P: Add options for larger sample sizes for White Balance Eyedropper Tool

[Todd-Russell]

Please see full thread here:

http://forums.adobe.com/thread/854734...

When adjusting images in LR I believe we all suffer from some degree of "color bias" or just plain "eyeball" inaccuracy! LR's single (1) pixel eyedropper white balance sampling tool provides very little assistance with "automating" white balance adjustments, since even normal camera “noise” causes readings to fluctuate widely. X-Rite's ColorChecker Passport plug-in samples a large area of each color patch in LR, and then uses the "average" of these pixel readings to calculate the 24 color patch values. The area sampled for my 5D MKII 21 Mpixel ColorChecker Passport images are approximately 100 x 400, or 40,000 pixels. This is a totally "automated" process, only requiring human intervention to establish the correct overall white balance with the new ColorChecker Passport camera profile. Unfortunately, LR's "limited" eyedropper tool makes this simple part of the process extremely cumbersome, inaccurate, and "unautomated!"

It would certainly make sense (and be very simple) for Adobe to add the option of using a larger user selectable area of the image with the eyedropper tool that is then “averaged” to determine the white balance. Without question, this will provide much more accuracy for those of us who use calibrated color targets like the ColorChecker Passport or even a simple 18% gray card, then using the current one (1) pixel eyedropper tool.

There may be times when a "single pixel" sample is the best way to determine white balance, but please do not tell me that "noisy pixels" in a large pixel sample area will make the readings inaccurate. That is the whole purpose of using larger sample size – To make the calculations more accurate. I am an EE with 45 years of analog and digital design experience, and have substantial experience with much more complex systems than LR's white balance tools. And lest we forget, film photographers have been using 18% gray cards for over 50 years. The analog colorimeters used for establishing white balance had a much larger sample size than anything near one (1) pixel size, and the accuracy was superb! Using a 100% digital workflow we should be able to do much better! By adding an option to the white balance eyedropper tool to select an area of pixels that I user adjustable I believe Adobe can achieve this objective.

[TheDigitalDog]

White Balance is dependent on the zoom level. If you want to average over a larger area, just zoom out.

[Bob_Peters]

It is possible to change the number of pixels shown in the white balance tool. However, that does not alter what appears to be a 5x5 sample size.

I would also like to see changes to the size of the sample so that the number of 'pixels' used agrees with the number of squares in the grid shown on the screen.

[tommaso ferrarese]

also I'd add ( actually I think it would be more important ) that wb sampling should be affected by NR setting, thus giving a reading that's pertinent to final render.

see full thread as linked above for better definition of the matter

[Rikk-Flohr: Do not Use]

LR 4 has zoom dependent White Balance Sampling.
http://blogs.adobe.com/lightroomjourn...

[Todd-Russell]

The LR4 Beta White Balance eyedropper tool can be set using the scale slider to sample up to a 17 x 17 pixel area (289 pixels) with averaging. I'm not sure what Adobe means by this statement "White balance sample area is now zoom-level dependent," since it appears the eyedropper 'Scale' slider setting is the only thing that affects the sample size. My ColorChecker Passport image readings are now more consistent with less variation when sampling different areas of the gray scale patches.

This is indeed a thankful improvement, but it would be better if Adobe allowed the scale slider to bet set to slightly larger samples sizes, perhaps up to 100 x100 (10,000) pixels. This shouldn't increase the eyedropper sample processing time significantly and will provide even better accuracy when setting the white balance on high ISO noisy images.

[Lee Jay]

If you are at 1:1, you get a smaller sample size than if you are "fit" view, even if you have the loupe at the same size in each case. At least, that's how I understand that statement.

[Todd-Russell]

Thanks Lee.

I went back and checked the Zoom function more carefully. The White Balance eyedropper loupe maps 1 to 1 pixel at 1:1 Zoom view. As you drop the Zoom level it does increase sample size linearly, so 1:2 view maps 2 pixels to each loupe box, 1:4 view 4 pixels, etc. At the minimum zoom level of 1:16 the White Balance eyedropper will sample an area of 272 x 272 pixels (73,984 pixels) with the maximum scale setting (17 x16 = 272).

I tried using the maximum sample size at 1:16 zoom on my ColorChecker and there is virtually no spot sample variation. It works very well! LR 4 Beta provides White Balance sample sizes from 5 x 5 pixels up to 272 x 272 pixels – This is great!

My complements to the Adobe Photoshop Lightroom Development Team!

[slarti3]

See simonsaith's last response in this thread:
http://forums.adobe.com/thread/947832...

According to him, now the entire WB window area should always be sampled. By "now" I mean the next LR 4 release.

[Todd-Russell]

Very good news – This will make the LR4 White Balance eydropper much more useful!

[Todd Shaner]

The White Balance Eyedropper tool appears to have changed in LR 5, 6, and CC2015. It now remains at a fixed 5x5 pixel sample area regardless of Scale or Zoom view settings. Please see the problem report here for further details:

http://feedback.photoshop.com/photosh...

[Todd Shaner]

UPDATE: It appears only the RGB values displayed in the eyedropper tool are incorrect. The applied WB correction uses the actual Scale setting sample area. See this post: https://forums.adobe.com/message/8929708#8929708

It still can cause confusion and should be corrected to show averaged RGB values in the eyedropper tool for the actual Scale setting sample area.