.  I use two.  Polarizing filters on lights.   Tube are UV coated.   Polarizing filter on the 100 mm Macro 2.8 L.   

White balance for Canon EOS 1Ds MK II.   Use the Illuminator by Westscott, first
the 3 shades, black, gray and white, to get exposure correct in the camera.  I am not leaning to the right (if you know what I mean).

Next I turn the Illuminator by Westscott over and have a full white target.    I image this target in camera, and create the WB.  I repeat this procedure.

Now, in my mind, the 1Ds MK II is not only WBounced, it has the correct exposure range.

Next, I image the CCP target, at the above illuminates and exposure range.
I use the image file to create the camera profile, using DNG profile editor.  This has it's own name in the camera profiles.

I find that the first white patch, bottom left, of CCP to have the same values as the Illuminator by Westscott's full white target.   What this tells me, is the instructions written and verbally given to me for white balance with CCP has been incorrect, misleading and false.

Verbal and written instructions tell me to use the almost white target of the CCPassport for my WB.   Just doesn't work for me.   Why?   I don't know.   This target is supposed to be better than plain typing paper or even the Illuminator by Westscott. 

Then for me, there is the issue of whether to use the CCP software to create a camera profile, or use the Adobe DNG Profile Editor.   From my experience, the Adobe DNG Profile Editor's camera profile more closely matches the original art that I must reproduce, faithfully.

These image files are not for my use.  But for client's use of their original art.   I print to canvas and fine art papers.   Everything is color managed.  NEC 271w, with SpectraView, printer/substrate/ink, and I hope the 1Ds Mark II with lens, lighting and environment.

So, I admit I don't know some of the terminology used in this forum.  I appreciate any help!

I also would appreciate help and comments on what I have written.   Especially regarding whether to be using the CCP software for creating a camera profile, or the Adobe DNG Profile Editor for creating a camera profile.

When adjusting exposure, after I have done the preliminary steps listed above, I change the diaphragm setting, not the intensity of the lighting.   And yes, I have recently been including the CCP target inside the frame of an image file for each different lighting exposure.

David










 
Logged
lpr
Newbie
*
Offline Offline

Posts: 9


« Reply #36 on: April 30, 2012, 11:57:27 AM »
ReplyReply


The main times you would want to do additional custom DNG profiles is where the illumination is somehow different than sun or tungsten. LED lights for example would need their own DNG profile. Or HMI lights. Calumet sells fluorescent light panels for stills or video. I would definitely make custom DNG profiles for these lights...

So, what if I am a capture one user?  Does this mean if I shoot with HMI lights I am condemned not to get the best our of my sensor because capture one does not work with DNG profiles?  Or is there something I should be doing in capture one, other than the WB of course, so adjust my shots to HMI light conditions?
Logged
Kaa
Newbie
*
Offline Offline

Posts: 4


« Reply #37 on: May 03, 2012, 10:07:24 AM »
ReplyReply

Well, reading this thread helped but I'm still confused.

First, how my confusion started :-) I took a picture of the CC Passport (under tungsten lights), converted the RAW to a DNG, and made *two* profiles from the same original image. One profile was made using the CC Passport software (drag-and-drop the image, press the button, done!) and the other one was made using Adobe's DNG Profile Editor (drag-and-drop the image, tell it it's a ColorChecker, export). One difference is that the CC Passport software made a single-illuminant profile and Adobe's PE seems to have forced me into a dual-illuminant profile even though only the 2850K color table was "edited".

Next I opened still the same original image that the profiles were made from, opened it in ACR and applied the profiles. And, um, the results from the two profiles are very different. In particular, the two profiles required quite different WB setting to achieve the grey patch neutrality -- e.g. for the Tint slider in ACR (green-magenta axis) the PE profile was more or less neutral (+1, I think), but the CC profile required +8 to get to neutral grey. I am not at all sure that's how it's supposed to happen and I'm not sure that I'm doing things right. Advice, comments?

Now, some theorycrafting according to my limited understanding. For any source of light you can measure/plot the distribution of energy across the visible spectrum (aka the spectral response, etc.). Some sources have a more or less continuous spectrum and some have spikes and holes. One notable abstraction (and/or model) of the spectrum is blackbody radiation at different temperatures. We think that this model approximates daylight (and incandescent lamps) pretty well. So far so good.

Let's take noon daylight, deep shade on a sunny day, and a sunset. The spectra of light in these three situations will all be different. The claim is -- as far as I understand it -- that we can use a simple function (called "white balance") to transform these different spectra into some baseline that we understand and that we know how our sensor responds to. OK. But let's take, say, a sunset and a tungsten bulb. Both can be more or less approximated by a blackbody spectrum at certain temperature, right? But someone in this thread mentioned that these two spectra are actually quite different and require different treatment. How so? Is the sunset spectrum more like the deep-shade spectrum than like a tungsten spectrum? In which ways?

The advice that I'm reading here is that most any continuous spectrum can be well approximated by a single "line" connecting two points -- daylight and tungsten -- with the white balance setting determining the specific position on that "line". But again, if applying a tungsten profile to a sunset image screws you up, this model of a line between two points stops looking like a good approximation and I don't understand under which conditions it breaks down (we're still talking only about continuous-spectrum sources, nothing spiky).

In terms of spectrum energy distributions, would it be correct to say that a DNG profile tries to match the *shape* of that distribution while the white balance determines the *location* of that shape on the energy axis..?
Logged
John Cothron
Full Member
***
Offline Offline

Posts: 170



WWW
« Reply #38 on: May 04, 2012, 10:15:55 AM »
ReplyReply

I've created single illuminant and several different dual illuminant profiles with the color checker.  For studio work I can see the benefit, it does make SOME difference.  IMO however, Adobe has done a pretty good job with their standard profile (at least for my cameras) and for the kind of shooting I do most of (landscape) it works real well as is. 
Logged

Electromen
Newbie
*
Offline Offline

Posts: 32


« Reply #39 on: May 06, 2012, 05:02:47 AM »
ReplyReply

It's my experience that if the ambient light is low and the scene is lit with 100% flash, then I only need one profile using the Passport.  I can use it with all my flash work from month to month.
If I'm using a slight fill flash, then the same profile doesn't work from scene to scene.  The ambient is the main light and it will change.  In mixed lighting like in a church, I'll may create several profiles during one wedding as the light changes.
Logged
Pages: « 1 [2] 3 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad