Pages: 1 [2]   Go Down

Author Topic: Gray card and exposure  (Read 6041 times)

shewhorn

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 537
    • http://
Gray card and exposure
« Reply #20 on: August 09, 2010, 04:11:01 pm »

Quote from: peerke
Hi Joe,

It was daylight and the gray card was in a light tent to make the light diffuse. The shots were taken after each other, so I imagine I can only blame fluctuations in a lightbulb some 149.600.000 km kilometers away. Come to think of it, even though it was quite sunny, moving clouds could give a more obvious explanation...  

Tom

Definitely blows my theory. Clouds as you mentioned can change the light temp. If something moved in relation to the light tent that could reflect a different light color and that could have an effect (especially if you were wearing a colored shirt OR... if for some reason you waited a certain period of time before taking the second shot (color temp of natural light changes throughout the day).

Cheers, Joe
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Gray card and exposure
« Reply #21 on: August 10, 2010, 12:04:03 am »

Quote from: digitaldog
Neutral by all means. While you could use a gray, its more likely you’ll end up with a cast and, because half all the data in a raw linear encoded file is in the highlights, the recommendation is to WB while in a gamma corrected image, the recommendation to gray balance.
This explanation is reminiscent of the erroneous rationale for exposing to the right because the brightest f/stop contains half the levels of a linear file. Actually, as Emil Martinec explains, the number of levels has nothing to do with the true rationale which is that the signal:noise increases with exposure. The advantages of ETTR are most apparent in the shadows rather than in the highlights.

If you read a brighter patch, you will have a better S:N, but the differences are not marked as shown below:

[attachment=23594:ck_patch2.png][attachment=23595:ck_patch4.png]

Actually, one can estimate the SNR for the green channel. The full well capacity of the D3 is about 65568 electrons. The ODs of squares 3 and 4 of the ColorChecker are 0.23 and 0.7, giving reflectances of 58.9% and 20.0% respectively. Assuming that exposure is such that 100% reflectance would yield full well, one would collect 38600 and 13083 electrons from the respective patches. Since shot noise is sqrt(electrons), the noise would be 196 and 114. The SNRs are 196:1 and 114:1 respectively. If you are using a point and shoot camera, the SNR would be considerably less and one would expect to get better results by reading higher luminance squares.

Regards,

Bill
« Last Edit: August 10, 2010, 09:14:25 am by bjanes »
Logged

Steve Weldon

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1479
    • Bangkok Images
Gray card and exposure
« Reply #22 on: August 10, 2010, 01:08:33 am »

Quote from: shewhorn
Definitely blows my theory. Clouds as you mentioned can change the light temp. If something moved in relation to the light tent that could reflect a different light color and that could have an effect (especially if you were wearing a colored shirt OR... if for some reason you waited a certain period of time before taking the second shot (color temp of natural light changes throughout the day).

Cheers, Joe
You mention a shirt..  and this brings to mind an induced shadow by moving close enough to the gray target to 'fill up' the recommended area necessary for an in-camera custom WB.. at some angles no shadow, other angles a shadow.  A bird flying over, wind moving a tree, there are a ton of explanations and they're all legit when you're talking small differences.. which goes to what DD was saying about 'accuracy', a word when working with color really needs to be quantified.

I think what works.. is for each person to understand the basics and then develop a style which fits them best.  Location matters because of surrounding trees/buildings/mountains/people, weather matters, time of day matters (shifting angles for shadows), and of course a camera doing an in-camera WB could vary.. and certainly operator intervention in post can vary.   So.. develop a style, pay attention to the details and variables, and do your best to be consistent.

I've found what works best for me.. is shooting the grey slab and doing an in-camera WB on location.  I'd guess this is because my camera is sampling a much larger area than an eyedropper in post samples and perhaps uses a more sophisticated custom WB scheme.. and this could vary by camera.  Then, my WB is correct for all subsequent shots including the color checker slabs.. so when building the color checker passport profile it's going to build the most accurate profile.  

This might not be the best workflow for everyone, but it works for me.  At least when shooting people towards the end of the day which is my habit.  I might develop slightly different workflows in other circumstances.

I will say this.. I did some pictures of a boy wearing a bright blue shirt.. the sort of bright blue you never expect to perfectly match.  The in-camera or post WB settings didn't quite nail it, it was still several shades off.  Hand tweaking couldn't nail it.. and I've been hand tweaking WB for a long time like most of you.  The CC Passport profile nailed it on the first try while keeping the skin tones exactly the same (to my eye) matching the color of the blue shirt.  A significant difference.

This tells me where color is considered, there is a lot more than the greys/blacks/whites to color balance.. and is exactly the reason Xrite developed this device.  It's contrary to our normal workflows and any conventional wisdom.  Its almost like magic.. in that I can't duplicate the results in post without the profile no matter how many hours I sit there trying.  I think this is because the profile is built taking the entire block of colors into account.. while manual adjustment we're using just a few.. or concentrating at once on just a few.

I encourage everyone to follow their directions to the letter.. in several different environments.. especially if there is a bright/odd color you're trying to nail.. and see how it works for you.

I would have scoffed at this a few weeks ago.. but when I wrote a review on the thing I tend to use the most scientific methodology I know.. and in this case I was surprised.  All my other WB devices save or the big Kodak gray card I use in the studio have been tossed in the "not gonna use it anymore but I don't want to throw it away" box..
Logged
----------------------------------------------
http://www.BangkokImages.com

peerke

  • Newbie
  • *
  • Offline Offline
  • Posts: 18
Gray card and exposure
« Reply #23 on: August 10, 2010, 02:03:40 am »

Well, this discussion has certainly been instructive for me so far. I Have been fiddling around with using the in-camera WB setting versus use of the eyedropper in LR, and I think I can get myself used to doing the first (I am not the most disciplined photographer around...). However only having to shoot the color card and use the second patch as a WB reference in LR as Andrew pointed out, should be sufficient as well.

What still puzzles me is that the colorchecker SW would need an image with the proper WB set, as Steve Weldon pointed out. With some effort I can imagine that for a jpeg, but not for a raw image. Unless the software would use the WB metadata at some point, but since it can use the color card itself as a reference, it wouldn't need that, right?

Coming back to my original question of differences due to exposure, I noticed that when I use the eyedropper in different corners of the same patch, I already get differences of up to 100, so a difference of 200 between shots with possibly different lighting conditions, albeit very minor, indeed should not come as a surprise.

Rgds,
Tom

Logged

Steve Weldon

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1479
    • Bangkok Images
Gray card and exposure
« Reply #24 on: August 10, 2010, 03:15:49 am »

Quote from: peerke
Well, this discussion has certainly been instructive for me so far. I Have been fiddling around with using the in-camera WB setting versus use of the eyedropper in LR, and I think I can get myself used to doing the first (I am not the most disciplined photographer around...). However only having to shoot the color card and use the second patch as a WB reference in LR as Andrew pointed out, should be sufficient as well.

What still puzzles me is that the colorchecker SW would need an image with the proper WB set, as Steve Weldon pointed out. With some effort I can imagine that for a jpeg, but not for a raw image. Unless the software would use the WB metadata at some point, but since it can use the color card itself as a reference, it wouldn't need that, right?

Coming back to my original question of differences due to exposure, I noticed that when I use the eyedropper in different corners of the same patch, I already get differences of up to 100, so a difference of 200 between shots with possibly different lighting conditions, albeit very minor, indeed should not come as a surprise.

Rgds,
Tom
1.  Setting an in-camera custom WB is not difficult by any stretch of the imagination.. the hardest part is remembering to do it.

2.  "Sufficient" yes.. more accurate.. I'd say it depend on the operator, raw package, and camera's ability to build an accurate WB.  I'll go out on a limb and say a decent camera, sampling a much greater area than an eyedropper, will consistently make a more accurate WB.   Of course it can be done in post if you forget.  I wouldn't worry about it either way.. but I will try to remember to do an in-camera because I find it renders a WB more accurately than in post.

3.  If you're just using the color checker and not the Passport Software.. then sure.. shoot the color checker, use the white square to adjust WB, and the rest of the squares will fall in line.  However, if you're using the software to build a profile things aren't so simple because there are some unknowns.  

a.  Does the software first check the white/black/grey squares and set a white balance.. and then build a profile?  I doubt it does.

b.  Or does the software build a profile based on an aggregate of all the colors in the color checker?  I think this most likely.. but I don't know.  Xrite knows if anyone wants to ask them.

c.  It should be obvious that a color checker shot at the wrong WB will not have the colors in the color checker at their optimum colors.. they'll either be too warm or too cool and who knows about the hue?  IF the software doesn't set white balance first.. then it's obviously essential to set the white balance of the color checker first.. before running the passport software.. to get the most accurate profile.

And this brings us back.. to getting the most accurate WB.  If you believe you can do better sampling an eyedropper size of the white square.. or maybe sampling several and averaging them.. then the camera can do using a much larger sample size.. then use that method.   This is where you'll need to try and compare.. and your results will only be as accurate as your methodology and consistency.. and of course as the firmware program written in your camera that does the custom WB.  I'll freely state the Canon 5d Mark II can build a consistently more accurate (using skin tone matching as the yardstick) custom WB using the gray slab of the CCP.. than I can do in LR3 or C1pro using the white/grey/black squares.. at least lots of work.  

Further, there's no way I can match the overall color profile this software creates poking around in LR3, CS5, and C1pro.. and I use these programs for hours daily.

There is obviously more than one way to achieve results.   Notice I didn't say the "same results."  The trick is in obtaining the results that work for you and your workflow.
Logged
----------------------------------------------
http://www.BangkokImages.com
Pages: 1 [2]   Go Up