1 2 Previous Next 47 Replies Latest reply on May 7, 2007 11:23 AM by Panoholic

    ACR 4: White Balance readings off of a target

    Level 1
      In the good old days (ACR 3.7 and previous), I could shoot a white balance target, then open the image in ACR and use the eyedropper to set the white balance. No more. Now, I get a message that states - "The clicked area is too bright to set the white balance. Please click on a less bright neutral area."

      Well, I'm using (and have been using) Danny Pascale's White Target that was made specifically for this purpose. (http://www.babelcolor.com/main_level/White_Target.htm)

      Something has changed, and it's not for the better. I've tried decreasing the exposure in ACR, to make sure the values are far below 255, to no avail. Interestingly, the first measurement "click" does nothing. The second gives the error. The is very reproducible. Something's not right...

      --Rich Wagner
        • 1. Re: ACR 4: White Balance readings off of a target
          Ramón G Castañeda Level 4
          Try a neutral gray card if your current target consistently gets blown out.

          >I've tried decreasing the exposure in ACR, to make sure the values are far below 255, to no avail.

          That's because presumably the tool looks at the raw image itself.

          > Interestingly, the first measurement "click" does nothing. The second gives the error.

          Yes, I can confirm that.

          In sum, it looks like you need to have a neutral gray that is not blown out 255.255.255.
          • 2. Re: ACR 4: White Balance readings off of a target
            Level 1
            Ramón,

            Even if I intentionally underexposeby a third (which I shouldn't have to do), this no longer works.

            This method worked well in previous versions of ACR and was very reliable. As I mentioned, the device I use was specifically made for setting the color temperature. Most gray cards are not spectrally neutral, and gray cards are not practical for me to carry in the field.

            Something has changed in the ACR color temp routine. There may be an error in the routine, since as you mention, the first click now does nothing. It is also possible (probable) that the code is now too conservative in "error-checking" the user for invalid data to use in setting the color temp, as this method has worked reliably for a long time, by a lot of people. The bottom line is the color temp eyedropper has been foobarred.
            • 3. Re: ACR 4: White Balance readings off of a target
              Thomas Knoll Level 2
              Richard: I far as I know, nothing has changed in this feature between 3.7 and 4.0. Are you ABSOLUTELY sure you are getting different behavor when using the SAME image in both versions and clicking on the SAME area?

              If so, please email me sample file with instructions on which area to click on.
              • 4. Re: ACR 4: White Balance readings off of a target
                Level 1
                Thomas,

                Absolutely is a strong word. Let me investigate further today and get back to you with a more detailed picture.

                --Rich
                • 5. Re: ACR 4: White Balance readings off of a target
                  Level 1
                  Thomas,

                  I have gone back as far as ACR 3.4, and I see *no* change in behavior, unlike what I reported earlier. The color temp eyedropper tool is very sensitive to exposure. I routinely set my exposure compensation to -1/3 and had not noticed the problem. My wife does not (different camera/subject material), and I shot with her D2X yesterday in a pinch and had not noticed this.

                  If the exposure is dead-on, the color temp cannot be set by the eyedropper on the BabelColor white target. If the image is underexposed (it doesn't have to be much), the white target works well, and the exposure can be adjusted up. (Yes, there is a down side to underexposure.) Using the same image and ACR 4 (CS3), ACR 3.7 and 3.4 (CS2), there is no difference in behavior of the white balance tool. If the white is "too bright" the first click beeps, and the second displays the error message dialog box.

                  Sorry for the confusion. I would still be interested in your opinion of the use of the BabelColor White Target, as it has essentially no color inconstancy and can also be used to set exposure. I will e-mail you 2 DNGs of shots containing the target - one that "works" and the other that doesn't, that you may use as needed.

                  Thanks,

                  --Rich
                  • 6. Re: ACR 4: White Balance readings off of a target
                    Ramón G Castañeda Level 4
                    Rich,

                    Have you tried the WhiBal White Balance Reference Card ? It has white, black and gray, and fits into a vest or shirt pocket.
                    c
                    • 7. Re: ACR 4: White Balance readings off of a target
                      Thomas Knoll Level 2
                      The reason for the warning in first place is if a camera sensor is driven into clipping or near-clipping for one or color channels (in native camera color space), then that point is not a good point for calculating white balance from.

                      Often the highlight recovery logic in Camera Raw can do a decent recovering enough detail from these areas so the image looks fine.

                      If you want to include a target in your image so you can do click white balance, I recommend a medium to light gray target, not a white target. If you use a white target, you might have to underexpose the image, which is not a good thing since it increases noise in the shadows.

                      Be sure to use a high quality gray card that is neutral. For example the "WhiBal" cards.
                      • 8. Re: ACR 4: White Balance readings off of a target
                        Level 1
                        Thanks.

                        I understand your concerns re: noise in the shadows with underexposure. It should be possible to intentionally underexpose *the test shot* specifically to measure color temperature (and not underexpose the "real" shots), but obviously this is not an ideal workflow. If the exposure is "perfect" is the sensor that close to channel clipping on a perfect white diffuser?

                        Unfortunately, a gray card does nothing to help determine ideal exposure, as does the White Target. Perhaps I should use both. ;-)

                        I assume that the WhiBal cards are spectrally uniform, although their literature does not address this. According to Robin Myers, the gray and black cards are, and their white card is not.

                        I'm also not sure of the value of having a "measurement" of the card. ("And every card is still hand measured, using a precision spectrophotometer, assuring that your WhiBal is calibrated to pure neutral.") I can do this as well, but if the card itself is not neutral, it's inconvenient to white balance off of it.

                        I ordered some gray cards from Robin Myers this morning, before I was these posts.

                        He has a comparison of gray cards posted here, including the WhiBal card:
                        http://www.rmimaging.com/information/neutral_references.html

                        Thanks again for the feedback, and sorry for the false alarm re: the color balance tool.

                        --Rich Wagner
                        • 9. Re: ACR 4: White Balance readings off of a target
                          Level 1
                          "Unfortunately, a gray card does nothing to help determine ideal exposure, as does the White Target. Perhaps I should use both."

                          Actually, yes, that would be proper...but you don't want to set white balance using a greycard which is actually too far down the scale. You want to set white balance with a non-specular "white" that is not clipped. That's what the WB tool is complaining about-the clipping.

                          Ideally, you want a source that will produce a tone that down a bit from "white" such as the second to brightest swatch on a ColorChecker card (which is what Thomas Knoll uses when calibrating Camera Raw).
                          • 10. Re: ACR 4: White Balance readings off of a target
                            Level 1
                            "Actually, yes, that would be proper...but you don't want to set white balance using a greycard which is actually too far down the scale. You want to set white balance with a non-specular "white" that is not clipped. That's what the WB tool is complaining about-the clipping."

                            ACR complains even when the white is *not* clipped, as best I can tell. I have no problem setting the WB with the same image using RAW Developer. From within ACR, a change of 0.10 exposure units brings the white to 254, but the tool will not WB on this white (regardless of the exposure adjustment). We're not talking "blown out."

                            If it's true that one or more of the channels are clipped (or nearly so) when the WB target hits 255, then it is a good exposure check to prevent clipping.

                            I have the ColorChecker card, and I've used it with Thom Fors' calibration script, but unfortunately it is not very field-worthy. The white of that card is also not neutral (avg. B* value of 2.3 from many cards), so using the gray for color balance makes sense. The BabelColor White Target is plastic, waterproof, spectrally uniform, it has a matte finish, and it is small and easy to use in the field. The *a, *b components are both less than one. It simply won't work well with ACR if the white hits (or nearly hits) 255. But that may be a sign that the exposure is "to the right" as far as one should go.

                            Thanks for your comments.

                            --Rich
                            • 11. Re: ACR 4: White Balance readings off of a target
                              Level 1
                              "The *a, *b components are both less than one."

                              Sorry, I'm dlysexic and I can't type. The a*, b* components are both less than 1.
                              • 12. Re: ACR 4: White Balance readings off of a target
                                Level 1
                                "From within ACR, a change of 0.10 exposure units brings the white to 254, but the tool will not WB on this white (regardless of the exposure adjustment). We're not talking "blown out." "

                                You realize that changing the exposure in CR is _NOT_ the same as compensating for exposure in the camera? If the target is too bright, nothing you do to alter the over exposed "white" will change the actual sensor data-which is what Camera Raw is set up to use. You need a nonclipped camera exposure to set WB.
                                • 13. Re: ACR 4: White Balance readings off of a target
                                  Level 1
                                  Yes, I do realize that, although I would re-phrase your conclusion to "You need a non-clipped area of the image to white balance on." You can still white balance if part of the image is clipped - you just can't WB on that part of the image.

                                  As I stated near the beginning of this thread, I usually set my camera compensation to -1/3. There are specular highlights in the image that "blow out" before the target (appears to be the red channel). The peak of the target is not at the edge of the histogram - although it's close. RAW Developer has no problem doing anaccurate WB on the same image. With the ACR exposure slider at 0, the white target is at 255. Bringing the exposure slider down does NOT change the exposure of the image (obviously), but the RGB numbers immediately begin to drop in the target (and the red channel of the highlight lags slightly). Leaving the exposure at zero, one can eliminate the clipped highlights with 20 of "recovery." This image is not drastically overexposed - it is near perfectly exposed, yet what I'm hearing is that this apparently indicates that the sensor is beginning to clip. If the CAMERA exposure is decreased by 1/3 stop, all works well. My take on this is that my exposure is "to the right" as far as it can go - and perhaps too far without camera compensation of -1/3.

                                  --Rich
                                  • 14. Re: ACR 4: White Balance readings off of a target
                                    Bill_Janes Level 2
                                    >Bringing the exposure slider down does NOT change the exposure of the image (obviously), but the RGB numbers immediately begin to drop in the target (and the red channel of the highlight lags slightly). Leaving the exposure at zero, one can eliminate the clipped highlights with 20 of "recovery." This image is not drastically overexposed - it is near perfectly exposed, yet what I'm hearing is that this apparently indicates that the sensor is beginning to clip. If the CAMERA exposure is decreased by 1/3 stop, all works well. My take on this is that my exposure is "to the right" as far as it can go - and perhaps too far without camera compensation of -1/3

                                    If you are exposing in daylight, the red channel in the raw file would not be clipped, since a multiplier of about 1.8 is applied to the red channel to achieve white balance and a multiplier of about 1.4 is used for the blue channel. The green channel is used as is. Here is the raw file histogram of the preferred square of the Color Checker exposed under daylight with a Nikon camera. The red and blue channels are relatively underexposed and the picture has a greenish cast. The WB multipliers for Canons are similar, but there is some variation among cameras.



                                    The red channel would not be clipped in the raw file, but clipping would occur when white balance is applied. In this case, use of the exposure control in ACR to prevent clipping of the red channel would be OK.

                                    For exposing to the right in all channels, some photographers use a magenta filter over the lens to hold back the green light to which the sensor is most sensitive. If you want your RGB histograms to display the status of the raw channels, you can upload unity (1.0) WB multipliers to the camera as a custom WB setting. However, this messes up ACR, which thinks that it is dealing with a multiple exposure. To avoid this problem you can use multiplers near 1.0 (such as 0.98).
                                    • 15. Re: ACR 4: White Balance readings off of a target
                                      Level 1
                                      Bill,

                                      Your post is interesting and thought-provoking to me.

                                      How is it possible to see the "raw" channels, to see what is/may be clipped at the sensor level? I don't have access to the raw numbers or an unmodified histogram. The image you show is a .psd - thus it's been converted. I have no way to "turn off" the processing in ACR or view the linear-gamma histogram.

                                      The "highlight" that I made reference to is not a gray patch - it's coming off of vegetation, so I'm not certain that you can conclude which channel will clip based on your gray scale patch and the correction factors that are applied to each channel.

                                      If I put a color sampler over the highlight, and decrease the exposure by 0.15, with the color temp manually set at Daylight, I get values of R:255, G:253, B:219 (in ProPhoto). Under the same conditions, a sampler on the White Disk gives 252,252,252, down from 255, 255, 255. So what is clipped in the highlight? Converting to a TIF gives the same result - the white disk at 252, and the red of the highlight at 255. Sure looks like red to me. If I select the area around the highlight in the converted TIF, and look at the histogram of that, the red channel is clearly clipped, and the green and blue are not.

                                      But this is moving away from the main problem - the inability to set WB in ACR off of the White Disk if the exposure is right-on (i.e., a spectrally neutral diffuser in full light giving RGB values of 255) - there is simply no lattitude. As I mentioned, I can do this ok with RAW Developer, but the values, of course, are not transferrable, and I do prefer ACR for the bulk of my work. If the white disk is under 255 (e.g., exposure compensation of -1/3), the white balance can be set ok. Most of the time this works well for me, but I shall continue to investigate.

                                      Thanks for your comments and for taking the time to post the image.

                                      --Rich
                                      • 16. Re: ACR 4: White Balance readings off of a target
                                        Level 1
                                        "If you want your RGB histograms to display the status of the raw channels, you can upload unity (1.0) WB multipliers to the camera as a custom WB setting. However, this messes up ACR, which thinks that it is dealing with a multiple exposure. To avoid this problem you can use multiplers near 1.0 (such as 0.98). "

                                        This is interesting. I'll look into this. Perhaps the red "clipping" I'm seeing is because of the WB multiplication. It would be nice if ACR could show the raw channels.

                                        It is certainly a disadvantage to be at the output end of the black box, and to not be able to see what happens every step of the way inside the box.

                                        Thanks again.
                                        --Rich
                                        • 17. Re: ACR 4: White Balance readings off of a target
                                          Bill_Janes Level 2
                                          >How is it possible to see the "raw" channels, to see what is/may be clipped at the sensor level? I don't have access to the raw numbers or an unmodified histogram. The image you show is a .psd - thus it's been converted. I have no way to "turn off" the processing in ACR or view the linear-gamma histogram.

                                          People have asked for a linear raw conversion with no white balance in ACR before, but thus far it is not supported. The two programs that I have used are DCRaw (available for both the Mac and Windows) and Iris (a powerful program directed to astronomers).

                                          >The "highlight" that I made reference to is not a gray patch - it's coming off of vegetation, so I'm not certain that you can conclude which channel will clip based on your gray scale patch and the correction factors that are applied to each channel.

                                          That's right. You should look at the linear raw file with no white convertion.

                                          >If I put a color sampler over the highlight, and decrease the exposure by 0.15, with the color temp manually set at Daylight, I get values of R:255, G:253, B:219 (in ProPhoto). Under the same conditions, a sampler on the White Disk gives 252,252,252, down from 255, 255, 255. So what is clipped in the highlight? Converting to a TIF gives the same result - the white disk at 252, and the red of the highlight at 255. Sure looks like red to me. If I select the area around the highlight in the converted TIF, and look at the histogram of that, the red channel is clearly clipped, and the green and blue are not.

                                          >But this is moving away from the main problem - the inability to set WB in ACR off of the White Disk if the exposure is right-on (i.e., a spectrally neutral diffuser in full light giving RGB values of 255) - there is simply no lattitude. As I mentioned, I can do this ok with RAW Developer, but the values, of course, are not transferrable, and I do prefer ACR for the bulk of my work. If the white disk is under 255 (e.g., exposure compensation of -1/3), the white balance can be set ok. Most of the time this works well for me, but I shall continue to investigate.

                                          If you include a neutral card in the picture to set white balance, I do not think that white is a good choice. For reasons Jeff Schewe explained, a light gray card would be a better choice. The WhiBal works well, but there are others. Even the Kodak 18% gray card can be used, but it may not be spectrally neutral and is a bit too far down the gray scale, but it can give good results for non-critical work.
                                          • 18. Re: ACR 4: White Balance readings off of a target
                                            Panoholic Level 2
                                            Richard,

                                            you need to realize, that you *never* see the original values gained from the pixel sites. *Some* WB is always applied; it starts out with "As shot". Plus, a "green pixel" does not yield a "green value", etc.

                                            If you send me such an image with co-ordinates of your WB objects, I can take a look at the *true pixel values* (Sch-T1@telus.net).

                                            Note, that using a gray card may not even come close to gaining identical or like values of the pixels with the different filters ("red pixel", "green pixel", "blue pixel").

                                            When I shoot a gray card with the Canon 20D in sunny daylight, the green values are about 70% higher, than the red values - but you would want to see quasy-identical values.
                                            • 19. Re: ACR 4: White Balance readings off of a target
                                              Level 1
                                              <you need to realize, that you *never* see the original values gained from the pixel sites. *Some* WB is always applied; it starts out with "As shot". ><br /><br />Well, back in my days in biophysics I could look at the raw data off the A/D converter...  you're correct - now, I can't.  You're also correct in that I cannot see the image in ACR w/o WB correction, unless I resort to the trick mentioned earlier.<br /><br /><If you send me such an image with co-ordinates of your WB objects, I can take a look at the *true pixel values* (Sch-T1@telus.net).><br /><br />See attached.  I'm curious how you do this.<br /><br /><Note, that using a gray card may not even come close to gaining identical or like values of the pixels with the different filters ("red pixel", "green pixel", "blue pixel").><br /><br />I assume you mean "pre-white balance."  Understood.  As long as the image data does not clip (specular highlights, who cares) and I can WB correctly, I could care less what the actual values are.  What I really want to know is when a channel clips.<br /><br /><When I shoot a gray card with the Canon 20D in sunny daylight, the green values are about 70% higher, than the red values - but you would want to see quasy-identical values.><br /><br />Thanks,<br /><br />--Rich
                                              • 20. Re: ACR 4: White Balance readings off of a target
                                                Panoholic Level 2
                                                Richard,

                                                I just added a half sentence to my post:

                                                "Plus, a "green pixel" does not yield a "green value", etc."

                                                and then I saw, that you replied in the meantime. Darn.

                                                [i]I assume you mean "pre-white balance[/i]

                                                I mean much more: [b]pre-demosaicing[/b]. A "green pixel" does not "accept" only green light rays (which is not a frequence but a range of frequences anyway), but all other light rays, to lesser degree. The same applies to the other color-filtered pixels.

                                                Consequently, he value of the red channel of a pixel comes from "red pixels", "green pixels" and "blue pixels".

                                                The effect is, that you [i]really[/i] can't rely on the displayed channel values in relation to the original pixel site values. This is a non-issue normally (and for most people always), but when the subject is [i]effective overexposure[/i], then this can't be ignored.

                                                [i]I'm curious how you do this[/i]

                                                As I am unsatisfied with what I get to see, I program it for myself. I could give you the program, however at the moment it works only with uncompressed DNG (i.e. if you send me an image, I will convert it in DNG). I am just working on the decompression, and processing the original raw files will come later (the manufacturers are keeping the structure of their data secret, and those, who decoded that already - like Adobe - keep the secret).

                                                DARN.
                                                I just received the message, that your file was 17021718 bytes long, and the limit of my mailbox is 15728640. What the hell is that long? Goodness, are you shooting with a satellite spy camera? :-)

                                                Anyway, see my related email.
                                                • 21. Re: ACR 4: White Balance readings off of a target
                                                  Level 1
                                                  Dear G,

                                                  I understand that interpolation occurs in de-moasiacing the image, and that this can affect effective exposure.

                                                  I sent 2 (then one) uncompressed DNG's made from Nikon D2X NEFs. Or were they compressed? I'll check and post them to my web site where you can d'load them.

                                                  I'm envious - I no longer have time for any serious programming.

                                                  Best,

                                                  --Rich
                                                  • 22. Re: ACR 4: White Balance readings off of a target
                                                    Level 1
                                                    Dear List,

                                                    I have prepared 3 DNGs from a Nikon D2x, shot at the same time with differing exposures of 1/3 stop. There is one image that is + 1/3, and another that is -1/3 compared with the ideal exposure. There are 3 white balancing targets in each shot: The small Gretag ColorChecker, the White Disk, and a 4x6 Digital Gray Card (DGC-100). I do not have a WhiBal card yet. The light is afternoon ambient daylight in Tucson.

                                                    You will see that the 3 targets are all capable of giving useful information needed to set the white balance. There is minor inconsistency between the various ColorChecker gray patch values and white, but all values measured in ACR and by the camera itself ranged between 4950 and 5100. Of all measurements, both in ACR and by the camera sensor, by far the largest variation is in the tint, which ranged from -1 (camera) to 10 (ColorChecker Gray3).

                                                    White balance measurement could always be made with the Digital Gray Card, regardless of the exposure. White balance could be made with the ColorChecker card even when the image was overexposed one-third stop, by using one of the gray patches. White balance cannot be made with the White Target if the image is overexposed by one-third stop.

                                                    When shooting outdoors, the White Disk gives valuable feedback on overexposure, in that the clipping will show up on the disk, which is easy to visualize in the viewfinder, even in bright daylight. Obviously the Digital Gray Card gives no feedback on exposure, and the white patch in the ColorChecker is not a reliable indicator of correct exposure.

                                                    In viewing the images, it is also easy to see that the black plastic case and velcro strap of the White Target gives valuable information on blacks. The ColorChecker black patch never gives a "true black."

                                                    I am not aware of any theoretical reason why adjusting the white balance based on a white patch should be inferior to using a gray patch, provided that the white patch data is not clipped. In my experience, both give comparable results.

                                                    Regards,

                                                    --Rich Wagner

                                                    (I hope I get the HTML correct for this Forum.)



                                                    Download DNGs
                                                    • 23. Re: ACR 4: White Balance readings off of a target
                                                      Level 1
                                                      " am not aware of any theoretical reason why adjusting the white balance based on a white patch should be inferior to using a gray patch, provided that the white patch data is not clipped. "

                                                      And what Thomas has stated is that it is the real image density of the raw exposure that determines whether an image has clipped whites points and therfore is suitable or unsuitable for using as a WB target. So, you can test all you want, but it's a battle you will loose...

                                                      Thomas uses the #1 grey patch as the basis for testing of Camera Raw. If you want to use something else, fine...just make sure NOTHING is clipped in the real exposure cause that don't work. Maybe somebody SMART might take a patch of grey paint (spectrally neutral) and paint a swatch on the White Disk so it WOULD be useful for white balance. But so far, your argument ain't getting a lot of traction, ya know?
                                                      • 24. Re: ACR 4: White Balance readings off of a target
                                                        Level 1
                                                        Sorry, Jeff, but I don't follow your reasoning. "just make sure NOTHING is clipped in the real exposure cause that don't work." Huh? If something is clipped and you're not using it to white balance, so what? For all intents and purposes, you could crop it out. Half the image could be completely blown out, but if that's not where the target is, and if the target is not clipped, the target will work fine.

                                                        " So, you can test all you want, but it's a battle you will loose... "

                                                        Lose based on what? If the White Disk has clipping in it, ACR rightfully will not allow you to white balance on it. If it is not clipped, it works, and it works well. Where's the controversy?

                                                        "it is the real image density of the raw exposure that determines whether an image has clipped whites points and therfore is suitable or unsuitable for using as a WB target."

                                                        Sorry, I don't understand what the phrase "real image density of the raw exposure" means. Can you explain?

                                                        Re: "your argument ain't getting a lot of traction, ya know? "

                                                        My only "argument" is that you can white balance successfully off of a White Target (as long as it is not clipped), and that additionally doing so gives you exposure information that you often otherwise don't have. (If it's clipped, you're overexposing.) It works fine for me (and others). If it doesn't work for you, and you've tried it, I'm curious why not, but it's not a big deal. I have a feeling we're talking past each other.

                                                        I still haven't heard why a white balance target should not be as good a target as a gray balance target if the white balance target is not clipped. X-rite actually sells a ColorChecker White Balance card made specifically for digital photography that is, you guessed it - white. Are they completely off-base?

                                                        --Rich
                                                        • 25. Re: ACR 4: White Balance readings off of a target
                                                          Ramón G Castañeda Level 4
                                                          From the X-rite link above:

                                                          ColorChecker White Balance

                                                          >The ColorChecker White Balance card is a full size version of the white reference square used in the standard 24-patch ColorChecker.

                                                          Isn't the "white" reference square used in the standard 24-patch ColorChecker a very light gray (241,241,241)? :)

                                                          c
                                                          • 26. Re: ACR 4: White Balance readings off of a target
                                                            Level 1
                                                            "If the White Disk has clipping in it, ACR rightfully will not allow you to white balance on it. If it is not clipped, it works, and it works well. Where's the controversy?"

                                                            You sure seem helbent on using your White Disk as a target for the white balance in Camera Raw...as you've found out, that's a bit less than optimal if your sensor exposure will likely produce a clipped channel since Camera Raw won't let you. What Thomas has stated and what I know to be a fact, is that the lightest grey of the CC is what the white balance tool is designed to use. It's light enough that it's not too far down the scale but dark enough to not worry about clipping. So, the ideal is to use the same patch that the white balance tool in Camera Raw was designed around.

                                                            So, what is it about your White Disk that has you so fixated on using WHITE (at the risk of clipping)? Yes, you can use it as long as it doesn't clip...is it optimal? No.
                                                            • 27. Re: ACR 4: White Balance readings off of a target
                                                              Level 1
                                                              "Isn't the "white" reference square used in the standard 24-patch ColorChecker a very light gray (241,241,241)? "

                                                              I'm not sure what color space your RGB numbers are from.

                                                              ColorChecker White
                                                              L:92.86 a*: -0.55 b*: 1.81

                                                              ColorChecker Gray1
                                                              L*: 81.65 a*: -1.00 b*: -0.04

                                                              ColorChecker Gray2
                                                              L*: 65.8 a*: -0.02 b*: -0.01

                                                              ColorChecker Gray3
                                                              L*: 50.13 a*: -1.27 b*: -0.63

                                                              White Disk
                                                              L: 96.28 a*: -0.17 b*: -0.37

                                                              Obviously, Gray 2 is the most neutral of the targets, at least of the targets I own and can measure. The White Disk is not far behind, and the difference when converted to RGB in ProPhoto is insignificant.
                                                              • 28. Re: ACR 4: White Balance readings off of a target
                                                                Ramón G Castañeda Level 4
                                                                > not sure what color space your RGB numbers are from.

                                                                Bruce Fraser. CLICK HERE
                                                                c
                                                                c Figure 1: The Color Checker, downloaded from the Web site and converted from Lab to ProPhoto RGB, with RGB values added as type layers.
                                                                • 29. Re: ACR 4: White Balance readings off of a target
                                                                  Level 1
                                                                  Ok...

                                                                  "You sure seem helbent on using your White Disk as a target for the white balance in Camera Raw...as you've found out, that's a bit less than optimal if your sensor exposure will likely produce a clipped channel since Camera Raw won't let you."

                                                                  Well, when I'm crawling around in the mud in Costa Rica, Ecuador, or Peru, soaked, the ColorChecker card doesn't fare well. I've gone through a number of them, and in *my* field use, they suck. Period. In a studio, they're fine. I still own several. I just measured one for the numbers I gave above. I'm helbent on having something that is reliable in my use. I could care less what it's called or who makes it. On the other hand, you seem insistant that the White Disk is inferior, and I'm trying to figure out why (other than the fact that Mr. Knoll used the ColorChecker card when designing ACR).

                                                                  "What Thomas has stated and what I know to be a fact, is that the lightest grey of the CC is what the white balance tool is designed to use. It's light enough that it's not too far down the scale but dark enough to not worry about clipping. So, the ideal is to use the same patch that the white balance tool in Camera Raw was designed around."

                                                                  Really? On my card, the lightest gray on the ColorChecker is not the most neutral. The second gray is. You can see the numbers above. Are you implying that Camera Raw will be more accurate making a WB measurement off of a patch that is not neutral, compared to one that is?

                                                                  "So, what is it about your White Disk that has you so fixated on using WHITE (at the risk of clipping)? Yes, you can use it as long as it doesn't clip...is it optimal? No."

                                                                  Argghhh... If I shoot a test shot and the White Disk clips, then guess what? I'm overexposing! Time to adjust the exposure before going on. This is easy to see on the White Disk because I get zebra stripes. I'f I'm shooting a gray card, I don't get that feedback and I may later discover that important stuff was clipped, and it's too late! Sure, I can color balance just fine, but important highlights are gone. Which is most important? I can always eyeball color temperature - it's subjective, anyway. Or I can measure off something neutral in the shot. Clipped channels? Data lost is data lost...

                                                                  Perhaps our shooting conditions are very dissimilar. I have found the White Disk to be a useful tool, just like many others that I have. It works (against all expectations?) and it works well. If I overexpose a neutral white target, I am often overexposing something else of interest. It becomes a red flag when checking exposure, since I often cannot even see the histogram on the back of the camera. I am certainly NOT recommending that everyone switch to using a White Disk. I have just been surprised at the response of those who feel that for whatever reason it cannot work as well as gray. I see no theoretical reason why not - still. Show me the math, or a give me clear-cut reason. My understanding (and experience) says otherwise. Or to put it another way, if it didn't work, I wouldn't use it, and neither would anyone else. If someone frequently overexposes their shots and relies on highlight recovery, it's probably not a good tool for them (unless it helps them correct their exposure).

                                                                  --Rich
                                                                  • 30. Re: ACR 4: White Balance readings off of a target
                                                                    Level 1
                                                                    Ramon,

                                                                    OK, they are in ProPhoto. If you want to call (241,241, 241) gray, then fine.. I think most would call it white.

                                                                    The chart that Bruce presented contains idealized reference numbers. I've measured a lot of cards, and none have had a perfectly neutral gray scale. Each card needs to be individually measured to obtain accurate reference values.

                                                                    See:
                                                                    http://www.rmimaging.com/information/colorchecker.html

                                                                    --Rich
                                                                    • 31. Re: ACR 4: White Balance readings off of a target
                                                                      Panoholic Level 2
                                                                      It's only for the understanding of the subject, without any practical relevance, but I take exception with the statement

                                                                      i the lightest grey of the CC is what the white balance tool is designed to use

                                                                      IMO there is no such thing. The white balance tool must work (and does work) with
                                                                      b anything
                                                                      what is not clipped.

                                                                      If I pick a red tulip, then that has to become white. It has no relevance if I pick white or gray and how dark gray.

                                                                      If I am wrong, then I would be thankful to be corrected with an accompanying explanation.
                                                                      • 32. Re: ACR 4: White Balance readings off of a target
                                                                        Level 1
                                                                        "If I am wrong, then I would be thankful to be corrected with an accompanying explanation. "

                                                                        Thomas Knoll, who wrote Camera Raw (was also co-author of a little thing called Photoshop) uses the ColorChecker card's lightest grey as the basis for creating camera profiles for white balance used in Camera Raw...so, if one where to opt to select the ideal object to use as the basis of setting white balance, I would expect that the lightest grey CC swatch would be ideal...wouldn't you say?

                                                                        Can you use something else? Certainly (although I would suggest that trying white balance off of a red tulip would give you somewhat less that optimal results).

                                                                        The phrase that Thomas Knoll and Bruce Fraser (who wrote Real World Camera Raw) have used is to "use a non-specular, textural neutral white or light grey as the basis for adjusting white balance". That describes the lite grey patch pretty well.

                                                                        As to why a middle grey (or darker) is less good? Well, because in a linear file, middle grey is WAY down the levels scale. Remember, in a linear capture, a "middle grey" will actually be in a much darker tone level until the image is gamma encoded. The results from white balancing from a middle grey may get biased based upon a shadow tint of the camera...(one of the reasons the Calibrate function has a Shadow Tint control).

                                                                        So, the bottom line? Use whatever works for the image you are trying to white balance...but it's optimal to use a "a non-specular, textural neutral white or light grey as the basis for adjusting white balance".
                                                                        • 33. Re: ACR 4: White Balance readings off of a target
                                                                          Panoholic Level 2
                                                                          >so, if one where to opt to select the ideal object to use as the basis of setting white balance, I would expect that the lightest grey CC swatch would be ideal...wouldn't you say?

                                                                          No, I would not. With all due respect, the fact that Thomas Knoll has used something is not an explanation for me. Perhaps he had some good reason to do so, and the explanation of that reason might be convincing. Perhaps there is no specific reason for that particular pick (this is my guess).

                                                                          >"use a non-specular, textural neutral white or light grey as the basis for adjusting white balance". That describes the lite grey patch pretty well.

                                                                          That describes everything from dark gray to 100% white.

                                                                          >because in a linear file, middle grey is WAY down the levels scale

                                                                          This is a serious mistake. The "location" of something on the level scale depends FIRSTLY on the lighting, SECONDLY on the exposure settings of the camera, and only THIRDLY on the texture/reflection grade of the object.
                                                                          • 34. Re: ACR 4: White Balance readings off of a target
                                                                            Level 1
                                                                            "With all due respect, the fact that Thomas Knoll has used something is not an explanation for me. Perhaps he had some good reason to do so, and the explanation of that reason might be convincing."

                                                                            If Thomas feels so inclined to answer the why, that's up to him. But for me, using the same tools he's using for the purpose of creating the white balance profile-he shoots the CC at D65 and Standard Illuminate A (2856K) and uses the lightest grey patch of the CC to determine the "Daylight" and Tungsten" white balance scales on a camera by camera basis.

                                                                            So, if I'm gonna spend the time to do a critical white balance, using the same swatch as he does to make the WB profile is certainly not stupid, right? In fact, I suspect it's optimal.

                                                                            "Perhaps there is no specific reason for that particular pick (this is my guess). "

                                                                            Uh no...he really doesn't work that way. I suspect that he uses the lightest grey because; 1) he has a known reference and 2) that reference is particularly suited to achieve his goal of determining the spectral responce of a camera at D65 and Illuminate A and 3) it's successful at accomplishing what he's trying to do...figure out how to set the white balance of a camera.

                                                                            "This is a serious mistake. The "location" of something on the level scale depends FIRSTLY on the lighting, SECONDLY on the exposure settings of the camera, and only THIRDLY on the texture/reflection grade of the object. "

                                                                            No...if you shoot raw, and the object you are shooting is say, an 18% grey card, it may have a 50% (middle grey) refelectance, but in a properly exposed raw file that tone it will be way down the luminance scale in linear gamma. Remember, what we are trying to achieve is a white balance, not a grey balance and the image is in linear gamma (1:1) which means we have a LOT of data in the highlights and very little data (by comparisson) in the shadows. So, using a darker grey is less useful right off the bat cause it's got less data in the capture. That really is why it's called a "White Balance" and not grey balance or cast removal...if you try to WB on a darker grey sample, it will be less accurate.
                                                                            • 35. Re: ACR 4: White Balance readings off of a target
                                                                              Panoholic Level 2
                                                                              >an 18% grey card, it may have a 50% (middle grey) refelectance

                                                                              >what we are trying to achieve is a white balance, not a grey balance

                                                                              I don't think there is any point to continue this discussion.
                                                                              • 36. Re: ACR 4: White Balance readings off of a target
                                                                                Ramón G Castañeda Level 4
                                                                                >I don't think there is any point to continue this discussion.

                                                                                Because?

                                                                                Because you agree or disagree with either position? If so, which one? Perhaps both? None?

                                                                                Because you don't like it?

                                                                                Not trying to be difficult, just that I'm genuinely puzzled by your cryptic remark.
                                                                                • 37. Re: ACR 4: White Balance readings off of a target
                                                                                  Level 1
                                                                                  Cause he thinks I'm stupid because I left the word reflectance in in my edit...didn't notice till after the period of time allowed for edits...

                                                                                  An 18% grey card reflects 18% of the light falling on it to product a middle grey...which still doesn't alter that fact that a middle grey, in a linear caputre is to dark to use as a basis for optimal white balance setting...

                                                                                  :~)
                                                                                  • 38. Re: ACR 4: White Balance readings off of a target
                                                                                    Panoholic Level 2
                                                                                    >what we are trying to achieve is a white balance, not a grey balance

                                                                                    Well, then we must not use any gray card, Rich's white disk is the only suitable subject for WB.

                                                                                    On the other hand, if we accept the banality

                                                                                    i white is a very light gray, and gray is a shade of white

                                                                                    then we come closer to the reality.

                                                                                    Really, what is white (in terms of reflectance)? It appears easy: a surface is white, if it reflects all or nearly all visible light.

                                                                                    Yeah, but what about diffuse reflection, exhibited by the vast majority of objects, among others the color charts and the gray cards? Only a portion of the reflected light can be captured by the camera; in other words, such surfaces appear gray.

                                                                                    So, not only a gray card, but a white card too is suitable only for achieving gray balance, but not white balance, right?

                                                                                    Luckily, it's not so bad, as white balancing has nothing to do with white; all it has to do with is the proportion of the light rays with different wavelengths. If a gray surface appears gray, then a white surface will (or is supposed to) appear white as well, and vice versa.

                                                                                    It boils down to the question: which luminosity of the surface is suitable for white balancing?

                                                                                    Of course it has to be unclipped and is noiseless; but otherwise the luminosity should not matter - or the raw processor has to be redesigned.
                                                                                    • 39. Re: ACR 4: White Balance readings off of a target
                                                                                      Level 1
                                                                                      "Of course it has to be unclipped and is noiseless; but otherwise the luminosity should not matter - or the raw processor has to be redesigned. "

                                                                                      Uh huh...but do you agree that a target, whose luminosity in a given exposure is too far down the linear scale, is less optimal?
                                                                                      1 2 Previous Next