Inconsistent white balance with LR eye dropper

I was under the impression that using the eye dropper on a neutral shade to set WB (works anywhere from white to black as long as it's not pure white/black)

But I find the WB will change depending on the brightness of the clicked point. (Like along a sleeve of a white shirt, from a bright point to a shadier bit). The shirt is all the same colour, but the eye dropper doesn't work as I expect.

Thoughts? I think using a grey card would make my like easier, but I mainly want to know if I'm using the eye dropper tool wrong.

A couple thoughts -
1. A white shirt partially in the shade is in different illuminations; outside, for instance, the well lit section is mostly illuminated by light from the sky, a clear blue sky perhaps, while the shaded portion is lit by light reflected from the environment - grass, trees, a painted structure, etc. The WB seeks to neutralize the predominate illumination.
2. Noise. The lower the S/N in the sampled area, the more the WB calculation can be thrown off by noisy pixels. It follows that a well exposed area will provide a better sample. One popular WB target, the WhiBal card, is 70% grey, rather than the traditional 18% card for determining film exposure, for this reason.
3. LR provides tools (attempting) to treat the problem.
A. Since LR 4 it has been possible to alter the size of the WB probe's sample from 5x5 up to (IIRC) 17x17 and to see the sampled area 1:1. A larger sample, by averaging a larger number of pixels, can reduce the influence of the noise.
B. The local adjustment tools can alter WB.
C. Split Toning can, in effect, alter WB in highlights and/or shadows.

I find LR's WB dropper fairly good most of the time, but it is finicky. Carry a sheet of white printer paper with you. Custom WB is way too easy to do to get caught up in the mess of fixing it in post. If I am not in AWB, I am in Custom. Everything else is a guess.

I will probably get considerable grief for this, but an 18% grey card is totally NOT necessary for custom WB. The 18% grey is for exposure. Just about anything white will work for custom WB. I've used linen napkins, paper towels, note book paper. My guess is toilet paper would work as well. You don't need 18% grey for WB.

Bassat wrote in post #17928982
I find LR's WB dropper fairly good most of the time, but it is finicky. Carry a sheet of white printer paper with you. Custom WB is way too easy to do to get caught up in the mess of fixing it in post. If I am not in AWB, I am in Custom. Everything else is a guess.

The process performed by the camera to create a CWB is essentially the same as that done by any Raw converter with a WB probe, although there can be differences in implementation and the sample used by the camera firmware is often larger than that used by some converters . But if you have a scene with mixed illumination, you have the same problem in deciding where to put your piece of paper as you would have later in deciding where to put the eye-dropper.

True, but you do need grey.

tzalman wrote in post #17928996
The process performed by the camera to create a CWB is essentially the same as that done by any Raw converter with a WB probe, although there can be differences in implementation and the sample used by the camera firmware is often larger than that used by some converters . But if you have a scene with mixed illumination, you have the same problem in deciding where to put your piece of paper as you would have later in deciding where to put the eye-dropper.

True, but you do need grey.

It seems you are way overestimating what I am doing for CWB. Pick up table napkin (or whatever). Fill frame. Release shutter. Done. I can't do anything about mixed illumination. I don't waste time worrying about it. WB for the subject lighting, forget the rest.

I have never used a grey card for custom WB. As mentioned, I use printer paper (90-94 bright, table napkin (paper or linen), toilet paper, or a beer coaster works, as long as it is white. I fill as much of the frame (usually 100%) as possible. I have never had problems doing CWB this way. IMHO, the '18% grey' is for exposure only. It has nothing to do with WB.

Ok, that's cool - WB isn't a big deal for you. Personally, I agree with you; in people pictures pleasing or appropriate skin tone is more important to me than "accurate" WB (they often don't go hand in hand). In scenics/landscapes/nat​ure subjects I shoot with a Uni-WB CWB that makes everything processed green and adjust WB by eye on a calibrated monitor. But that's not what the OP asked. Clearly he is concerned about WB accuracy (rightly or wrongly) and is wondering about inconsistent probe readings in LR.

IMHO, my last Kodak 18% Grey Card became irrelevant for exposure calculation the day I retired my film cameras.

Agreed.

Just to re-iterate. I find the LR WB tool to be fairly good, but sometimes inconsistent and finicky. Which leaves me 'guessing' as to what is correct. I usually end up settling for what looks good enough. Which is why I am such an advocate of CWB. Oh, when I have to shoot CWB, I also have to shoot raw. I want the latitude.

Agree about the film camera exposure, too. AE is the only thing I value more than AF. As my eyes continue to deteriorate, that order may change.

tzalman wrote in post #17929012
IMHO, my last Kodak 18% Grey Card became irrelevant for exposure calculation the day I retired my film cameras.

Yep. I used to do all sorts of things to get exposure correct in tricky situations. One of my favorites was to turn the reflective meter in my AE-1 into an incident meter by putting a styrofoam cup over the lens. It worked, but I am so glad that I don't have to mess with that any more.

Bassat wrote in post #17929026
Agreed.

Just to re-iterate. I find the LR WB tool to be fairly good, but sometimes inconsistent and finicky. Which leaves me 'guessing' as to what is correct. I usually end up settling for what looks good enough. Which is why I am such an advocate of CWB. Oh, when I have to shoot CWB, I also have to shoot raw. I want the latitude.

Agree about the film camera exposure, too. AE is the only thing I value more than AF. As my eyes continue to deteriorate, that order may change.

Bassat, I recently spent a considerable amount of time creating LR 5.7 HSL and Camera Calibration profiles for my 5D3. I found that after I got the camera profiled, that WB fine-tuning by eye became much easier (since the deep reds and violet-purples were more accurate).

Something you might consider...

Keep it up people... I have lots to learn here. tzalman, thanks for the input!

Would changing the sample size help?

Changing to a sample size averages more pixels so statistically you ought to be closer to 'correct'.

If we think about it, a sensor is a collection of 'sensels' each of which is sensitive to ONE color in the R-G-B triad. And then Bayer processing looks at the values of the sensel's neighboring sensels to determine a single R-G-B value to assign to that location and call it a 'pixel' rather than a 'sensel'. So a bunch of math is used to determine each R-G-B reassignment, so it is no wonder that any single pixel is not necessarily perfectly R=G=B when we sample a neutral area of our target.

By averaging a larger sample of pixels, each one deviating from R=G=B by a different direction (color) and amount (magnitude), we statistically can be closer to getting a neutral 'mean' (R=G=B)