Using X-Rite Colorchecker as WB reference

[forward06]

Hello everyone,

I use X-Rite Colorchecker for my WB refernce, but in the shade images look to yellow to me. (I am using D65 calibrated monitor)

In the comparation above I adjusted wb manualy on the left, and used natural wb patch from colorchecker on the right. (colorchecker was captured in the previous frame)

When I look at RGB reading in the photoshop on boys shoe (white area), reading is more natural on manual wb.

Which photo look more neutral to you (on calibrated monitors)?

Does somebody else experianced that gray patch wb not work well in some conditions?

PS. Both photos are shoot in RAW developed in LR, only diference is that I manualy moved temperature slider on the left image.

Thank you.



 

[Bart_van_der_Wolf]

I use X-Rite Colorchecker for my WB refernce, but in the shade images look to yellow to me. (I am using D65 calibrated monitor)

Hi,

The adjusted image looks too red/orange to me. Which raises the question, how did you shoot the ColorChecker?

When in a natural surrounding, your CC will be illuminated with lots of ambient (grass or soil color, tree bark and canopy, blue sky, white clouds, etc.) and direct illumination, although in the shade it's mostly a mix of reflected light. That means that the angle/positioning of the CC becomes important. IMHO, one should position the CC at the subject position and perpendicular to the optical axis of the lens.

Then that will allow to create a neutral WB, but that's only the starting point. From there you can add ambiance, e.g. add a bit of a green tint when under tree canopy, or a bit of blue if you want to stress the shadow appearance, or leave it at correct color balance if that matters more.

Also note that the brightest step (#19) of the colorchecker is not as neutral as the second brightest step (#20).

Cheers,
Bart

[forward06]

Hi Bart,

Thanks for the reply. I beleve that most of the light is coming from the blue sky (that can be seen in the subject eyes when review photo in 100% magnification). I shouldn't have strong color cast from anbient, because all around is concrete which is almost neutral in color. Light colored sand beach was maby 30m away, I don't think that's enough reflection to introduce strong color cast.

I didn't shoot colorchecker at angle that you suggested, but I use step #20 as neutral reference.

I generally have a feeling that photos are on the warm side when using gray card as a reference in the shade.

Maybe my brain is used to see shadow as more bluish, so actually good color balance look yellowish to me...

[Bart_van_der_Wolf]

I didn't shoot colorchecker at angle that you suggested, but I use step #20 as neutral reference.

I've found that it can make quite a bit of difference how the WB reference is tilted. of course positioning it at the subject is not always possible, but it helps when there are differently colored reflecting or light emitting surfaces near the main subject. And of course three-dimensional subjects will have different color balance as well, at varying reflection angles to the illumination sources.

I generally have a feeling that photos are on the warm side when using gray card as a reference in the shade.

There should not be any bias when you click-WB on a spectrally neutral reference, so the only plausible reason (other than a CC that's dirty or old) can be in the reference shot being different than the actual subject. Try and find/eliminate any difference between the shooting conditions of both, and predictable results should follow.

Maybe my brain is used to see shadow as more bluish, so actually good color balance look yellowish to me...

Psychology certainly plays a role in color vision, because our vision continuously adapts to scene content, but you should be able to verify with an RGB sampler whether the actual color is more or less neutral as a starting point.

Cheers,
Bart

[AlterEgo]

Which photo look more neutral to you (on calibrated monitors)?

my $0.02 - ignore what is "neutral", instead adjust WB to make skin tones look like you like them to be - you don't really care about the color of their clothing or tree's trunk or pavement, do you ?

[AlterEgo]

I didn't shoot colorchecker at angle that you suggested, but I use step #20 as neutral reference.

BabelColor website has averaged data from 30 colorcheckers (not passports) :

ID#, name, L*, C*, h*

24   F4   20.813   0.38   274.3
12   C4   66.484   0.53   180.1
16   D4   50.828   0.66   191.7
8   B4   81.210   0.70   156.6
20   E4   35.854   0.73   222.1
4   A4   96.529   2.47   101.1

so for WB you typically want a good exposure (to minimize S/N) of the patch to be used and typically a brighter patch from the bunch and most neutral of course, hence this one is good "12   C4   66.484   0.53   180.1" (3rd from the left in greyscale row), when exposed properly

[AlterEgo]

I generally have a feeling that photos are on the warm side when using gray card as a reference in the shade.

if you have more blue in the light reflected in the end from the patch towards you(r lens) for whatever reason then WB off that patch in that shot will get you more yellow ( RGB - B = RG = yellow = warm ) cast...

[Some Guy]

Pass this around and you'll get a whole bunch of "This one is better than that one."  Very subjective to the eye.

My monitor is either D50 or D55 since I'm a printer.  The warmer one on the right is what I'd go for as the left looks too cold and flat for me.  Just imho though.  Maybe a 50-50 split too.  Sometimes I've printed an image and stared at it for days debating which way to go or if it's even right.

Guess you could use a color temperature meter and dial that into your camera.  Software has too many designer's theoreticals as to what Kelvin it ends up being.  Move around between various sofwares like Corel PaintShop, Adobe, Zoner Labs, and the Manufacture's converter and you may see it shift by 1,000K between them too.  Maddening!

Expodisc 2 can be used with some tunable cmaeras that show a RGB color histogram where the center spike (If the exposure is correct to center the spike!) may show a color temp offset in that the RGB colors will gather around the spike and not be contained within it sharply so it is one white spike and no colored shoulders.  If you tune the Kelvin and Tint, the spike can be cleaned up and you'll know the camera's set about as good as you can get for the lighting.

If you went by the book on CMYK coloration percentages following Lee Varis' "Skins II..." book, the values on the right might be closer to that in the book.  That's if you like to play with by-the-numbers than subjective looks too.

Good luck!  And why one cannot get a lab to even agree on what's right if you send same image to 10 labs you get 10 differing results.  Even re-submitting the same image comes back different looking at times from the same lab.

SG

[Tim Lookingbill]

I deal with this kind of shaded/overcast WB issue on a daily basis. I wish I could develop a one click neutral target reference WB system but it rarely gives the desired results just as you're experiencing.

The CC works perfect and consistent in controlled studio situations but outdoors I've noticed hue shifts throughout the day both in shade, overcast and daylight that rarely give consistent results. Maybe I get lucky usually in bright cloudless midday lighting by shooting Raw and selecting either Daylight or AWB in camera.

With this setting ACR/LR most of the time interpret the WB but a lot of times the green/magenta tint slider needs most of the adjustment. Lately I've been leaving the incamera WB setting on Daylight where shade/overcast tends to look like your warm version on the right.

Since I now know of this issue out in the field I just take note of the level of saturation/hue of objects that should look blue (i.e. the white wall behind your two subjects) which from my observation should have a noticeable magenta (violet, not cyan) hue of blue as in your CC shot on the left which looks too cyan blue and overall too cool looking. Shade is somewhat warm but not as much as the one on the right though.

Also not mentioned is the profile (or Picture Style) chosen in your editor if you happen to use one when shooting Raw. My custom profile makes Caucasian skin tone Lab readings match closely the 60L, 18a, 18b numbers of the CC chart skin patch making them appear kind of reddish in overcast warm-ish light. So I revert to ACR4.4 profile which makes them yellowish orange and adjust the tint slider from +13 to +18 adding more magenta but it's different with every shot when shooting outdoors depending on surrounding greenery and whatever particles are floating in the air and overcast clouds that makes the camera shift hue.

[Tim Lookingbill]

A correction, I meant to say the Xrite CC version is too warm, the one on the right. I got the two flipped.

Just look for something in similar scenes that should have some blue and adjust from there.

For example note in the screenshots below the color of aluminum silos in this overcast scene should look somewhat blue while making sure the green tree leaves don't look cooked while the overcast cloud shadows should have some blue but not too much.

The first is As Shot (5250K/+13) with incamera WB set to Daylight.

The second is setting WB in ACR to Daylight (5500K/+10) which warms it up.

The third is clicking for R=G=B on the cloud shadow which slightly shifts the tint to magenta & warmer (5700K/+14) all three using a custom DNG profile.