31 Replies Latest reply on Apr 12, 2011 8:29 AM by ssprengel

    ColorChecker Passport-Color Profiles

    gmurillofdzp

      Hi there,

       

      I am photographing paintings with Elinchrom flashes with an even light and I use the ColorChecker Passport from X-Rite to keep a reference in order to match the colors.

       

      Till now, I've been using the software DNG profile manager, provided with the ColorChecker Passport from X-Rite, which creates a camera profile that should be helpfull to get the right color. Once I've applied the new camera profile (after restarting lightroom), I do the white balance. It's recommended on the X-Rite tutorial to use the target number 20 called "neutral 8" to do the white balance. I've also tried the target 22 called "neutral 5", and it seems to get slightly better results.

       

      I've noticed that even creating a color profile and doing the white balance the colors don't match. I am using a mac laptop screen that I calibrate quite often, which is not very helpful for obvious reasons, but that helps to keep track of the general look of the image, so that's why I keep the color adjustment to the RGB numbers.

       

      I've downloaded the following PDF; here there is some information about the RGB values of various ColorChekers:

       

      http://www.google.com/url?sa=t&source=web&cd=2&ved=0CB4QFjAB&url=http%3A%2F%2Fwww.babelcol or.com%2Fdownload%2FRGB%2520Coordinates%2520of%2520the%2520Macbeth%2520ColorChecker.pdf&rc t=j&q=ColorChecker%20rgb%20values&ei=wOCdTbG7NIyDhQelj5G8BA&usg=AFQjCNGj3wUpWC_uMxgMCeR3GI yN7kDU0A&sig2=mPf1grSAULxyMs_SanYVNw&cad=rja

       

      Using these values and a selective color layer in photoshop I get close to the real colors, although it is not good enough.

       

      My questions are:

       

      1-Does anybody know the ColorChecker Passport RGB (Adobe 98) values?

      2-I've tried to use the DNG profile editor because it seems a precise tool for this, but it doesn't recognise the 5D Mark II files. Is it because this application hasn't been updated? It also seems to work only with 6500K or 2850K, which is not the case of the Elincrhom 600 monoblocks I am using (which I think are around 5000K.

      3-Could anybody recommend a workflow that's better to match the colors?

       

      I have more questions, but I think these three are a good start.

       

      Thanks!

        • 1. Re: ColorChecker Passport-Color Profiles
          ssprengel Adobe Community Professional & MVP

          1 - AdobeRGB numbers are available in a spreadsheet on the BabelColor website:

          http://www.babelcolor.com/main_level/ColorChecker.htm#spreadsheet

           

          2 - The two colors of light you're talking about (6500K, 2850K) only apply to a dual-illuminant profile.  If you are creating a single-illuminant profile it's ok to use colors other than those two.  When you say your 5D.2 files don’t work with the 3.5 year old DNG Profile Editor, are you converting to DNGs?  If so, maybe you need to convert to an older ACR-compatibility, but I wouldn't do this if you don't need to.

           

          3 - When you say the colors don't match, what are you looking at, the colorchecker reference image, or a painting, and are you comparing with your eyes or with RGB numbers in Photoshop?  I would think eyecdroppering colors in Photoshop are going to be a problem since the contrast and toning curve could change the numbers quite a bit.  Do you have a way to normalize your exposure across the gray patches before eyedroppering the other colors?

           

          Can you post a RAW or DNG color-checker image taking with your Elinchrom lighting?  If you don't have your own file-storage website, use something like www.yousendit.com send the download link to your own e-mail and post that link, here.  I'd like to compute the error of the profile compared to published CC values.

          • 2. Re: ColorChecker Passport-Color Profiles
            Tim Lookingbill Level 1

            I never got really accurate results using a dual illuminant DNG profile. What this type of profile does is average the amount of yellow and luminance in reds and greens between tungsten and midday overcast lighting which is what 6500K looks like. It's for those who want a one size fits all approach to get pleasing color under both types of light. I'ld suggest using the single 6500K illuminant profile.

             

            Now as for getting dead on accurate colors compared to what you see on your display and the scene colors you're matching to, you're going to have to do some HSL and color temp tweaking. There's not one light source I've come across including direct sunlight that will render all colors from a wide range of objects accurately.

             

            You're going to have to compensate for the interplay of spectral reflectance characteristics inherent in the light and what it does to some oil, watercolor and acrylic pigments and suspension medium and how the camera records this interplay. See the differences in the amount of green and red in the image of the strawberry under two different light sources below. Both are dead to nuts accurate to how I see them under their respective light source.

             

            Not familiar with Elinchrome flash. The flash on my $500 Pentax K100D DSLR makes colors look pretty accurate except in some reds and oranges which can be fixed using the HSL. Since the Elinchrome flash isn't a constant light source to judge whether color is accurate to the display, I'ld suggest you view the painting for comparison purposes under the light source the print will be viewed under like a Solux or museum grade halogen. Fluorescent lights like the Alzo demonstrated below is not an ideal constant light source for exact reproduction yellows, greens and reds will look drastically different, but still pleasing.

             

            Don't know what kind of painting you're reproducing and what substrate and printer used, but you're going to have to contend with the painting's own substrate color like canvas or illustration board in whether that color becomes the color of the printer substrate or you'll have to white out to 255RGB in the image where you want the print substrate to show.

             

            This usually isn't a big concern, but some substrates do have optical brighteners that change hue under different lights the print is viewed under AND change hue in the original capture according to the spectral reflectance of the substrate of the original painting.

             

            StrawberrySpectralDif.jpg

            • 3. Re: ColorChecker Passport-Color Profiles
              gmurillofdzp Level 1

              1-Thanks for the link to the color checker AdobeRGB numbers.
              2-I was using the Canon 5D MarkII CR2, as I wanted to avoid converting to DNG. So I guess the best option will be not using the DNG profile editor then.
              3-When I say the colors don't match I am looking with my eyes both at the colorckecker and the painting in the laptop screen. The RGB numbers in Photoshop doesn't match either. I've used a lightmeter to make an even exposure if that's what you mean by "normalize the exposure"; the pictures have been taken at f8-1/125sec-ISO160, exactly the same reading that I was getting in all the areas of the painting with the lightmeter.

               

              Find below the original CR2 with a ColorChecker in it. I've also included the profile created by the XRite DNG profile manager for this specific file.

               

              https://www.onlinefilefolder.com/2flzEFXCt8Hzmx

               

              I am using this lighting: http://www.elinchrom.com/products.php?cat=2

              • 4. Re: ColorChecker Passport-Color Profiles
                Tim Lookingbill Level 1

                You keep saying the colors on your display don't match the painting even after applying the profile.

                 

                What light source are you using to view the painting? Does the Elinchrome stay constantly on or is it just a flash?

                 

                For accurate color reproduction you REALLY need to understand what a constant light source does to the perception of color compared to a flash exposure.

                • 5. Re: ColorChecker Passport-Color Profiles
                  Jeff Schewe Level 5

                  gmurillofdzp wrote:

                   

                  2-I've tried to use the DNG profile editor because it seems a precise tool for this, but it doesn't recognise the 5D Mark II files. Is it because this application hasn't been updated? It also seems to work only with 6500K or 2850K, which is not the case of the Elincrhom 600 monoblocks I am using (which I think are around 5000K.

                   

                  DNG Profile Editor can not open native raw files, only DNG files. As indicated in the docs, you'll need to use a DNG Converter that supports your camera (I would suggest using the current shipping DNG Converter 6.3). Note, the step requiring the DNG Conversion is ONLY required to open the images in DNG Profile Editor...for regular processing you don't need to convert to DNG.

                   

                  In terms of a fixed light setup, you don't need to use dual-illuminate profiles. You'll want to profile with the exact lighting you end up using.

                   

                  gmurillofdzp wrote:

                   

                  3-Could anybody recommend a workflow that's better to match the colors?

                   

                  If the DNG Profiles are not giving you accurate color rendering for all colors, you might consider editing the profiles in DNG Profile Editor and tweaking the colors that are off. It's important to not do too many edit points as this will make color rendering less smooth. The least number of edit points the better.

                   

                  In terms of technical analysis of the final color rendering accuracy you should note that a weak point in your system currently is the laptop display. You should think about getting a higher quality display to use as a second display that can be more accurately profiled.

                   

                  Also note that when shooting paintings, paint and pigment may not have an accurate spectral response with a digital camera sensor. The way the eyes see a color is similar but not exactly the same as the way a sensor sees color. While I encourage you to get the capture and raw processing as accurate as possible, if you are trying to nail specific colors, you may need to do that in Photoshop which has more tools for very specific color corrections. It's inevitable that you'll need to tweak a bit after raw processing...

                  • 6. Re: ColorChecker Passport-Color Profiles
                    ssprengel Adobe Community Professional & MVP

                    1 - I think the Excel spreadsheet was one link down from the PDF you posted the link to.  I'd seen the spreadsheet before so I just kept looking until I found it.

                     

                    2 - The DNG Profile Editor only operates on DNGs however the profile thus computed works for the CR2s and DNGs.  It is independent of the container of the original RAW data.  You should have no hesitation converting a copy of your color-checker CR2 to a DNG to try the Adobe-supplied product.  You won't have to convert all your photo CR2s to DNGs if that's what was causing hesitation.

                     

                    3 - Are you able to view the painting under the color of lighting you're photographing it in?  Otherwise you shouldn't expect the colors to match.  By normalizing, I mean adjusting the toning and whatever else in ACR until the RGB numbers measured in Photoshop are correct for all of the gray patches, before you attempt the color patches.  If you don't then you're not going to be seeing anywhere near the same RGB numbers. 

                     

                    I'll take a look at your color-checker and see how much off the colors are from the standard using your profile.  Are you using Photoshop CS5 with ACR 6.3 or 6.4RC or something earlier?

                    • 8. Re: ColorChecker Passport-Color Profiles
                      gmurillofdzp Level 1

                      1-Ok, I'll use the PDF you've provided with the ColorChecker Adobe  RGB values. In this document it recommends to use table 2, which is for  8bit instead of table 3, which is 16 bit because "there is no color  picker that offers 16 bit resolution", as "Photoshop processes 16-bit  file as if 15-bit and resaves the file as 16-bit" (page 4, point 3). Do  you recommend to convert the CR2 to 8 bit in ACR in order to match the  ColorChecker values later on?
                      2-Ok I'll try to convert the CR2 to DNG and creating a profile with the  Adobe DNG Editor and will let you know about the result I get.
                      3-What I was doing to normalize was only doing a white balance in target number 20 (neutral 8).
                      So then first of all I need to match in ACR the gray patches to the values  there are in the PDF you've provided with the ColorChecker Adobe  RGB values?
                      By gray patches I understand you mean targets number 19 (white 9.5), 20  (neutral 8), 21 (neutral 6.5), 22 (neutral 5), 23 (neutral 3.5) and 24  (black 2)?

                      I am using Photoshop CS4 with ACR 5.5.

                      • 9. Re: ColorChecker Passport-Color Profiles
                        Tim Lookingbill Level 1

                        Here's links to a site I recently stumbled across at Luminous Landscape addressing the same issues about accurate color reproduction...

                         

                        http://www.cdiny.com/white_papers.html

                         

                        A spectrally flat constant LED lighting technology...

                         

                        http://www.cdiny.com/LEDproducts.html

                         

                        It may not help you but it's an indicator someone's trying to address this issue. Geffert worked with the Metropolitan Museum of Art for digital color reproduction of paintings. It might lead you to other understandings. Just thought it was a cool site.

                        1 person found this helpful
                        • 10. Re: ColorChecker Passport-Color Profiles
                          gmurillofdzp Level 1

                          Good point. I am comparing the painting using tugsten constant light source, as the Elinchrome is a flash.

                           

                          In  terms of display setup I understand I have limitations both with the  light source I am using to compare the results and the screen I am  using; that's why I am trying to edit everything via the RGB values and  not by eye.

                          • 11. Re: ColorChecker Passport-Color Profiles
                            Tim Lookingbill Level 1

                            You should note about going by the numbers with the Color Checker. Even the Lab numbers can be off as much as 5 points without seeing an extremely noticeable difference, with some colors more noticeable than others. The eyes will take in the overall look of the ColorChecker as looking accurate/normal.

                             

                            I've studied this shooting the Color Checker under a wide range of different lights and can barely see a difference when the numbers change drastically. Watch out mainly for a big change in the green channel assessing RGB numbers which will usually show the biggest shift.

                             

                            Check out how off the CC chart is in the YouTube screenshot below of the eHarmony ad that's been running lately. The people look pretty normal looking even though the CC chart is off the charts.CCchart-eHarmonyYouTubeAd.jpg

                            • 12. Re: ColorChecker Passport-Color Profiles
                              gmurillofdzp Level 1

                              Great article, thanks a lot, I am sure it will be very helpful to get a better understanding of fine art reproduction.


                              • 13. Re: ColorChecker Passport-Color Profiles
                                ssprengel Adobe Community Professional & MVP

                                Comparing a subject illuminated in tungsten verses daylight (your flash is approximately this) is not going to look nearly the same.

                                 

                                This is because taking a picture of things in a reddish light and white-balancing will cause all the reddish colors to be more grayish--red color viewed in red light will look the same as a gray-color viewed in red light, so when you white-balance to the gray color the red color will also look gray. 

                                 

                                Take a picture of the color-checker in each lighting and then white-balance on the same gray-square to compare the differences visually side-by-side in Bridge (or Lightroom).

                                 

                                Before you can think about checking the RGB colors of the colored squares you'll need to adjust the toning to get the RGB values of the neutral squares to be right--the white-point and the black-point the same, and the middle gray squares also the same; otherwise, it's futile.

                                • 14. Re: ColorChecker Passport-Color Profiles
                                  Tim Lookingbill Level 1

                                  You got that right! I couldn't have said it better.

                                   

                                  There are also different flavors of tungsten. I've been experimenting with regular 100 watt GE soft white and 90 watt GE Edison halogen flood bought at Home Depot photographing dark desaturated bluegreen gray fabric and my DSLR sees it as just R=G=B gray. My pop up flash renders it perfectly and the Alzo CFL makes it appear more vibrant, direct sunlight desaturates it but can still make out the hue slightly. Warm colors don't seem to be that affected.

                                  • 15. Re: ColorChecker Passport-Color Profiles
                                    gmurillofdzp Level 1

                                    I totally agree about the comparison with tugsten versus daylight being  futile. That's why I would like to keep the corrections to the RGB  numbers.

                                     

                                    -In the PDF you've provided with the ColorChecker Adobe   RGB values it recommends to use table 2, which is for  8bit instead of  table 3, which is 16 bit because "there is no color  picker that offers 16 bit resolution", as "Photoshop processes 16-bit  file as if 15-bit and resaves the file as 16-bit" (page 4, point 3). Do  you recommend to convert the CR2 to 8 bit in ACR in order to match the  ColorChecker values later on?

                                    -How do you normalize the exposure across the gray patches? Is it matching in ACR the gray patches to the  values  there are in the PDF you've provided with the ColorChecker  Adobe  RGB values? By gray patches I understand you mean targets  number 19 (white 9.5), 20  (neutral 8), 21 (neutral 6.5), 22 (neutral  5), 23 (neutral 3.5) and 24  (black 2)?

                                    • 16. Re: ColorChecker Passport-Color Profiles
                                      Jeff Schewe Level 5

                                      gmurillofdzp wrote:

                                       

                                      -In the PDF you've provided with the ColorChecker Adobe   RGB values it recommends to use table 2, which is for  8bit instead of  table 3, which is 16 bit because "there is no color  picker that offers 16 bit resolution", as "Photoshop processes 16-bit  file as if 15-bit and resaves the file as 16-bit" (page 4, point 3). Do  you recommend to convert the CR2 to 8 bit in ACR in order to match the  ColorChecker values later on?

                                       

                                      No...the reference to 8 bit is simply that you can't use full 16 bit numbers in Photoshop. Photoshop goes 0-32768 for a total of 32769 levels which is what 15 bit plus one level contains. The reason you want to use the 8 bit numbers is simply for the readouts which in ACR is based on 8 bit numbers. You can have 16 bit images in Photoshop also readout in 8 bit numbers. So DON'T go from 16 bit to 8 bit simply for the numbers!

                                      • 17. Re: ColorChecker Passport-Color Profiles
                                        gmurillofdzp Level 1

                                        Ok, I'll keep the image in 16bit and will use table 2 with the 8 bit numbers to match the readouts both in ACR and Photoshop. Thanks for clarifying.

                                        • 18. Re: ColorChecker Passport-Color Profiles
                                          gmurillofdzp Level 1

                                          Ok now more ore less I am matching the ColorChecker numbers to the table  2 (8bit) in the PDF provided by Ssprengel with the ColorChecker Adobe  RGB values.

                                          This is the workflow I've followed:

                                          1st-Create color profile with the DNG Profile Editor and apply it to the CR2 in ACR.
                                          2nd-Do a white balance in target number 20 (neutral 8).
                                          3rd-Get color samples of targets 19 (white 9.5), 20  (neutral 8), 21  (neutral 6.5), 22 (neutral 5), 23 (neutral 3.5) and 24  (black 2) and  match the numbers to the table 2 (8bit) in the PDF provided by Ssprengel with the ColorChecker  Adobe RGB values. Here I have a question, should I match the grays in  ACR or in Photoshop? In case it is ACR, which tools are the most  suitables? I don't manage to change much with HSL. The tool which I find  more useful is the ACR curve, but I am not sure how precise is that.
                                          4th-Once the exposure is normalized across the gray patches, open the  image in Photoshop and match the ColorChecker colors numbers with  selective color.

                                          Do you think there is a better workflow or any weak point here?

                                          • 19. Re: ColorChecker Passport-Color Profiles
                                            ssprengel Adobe Community Professional & MVP

                                            I think I provided a link to an Excel spreadsheet, where in the RGB_8_bit worksheet Table 1 is the Average from 20 CCs, Table 2 has the Color Checker 2005 numbers, and Table 3 has the Color Checker 1976 colors.  The note next to Table 3 that says to use Table 2 I merely because the Table 3 numbers are from an obsolete version of the color-checker.  It is not about using 8-bit vs 16-bit.  As Jeff says you can work with 16-bit files and show 8-bit numbers.

                                             

                                            Use the toning sliders to adjust the neutral patches: exposure, brightness, contrast, black-point, tone-curve, before attempting anything with the color patches.  I would expect the toning curve needs to be set to linear, and then the other four sliders tweaked but I could be wrong.

                                             

                                            To answer the question about gray RGB values in ACR vs Photoshop, for the same image, when you eyedropper patches in ACR do you get the same numbers as when you eyedropper patches in the exported TIF or not?  If they are the same, then using the ones on ACR would be faster because you don't have to export.  If they are not the same, then use the ones in Photoshop.

                                             

                                            The main issue in the workflow is that it'll take forever to get the right values, because adjusting one slider changes multiple patch colors.

                                             

                                            Being able to eyeball things would be a lot faster if you could figure out a way to view and photograph your art in the same continuous daylight-balanced lighting, rather than using flashes.

                                            • 20. Re: ColorChecker Passport-Color Profiles
                                              Tim Lookingbill Level 1

                                              Since you're photographing paintings, I found a site where someone actually photographed hi-rez captures of Denmark artist Peder Kroyer's work. The reason I'm showing you this is to illustrate another aspect of artwork reproduction using a digital camera you may or may not be aware of. It's mainly to do with capturing tertiary=(color hues next to each other on the color wheel) detail inherent in paintings that is limited by the camera's resolution. Perceived color accuracy of the overall painting can be skewed because of a lack of resolution depending on the size of the painting and how close the camera and the amount of resolution available.

                                               

                                              See below the two links of a Kroyer painting taken with a Canon P&S. The first one is the entire painting and the second one is a close up detail section. Compare the detail and color differences in the close up to the full frame version of tertiary colors of cool and warm grays the artist implemented in his brush strokes to add depth to a flat, wet beach surface. If you're planning on enlarging this capture to do full size reproductions, you're going to have to contend with this lack of detail.

                                               

                                              Full frame...

                                               

                                              http://2.bp.blogspot.com/_fvht8G023ow/S4Givdudl9I/AAAAAAAABIc/xzOWnFMiN7c/s1600-h/IMG_0377 .jpg

                                               

                                              Close up detail...

                                               

                                              http://4.bp.blogspot.com/_fvht8G023ow/S4GiwZ9bN8I/AAAAAAAABIk/MWxddE48tls/s1600-h/IMG_0377 b.jpg

                                               

                                              Here's the original blog of the photographer who took the shots which shows other full rez paintings...

                                               

                                              http://timpaints.blogspot.com/2010/02/here-are-some-high-rez-phots-from-my.html

                                              • 21. Re: ColorChecker Passport-Color Profiles
                                                Vit Novak Level 3

                                                Well, you are limied by the camera resolution in most photos, aren't you ?

                                                Excepth portraits for instance, where you want to hide some imperfections on the skin etc

                                                • 22. Re: ColorChecker Passport-Color Profiles
                                                  Tim Lookingbill Level 1

                                                  Well that's a given, Vit.

                                                   

                                                  But when "accurately" reproducing artwork on prints at actual size, expecting a color profile to take care of all the details is asking too much. I mention this from experience working with a gallery owner who wanted to reproduce as exact as possible his client's paintings on a wide format HP inkjet capturing the paintings using a Canon DSLR and processing in Raw. He didn't even get into profiling the camera.

                                                   

                                                  He showed me a side by side comparison between the original and the print and how bad the print looked even though the colors were almost exact. However, the print just looked like it was made of plastic where I saw brush strokes of varying colors on the original painting as shown in the linked close up, the print rendered it as one flat color with barely noticeable brush stroke texture. Depth and vibrance was lost.

                                                   

                                                  The gallery owner and I determined it was from a lack of camera resolution and so he invested in a wide format scanner and pulled the canvas paintings off their frame and scanned them into his computer and processed them in Photoshop. The prints looked much better to the point some artist's of the original preferred the print over their own painted version.

                                                   

                                                  I can tell you for a fact Lab readings weren't even mentioned in our discussions working through this issue. The gallery owner relied mainly on his own eyes in getting the color to match the originals on the prints.

                                                   

                                                  It wasn't an efficient way to work, but he got it done and the clients were quite happy.

                                                  • 23. Re: ColorChecker Passport-Color Profiles
                                                    Vit Novak Level 3

                                                    Well, that's a problem of bayer sensor and interpolation + sharpening + noise reduction needed to get a picture from it. On small details, color is not accurate and they have plastic look because of that. Color profile, whether dng or icc, has nothing to do with lack of details - it just transforms color (RGB)1 to (RGB)2 no matter what ...

                                                     

                                                    Not long ago many people were amazed with 3 Mpix DSLRs, while those photos looked like sh**** to me - sharpening halos were 1mm wide even on 10x15cm print. Today's budget DSLRs are much better, but are still far from medium format film in most aspects, although many people won't agree with me about that ... But anyway, I'm on digital for several years, much more practical and I can live with it ...

                                                     

                                                    Scanners are working different way - they usually have 3 rows of pixels, each row for one of primary color, so there is no interpolation

                                                     

                                                    There were also some (quite expensive) digital cameras working on the same principle, with scanning backs. I don't know if they are still being produced ...

                                                     

                                                    As about photos of painting, if one photo doesn't have enough details, there is an option to take 2x2 or 3x3 or 4x4 photos and merge them, if you have good lens profile to eliminate vigneting and distortion ...

                                                    • 24. Re: ColorChecker Passport-Color Profiles
                                                      Jeff Schewe Level 5

                                                      Tim Lookingbill wrote:

                                                       

                                                      He showed me a side by side comparison between the original and the print and how bad the print looked even though the colors were almost exact. However, the print just looked like it was made of plastic where I saw brush strokes of varying colors on the original painting as shown in the linked close up, the print rendered it as one flat color with barely noticeable brush stroke texture. Depth and vibrance was lost.

                                                       

                                                      That's as much a function of lighting the painting for evenness and texture than anything else. Typical copy lighting wipes out surface detail even though the lighting is very even-actually it's the evenness of a typical setup that kills the texture. I suspect the reason the scanner showed more detail was because of the more directional lighting a flatbed scanner can have. Scanners often over-enhance small textural detail.

                                                       

                                                      There are ways of lighting paintings to not only have even lighting but also maintain a degree of texture. It usually involves deciding the directionality you want and using a mix of hard and soft lighting to cast actual shadows and have a high-lit edge to the texture. A bit of an art form actually...

                                                      • 25. Re: ColorChecker Passport-Color Profiles
                                                        Tim Lookingbill Level 1

                                                        I know what you' re talking about, Jeff. I thought the one color flatness was either caused by the light used to expose for the DSLR or the light used to view the print. It wasn't necessarily texture we were going for BTW but the subtle color differentiation demonstrated in that close up of the Kroyer painting that was rendered as one color on the print. The differentiation wasn't in the digital file either. I checked.

                                                         

                                                        The gallery owner said he exposed the original painting with lights he used to light his paintings in his gallery. The painting was around 3x4 feet in size captured with an 8MP Canon DSLR. I suggested he move the camera closer and shoot the top half and bottom half separately and combine them in Photoshop, but he didn't want to go that route because by the time I was brought in and suggested it, he'ld already bought the scanner.

                                                         

                                                        And in hindsight I do remember the gallery owner indicating he switched his halogens to fluorescent CFL halogens which had the corkscrew spiral in an enclosure. Not sure if he used these with the DSLR. Also at the time I didn't know what I know now about fluorescent lights and how they cause this lack of subtle color definition and differentiation compared to hot lights. 

                                                        • 26. Re: ColorChecker Passport-Color Profiles
                                                          Tim Lookingbill Level 1

                                                          I checked the EXIF data on those linked jpegs of the Kroyer paintings and flash wasn't used and WB was set manually, so he was using the gallery's available lights. Also focal lengths are the same 5mm which means he moved in closer to get the separate close up shot so it's not a crop from the full frame version. They also have separate IMG...jpg numbers which supports they're separate shots.

                                                           

                                                          I'ld say for a P&S jpeg it did a pretty good job at capturing a lot of color subtlety even for the full frame version.

                                                           

                                                          Thinking it over it probably was the lights used with the DSLR that caused the gallery owner's lack of detail and not lack of resolution. It's still good to check for this kind of detail though even using a one shot full frame off a DSLR.

                                                          • 27. Re: ColorChecker Passport-Color Profiles
                                                            ssprengel Adobe Community Professional & MVP

                                                            Here are color-error analysis plots for several camera calibrations using the fine-art color-checker calibration shot uploaded, earlier.

                                                             

                                                            The three profiles tested were:

                                                             

                                                            1)  A custom Elinchrom Flash profile I created from the color-checker using the DNG Profile Editor,
                                                            2)  The Adobe Standard camera profile,
                                                            3)  The custom profile provided by Gustov.

                                                            ProPhotoRGB_BP0_Linear_Elinchrom-AdobeStandard-CustomGustov_colorerror.png

                                                             

                                                            My general comment would be that the Adobe Standard profile has about the same error as the custom profile I computed using the DNG Profile Editor but the Adobe profile has a more even distribution of the color error.

                                                             

                                                            The profile Gustov provided seems to be very close in the yellow area, but quite far off in the reds, blues, and violets. This can also be seen as a severe shift toward the green of the WB Tint slider value when the profile is selected for the As Shot WB as compared to the almost zero value when using the Adobe Standard profile.

                                                             

                                                            These color-error plots were created using the ColorCheck module of Imatest Studio 3.7 from www.imatest.com.  The program compares the measured color values from a color-checker image to the standard color numbers of the Color Checker then plots the difference on a CIE chromicity diagram, where the squares are the standard color position, the circles are the measured color position, and length of line in between is the color-error.  In general, error in-and-out from the center is saturation error, while error around the center is hue error.  In a 3D plot luminance error is also shown but harder to visualize without a way to rotate the plot around.  Imatest provides this visualization in its Multicharts module.

                                                             

                                                            I used the ProPhotoRGB colorspace when saving the TIF out of ACR because the standard ProPhotoRGB color-checker numbers in Imatest are for D50 lighting whereas the AdobeRGB standard numbers are for D65 lighting.  As the image-filenames suggest, I used a linear toning curve and a black-point of 0 when saving out of ACR, but left the brightness at 50 and the contrast at 25.  I tried zeroing out the brightness and contrast or using the Medium Contrast toning curve but things were way off.  Using a blackpoint of 0 instead of the default of 5 also made things slightly better.  There are scripts that can compute the optimal toning values for the gray patches but I was only comparing profiles, not trying to optimize them completely.

                                                            1 person found this helpful
                                                            • 28. Re: ColorChecker Passport-Color Profiles
                                                              gmurillofdzp Level 1

                                                              Thanks for this impressive analysis. Is there any way to create a color  profile manually (with the help of the ColorChecker) in order to get a right color profile? According to the data you've provided, neither DNG  profile manager, provided with the ColorChecker Passport from X-Rite nor  Adobe DNG Profile Editor seem able to create correct color profiles (at least in this particular case).

                                                               

                                                              I bet the best solution to solve this issue will be using a proper  screen to retouch and a proper light to lit the paiting to match the  colours next time I am shooting. An expensive solution though.

                                                              • 29. Re: ColorChecker Passport-Color Profiles
                                                                ssprengel Adobe Community Professional & MVP

                                                                An area for further experimentation would be to use Rags-Gardner's Version 8 script to further tweak the color matrices after computing what the DNG Profile Editor thinks is optimal. 

                                                                 

                                                                http://www.rags-int-inc.com/PhotoTechStuff/ColorCalibration/

                                                                 

                                                                Tom Fors originated these scripts after talking with Adobe folks about how they did their manual calibrations.  Both Tindemans and Rags-Gardner created their own versions, but after Adobe released their DNG Profile Editor only Rags-Gardner continued to update the scripts to keep up with the changes to Camera RAW, with the others stopping with CS3/ACR4.

                                                                 

                                                                As an example of the improvements a script might make, here is a gallery of the color-error using various profiles where the results from using the script are in the columns that have the word Tindemans in them:

                                                                http://www.pbase.com/ssprengel/_tech_adobecalibration

                                                                 

                                                                 

                                                                For reference, here is Tindemans' page which has a link to Tom Fors' page:

                                                                http://21stcenturyshoebox.com/tools/ACRcalibrator/

                                                                 

                                                                 

                                                                I stopped being interested in these scripts about time the DNG Profile Editor came out, too, if for no other reason that the color-error depended on the color of the lighting and was never going to be zero and my RAW adjustments in Lightroom or ACR of my nature scene photos lead to colors that were significantly more saturated than the rather dull "correct" colors.  However, I do almost always use my custom color profiles just to get the balance of saturations and hues about right for various lighting scenarios, with indoor fluorescent or other arena lighting benefitting the most, and just don't worry about further optimizations with a script.  At the time my computer was dirt-slow and sometimes the scripts would run for hours and hours or overnight.

                                                                 

                                                                Back in the day, I remember worrying about shooting the target as large as possible to minimize color-noise variations, but not so large that the lens vignetting came into play, and also worried about wearing dark, neutral-colored clothing and what was behind me to minimize any reflections from the CC target.  Assuming your flashes have repeatable brightness from one shot to the next, and you're not relying on some sort of TTL that can vary for each shot, then set your exposure and shoot the target a bit larger than you do, now, if possible.

                                                                 

                                                                The native WB of your supplied being 4600 makes me suspicious that some of the incandescent lighting might be bleeding in, but maybe that is just the color of your flashes, if the other lights are already turned off.  If they remain on when you take the pictures, you probably should test to see how much they change the WB by taking at least one shot with them off, only illuminated by the flashes, and compare that eyedroppred WB to your regular photos. 

                                                                 

                                                                I found the color profiles for lighting in the range of 5000K to 6500K were much closer than for lower temperatures like 2600K incandescent lighting which makes sense because the color checker standard numbers are usually for D50 or D65 lighting.  Another thing to remember is that the toning curve, and contrast settings change the color saturation so the colors are most "correct" with the toning slider values in ACR set exactly the same as when computing the profile.  I would guess that the scripts require the toning curve to be linear, so using the calibration for photos where it is set to the normal Medium contrast isn't going to show the "correct" colors, but they should be off only in saturation not in hue.  It really depends on what you're trying to do with your photos when you perfect them. 

                                                                 

                                                                What these scripts do is methodically tweak the 6 or 7 color-matrix sliders underneath the color profile, then read the color patches after opening the image in Camera RAW and then Photoshop, then compute the color error, and tweak the sliders differently, using the well-known Simplex heuristic, trying to minimize the color error. 

                                                                 

                                                                You probably always want to test the color error mathematically to make sure your process is free of mistakes, but it helps to have a visual check as well.  I like the Imatest color-error plots because I can easily see which have more error, but the Rags-Gardner page has Read scripts to help test the error as well.  You'd want to just use the CC24 scripts unless you actually have some of the other targets.  I like the number of colors in the IT8 target, but the fact it is glossy is a big problem.

                                                                 

                                                                --

                                                                 

                                                                If you want to try these scripts and have some questions about how to get them to work, you are welcome to send me a private e-mail, if I miss a response sent to the forum, although others may be interested in your results with the scripts, so update the thread after you've tried things:

                                                                 

                                                                My e-mail is:

                                                                 

                                                                steve @ sprengels . com

                                                                 

                                                                (without the spaces)

                                                                • 30. Re: ColorChecker Passport-Color Profiles
                                                                  Marco N. Level 1

                                                                  >An area for further  experimentation would be to use Rags-Gardner's Version 8 script to  further tweak the color matrices after computing what the DNG Profile  Editor thinks is optimal.

                                                                   

                                                                  I suggest to use the script BEFORE using DNG PE, because if the matrices are in OK the tables have to make less work, lowering the possibility to introduce posterization.


                                                                  >I  stopped being interested in these scripts about time the DNG Profile  Editor came out,


                                                                  The scripts and DNG PE work in a totally different way, the first produces a matrix that corrects the Adobe matrices, the second append a LUT to those matrices, leaving them untouched: here I have write my point of view about this question


                                                                  >my RAW adjustments in Lightroom or ACR of my nature scene photos lead  to colors that were significantly more saturated than the rather dull "correct"

                                                                  colors.

                                                                   

                                                                  Maybe for language problem I can fully understand this interesting (for me) point. Can you explain me it better?

                                                                   

                                                                  >At the time my computer was dirt-slow and  sometimes the scripts would run for hours and hours or overnight.

                                                                   

                                                                  Tindemans script works perfectly with the last version of ACR and with a decent computer a script run takes about 15 minutes.

                                                                   

                                                                  >Back in the day, I remember worrying about shooting the target as large as possible to minimize color-noise variations

                                                                   

                                                                  I think that this is not a big problem because the scripts average the center of each patch before going into calculations

                                                                   

                                                                  >Assuming your flashes have repeatable brightness from one shot to the next

                                                                   

                                                                  From my experiences if the exposure dosn't change too much (below 1 stop) this is not a problem

                                                                   

                                                                  >The  native WB of your supplied being 4600 makes me suspicious that ...

                                                                   

                                                                  4600 Kelvin is a too much synthetic index (CCT), it shoud be more interesting to see the spectral power distribution of the light

                                                                   

                                                                  >I found the color profiles for lighting in the range of 5000K to 6500K were much closer than for lower  temperatures like 2600K incandescent lighting which makes sense because  the color checker standard numbers are usually for D50 or D65 lighting.

                                                                   

                                                                  You can calculate them starting from spectral data, I link a page in Italien but if Google translator can't translate it in a good form, the images should be enough to understand the method

                                                                   

                                                                  Marco

                                                                  • 31. Re: ColorChecker Passport-Color Profiles
                                                                    ssprengel Adobe Community Professional & MVP

                                                                    I agree the DNG Profile Editor and the scripts work the issue from two different aspects which is why doing both can be better than doing only one.  I don't know enough to think properly how doing the script first would work better.

                                                                     

                                                                    The OP seems to be wanting to make things match visually on screen and real life using a custom profile followed by further tweaking.  I was saying that I like my photos of natural subjects, like green plants, colorful bugs and birds, etc, to be higher contrast and have more saturated color than what applying a custom profile with minimal color error and not doing much toning adjustments afterwards because it would increase the color error.  In other words, if I shoot a color-checker, create a profile, apply that profile to a natural scene, then adjust the toning to my liking of that scene, then apply the same toning to the color-checker photo, and compute the color error, it will likely be quite far off.  My point was that since I almost always adjust the toning on images to make them look better, having a perfectly optimal color profile might be overkill if, especially when it took hours and hours to compute on my slow computer back then.

                                                                     

                                                                    With a small target, where the OP's was only 600x400-something, the number of pixels averaged wasn't that many, so a sensor flaw in one patch could skew things a bit.  I like to minimize the issues in the parts of the process I can control so shooting a larger version of the target is one thing I do.  For the normal profile-per-photoshot things, I probably wouldn't be that picky about the size, either, but the OP is shooting fine art and has a special interest in making things as perfect as possible, so a larger target would be a thing to make things slightly better.

                                                                     

                                                                    The comment about repeatable flash exposure was in the context of having the white-balance change in a mixed lighting situation if both flash-illumination and ambient incandescent were involved.  The WB and the color-profile would change if the flash was brighter and dimmer in relation to the fixed amount of ambient lighting.

                                                                     

                                                                    The "color" of the lighting being 4600K suggested a reddening from mixing in incandescent of a supposed 5000K flash temperature.  As you say the spectral profile of the lighting is important because a light source with a few spikes of certain colors could have the same average color-temperature as one with a more even amount of each color like daylight has.  The color spikes in artificial lighting, especially fluorescent-type lighting, can cause weird hue changes for only certain colors, and is why I compute a specific color profile for those situations.