Can't get decent reds with digital cameras

Discussion in 'Digital Darkroom' started by john_dole, Mar 5, 2009.

  1. I shoot people's art work and can never get the reds right. I use a D-300, set the custom white balance using a digital gray card and studio flash (PS shows the grays in a Macbeth card to be all the same numbers for R G & B). I set the camera to neutral--although the problem shows up no matter the setting--and my screen is a hardware-calibrated NEC. But it doesn't so much matter what screen or computer I use (including the camera-back screen): All colors are quite good except reds. They are too pink or too orange, compared to the object viewed in daylight. Sometimes a deep maroon comes out brown. What Photoshop calls pure red (255, 0, 0) looks like orange to me. In fact I can't find what I would consider a real, fire engine red on the web anywhere. No matter what I do in PS, including in Camera Raw, reds are either tomato, pink or orange. Is this just a limitation of video rendering? Any help appreciated.
    JD
    00SfPD-113517584.jpg
     
  2. Hi John, what lens, or lenses do you shoot with?
     
  3. Hmm. Thought I posted this but it didn't show up: I shot this with a Nikon 18-200 VR. Have had the problem with other lenses, however, and my old D-70 as well. Will experiment with other lenses.
     
  4. Ray House

    Ray House Ray House

    In PS I went to Image-Adjust-Selective color (Red with come up as the default) and just moved some sliders around to taste.
    00SfQ7-113521784.jpg
     
  5. Is this the type of red you're going for?
    Did a hue, saturation and luminance adjust in ACR 4.6 adjusting the red slider in the HSL panel. Shooting fabric can be tricky in getting the color right especially red colors.
    00SfQg-113525684.jpg
     
  6. Thanks, guys. They are both somewhat better. I will fool with those controls.
    Do you know what I mean about a more fire engine red though?--and this pic may not be a good example, because that red, while way off--with respect to the other colors--was not really fire engine red. It seems like other colors just fall into place when I am careful about lighting balance. But reds drive me crazy. I did find some reds on the web that look pretty close and PS pegged them at about 150, 0, 0. So maybe a lightness/darkness issue to some extent. But they are still a bit orange-tomatoey (see darker parts of example image). I know red always metered funny when I shot slide film--If you metered red for mid-tone it came out overexposed.
    00SfRc-113533584.jpg
     
  7. Sure sounds like you have a monitor problem if you can't find what you feel is red anywhere you look. You said that even if photoshop says a red is 255,0,0 it doesn't look red, this would indicate your monitor is off. The red in your posted photo sure looks like red to me.
     
  8. I'm with Don on this one - all your posts look perfectly "red" on my monitor.
    What do the images print like - if ok, it looks like it may be time to upgrade your monitor.
     
  9. Some thoughts:
    If you're used to the "fire engine" reds of slide film then digital red can be a bit problematic.
    Ray's edit is more down to earth than your original. It doesn't "glow". Is it more accurate? I don't know, haven't seen that bag.
    PS 255 red can look a bit orange / too bright, so yes, I think I know what you mean.
    Color is subjective and different colors next to each other affect how we perceive them.
    It could be that your monitor (or simply calibration) is going belly up but somehow I feel this is more a matter of personal perception.
    Are you happy with your prints?
     
  10. rnt

    rnt

    Looks red to me too on my calibrated LCD...
     
  11. I've never been able to get the same reds from ACR that I get from Nikon NX/NX2. That includes using all of the beta profiles available for ACR. Perhaps it's my fault, all I know is that the Nikon software produces better reds on my calibrated Macs. I actually prefer the work flow of ACR, but I hate the reds. Suggestions?
     
  12. John and Ray..this red wont print in CMYK, and maybe not even on a external lab. Be cautious about teh use of the hue saturation.
    Tim, the red now look good and printable, is it the right one? maybe yes / no..but you can print it.
    As for the limitation, i would say its a user / your camera brand *problem*, not a digital camera problem. If you can see it good in NX2 theres is NO eason why you cant get it rigth in any other raw developper, other than your own knowledge.
     
  13. Also, in many situation like this i prefer the use of the selective color vs the hue saturation, i find it pretty fast and intuitive for many user..more like the Ligthroom color panel.
     
  14. Reds are tough because digital cameras are overly IR sensitive compared to human eyes--they see reds that are too long in wavelenth for the eye to see. There is also a wide variation in how people perceive red--color blindness is a common problem, affecting roughly 10% of the population on this board. There may be the reverse problem a the other end of the spectrum, involving UV and blue light.
    The most practical solution may be to shoot in RAW and adjust the curves for less red and more blue, until it looks right. You need all the dynamic range possible to avoid overexposing the red channel. In theory, better filters would help, to make the camera's spectral sensitivity exactly match that of the photographer.
    A more difficult problem is florescence--it is well known that some minerals will take UV light and convert them to visible light--there are Natural History museums that display this phenomena. To fix this you may need to filter the light source, so the unwanted wavelengths are not converted to other colors.
     
  15. You're right, Patrick, it's my lack of knowledge. I cannot address how ACR renders Canon images, only Nikon's. We also have to remember that Adobe does not copy Nikon's algorithms with their camera profiles, they only approximate them. So, if I knew how to get the same reds using ACR, I would, but I don't know. I do know that I don't have to do anything to NX/NX2 to get correct reds. ;o)
     
  16. What are you using as a light source?

    Which version of ACR and Photoshop) are you using?
    If Photoshop CS4 have you tried Adobe's beta profiles?
     
  17. I use CS 3 w/ACR 4.6.0.30.
     
  18. jtk

    jtk

    IMO the answer has to do with light source and the nature of the fabric.

    Some of the aniline dyes used on especially good wool (as in some hand-woven items like your examples) can be downright bizarre when photographed in daylight or strobe.

    You may be happiest with tungsten light because daylight and strobe pump out a lot more UV.

    Using a strobe or daylight, compare with and without UV FILTER on camera...might help, might not.

    This is also a well-known problem with white clothing...it's often got "whiteners" (or its been washed with whitening detergents) that glow under strobes, making it a virtual light source rather than simple reflectant white...the well-known fix for that, if you're shooting a catalog with studio strobes, is to refit them with optional strobe tubes with UV filters.
     
  19. Looks red to me. Eyedropper in PSP says there's not a speck of green on the image so it can't be orange. Are you sure your screen's adjusted right and/or you aren't looking at an image in absolute colorimeric or something?
     
  20. Wow. Thanks everyone for your thoughtful comments. I am using old Paul Buff strobes, so I suppose there could be some lighting issue, esp re fabrics. I could also be a little colorblind, but when I shoot things with reds, people will often comment that everything is correct except the reds. That's on the camera back, my monitor, their monitor, etc. D300 or D70 (CCD, CMOS). I am mostly working for the web, so I don't have much to add re print color. I just shot quickie of a Macbeth color card in window light, after first doing a custom white balance. To my eye all the colors look pretty close to the card--except the red. Even when darkened a bit (spot in middle), it is just not close to the card. It is way more orange. Looks the same on the camera screen, and the D300 has a decent screen. Maybe I am crazy. Again, all comments have been greatly appreciated.
    00Sfe4-113595784.jpg
     
  21. Given that you have a Macbeth ColorChecker, there's a simple cure for your problems: get the Adobe DNG Profile Editor and feed it appropriately-lit shots of the ColorChecker. It can use those to produce a profile for your camera. This should give you reasonably accurate color rendition.
     
  22. jtk

    jtk

    This is interesting and it may prove a point.
    IMO it's not likely that a "red" on the Macbeth ColorChecker is equivalent to "red" coming from aniline dyed, high quality wool...which "looks" red but doesn't photograph that way, even when everything else is perfect.
    If you're having the same problem with non-woolen reds or factory-woolen-reds, I yeild the point.
    By "fine wool" I'm referring to the best from New Zealand, Great Britain, and Scandanavia....although small batches of the best are also grown in the US. It's sold specifically to people who do hand-weaving, who sometimes spin their own. This stuff is almost prismatic, not at all like the wool we're used to in factory woolens.
    Since you're using Buffs, you do have modeling lights...you might try exposing with them alone, no flash...that'd test the tungsten idea I suggested earlier.
     
  23. If all you have problems with are the reds, do the simplest thing first and that is use ACR's HSL panel. Concentrate on using the red hue and luminance slider. The splotch you have on your CCchart posted can easily be attained using the HSL panel. Don't use the Calibration panel sliders. They affect a wider range of colors where as HSL isolates to a specific color without affecting other colors like oranges when you want to change reds.
    You can do this on jpeg or raw. The Calibration panel sliders won't work on jpegs.
    The HSL panel is engineered VERY differently and for the better than any color tool you can find in Photoshop. The luminance slider retains hue without adding green when increasing luminance which has been the primary cause of orangy looking reds when increasing saturation. The saturation slider also makes reds orangish increasing saturation using Photoshop's Hue/Sat tool. The HSL panel detaches this green channel influence for attaining richness, vibrance and luminance in reds without shifting to orange.
    You need to give this a try before you do anything else. It's the easiest solution. Try it using any number of profiles selected in the Calibration panel to get you as close as you can.
    Pick the profile that makes everything the closest and work from there on the reds.
     
  24. I have three suggestions:
    (1) If you're using studio flash, try using the Daylight white balance preset or a manual setting that's a tad warmer (5500-5700K).
    (2) Try using the Vivid optimization preset.
    (3) Try underexposing slightly (during capture or during post-processing). Normally-exposed red looks a little to "cheerful" to my eyes. I like a slightly darker, more somber red.
     
  25. Looks Red to me, have you thought about getting your eyes tested for color blindness? Thats why all these discussions about sharpness, the merits of blue ray DVD vs whatever, H Definition screens, HHD, Super Duper Extra High Def etc etc etc are all Bullcrap, If the Human eye had perfect vision, and Opticians the world over went Bankrupt, then there might be some point to these discussions.
     
  26. The image at the bottom was shot within a foot of a 50watt Solux clip-on task lamp. It is a spectrally flat halogen MR16 bulb with a rated 4700K, 98CRI. The neutrality is exactly as it appears meaning I didn't click for R=G=B on the WhiBaL target.
    It is edited to look as it appears only the artificial red peppers are much more redder, intense and brilliant under these lamps causing the red to be beyond the gamut of my display so I have to fool the eye into seeing a "richer" looking red using ACR's HSL.
    The only difference between the one on the left is that I decreased the red Luminance slider by (-25) and the red Hue slider by (-30) with Saturation at max (+100). The one on the right had an increase to the Luminance slider by (+20) and Hue was at (0). The profile I had selected in the Calibration panel was Adobe Standard which rendered the overall image closer to what I saw than the DNG Profile Editor CCchart Wizard Profile and the default ACR 4.4 profile.
    00SfoG-113627584.jpg
     
  27. Thanks, Tim L. That is a great comparison. The left one is much better, and redder. I see your point. Everyone else has good suggestions too, and I will try the Adobe DNG Profile Editor, and fool around with exposure and light sources. Tim E., you are right I should get tested. But by my "standard" I sill see a big difference between object and screen. Here is a color chart from the Adobe site Jerry referred to. Not sure if it is a Macbeth. Isn't exactly the same as mine (which is old), and it is clearly optimised differently. But look at the red. It is better than mine, even taking density into account. That red, while still a bit tomatoey, is, like Tim L's left pic, closer to what I am looking for.
    00Sfog-113629584.jpg
     
  28. I would say you are approaching magenta.
     
  29. I've seen too-orange reds in many Nikon images. I believe it is due to clipping in one or two of the channels.
    I spent about a year playing with ACR's calibration tab, to get reds to look right:
    [​IMG]
     
  30. John, your downloaded Adobe chart is a good representation of a real Macbeth Colorchecker, in sRGB (this means you must assign an sRGB profile in Photoshop). If you can approximately match these values from your camera, it should do a pretty fair job of reproducing artwork. In fact, if you sent output to a well-profiled printer, it should look very similar to a real Colorchecker.
    BTW, earlier today, I read values from the chart you photographed. Your problem is more than just red, although it IS the worst one. The next four, with roughly the same amount of color error, are: Green, Blue, Dark flesh tone, and dark green. On the flip side, the three BEST colors (aside from the neutrals) are Cyan, Magenta, and Yellow. When I say color errors, I mean in terms of Tim's HSL (hue, saturation, and lightness), but ignoring the L part. Mostly, you have too much saturation, but not on everything, so it is not easily corrected. I really think your camera has an odd setting, perhaps a saturation boost or enhanced red-green tones. I don't know Nikons, so can't even make a suggestion here, but if you find it and are able to roughly mimic that downloaded chart, your artwork should come pretty close right out of the camera.
    By the way, the settings you use to reproduce artwork will NOT work well for general photography; they'll look too dull. So for general shooting, you need to use different settings, that have a little steeper tonal response (more contrast). Good luck.
     
  31. Aa Lad partially suggested, you could be blown out your red channel, giving at a nasty look. Especially in the lighting shown in that shot that is predominantly red, skewing everything towards red. Open up your channels pallet and look at your red channel and see if it is overly white or blown out. My guess is if youre getting a reading of red 255, you have.
     
  32. I have a shooter friend that uses the Nikon (forget the model), but keeps asking me about his reds being so off. He upgraded to an new Nikon as the store took it back, and he still had the issue. This maybe some batch of defects, but he gave up and now switched to a Canon. I only had Nikon F film body, so i don't know much about it, but you are not alone.
     
  33. Once again, thanks all. I have been reading about digicams, and the D300 in particular, over-exposing reds. I am realizing that this is an issue, which adds to the hue issue (I still think screens have trouble with "pure" reds) , and I am liking the ACR > HSL > Luminescence (and Hue also, although my example just changed Luminescence) approach.
    Thanks Tim L. This really is the easiest way. I guess I can reduce red saturation in the camera, but I expect it will not work as well.
    00Sfvo-113665784.jpg
     
  34. From a LED wide gamut perspective, in SRGB it looks great!:)
     
  35. If I use this monitors SRGB mode, colors look dead. In the Calibration mode which I used the Eye1 to calibrate the LUT, the colors look vivid.
     
  36. Hi there . my opinion realy simple I don't know much about color chart but I alway playing fun with COLECTIVE COLOR in photoshop some how if you want RED but not mean you have to adjust RED maybe blue/green or neutral depent on your file w/good white ballance so that mean I alway white ballance to the lighting I shot not on auto , some people said you can auto white ballance then adjust later on PS to me NO very diffenece when you shot w/right that time .
    00SfzL-113681584.jpg
     
  37. I suspect orangey reds are an endemic problem with Bayer-pattern sensors because they have twice as many green photosites as red. Reviewers obsess about high-ISO noise but inaccurate color is mentioned only in passing, although results are there for everyone to see. My suspicion is that what you have here is an Adobe and Nikon problem. Look at the dpreview.com D300 Macbeth chart and note how Bibble has the most accurate red rendition, followed by ACR, with all three Nikon renditions far behind, way too orange. The color dropper shows this to be true, no matter what your monitor calibration. BTW, my Dell Ultrasharp monitors at work show oversaturated pure reds, so this is possible for LCD.
    Doubling gamma (similar to raising Curves) accentuates Tim's red vs orange pepper, but what is very troubling is the color banding. Hard to say why this happened, the EXIF is bare, but I could see it even without doubling gamma.
    00Sfzh-113683584.jpg
     
  38. Apparently neither sRGB nor adobe-RGB (monitors) can show real reds; something that kodachrome apparently does. In my experience, shooting slides and projecting on the wall really gets things right with minimal effort. I've been fooling around with color spaces and expose-to-the-right (digital raw) for some time now and am beginning to think I'm retarded. The colors simply do not add up. Maybe when a prophotorgb monitor comes out... Interestingly, the DSLR sensor may already be prophotorgb (or close). My HP LP2275w LCD (adobe-rgb) does show minutely more saturated colors than my fujitsu A3130 laptop (srgb). See this:
    http://www.drycreekphoto.com/tools/printer_gamuts/gamutmodel.html
     
  39. In painter's terms, of the two red peppers above one is cadmium scarlet, at right. The other is crimson, which has a bluish tone. The bikini bottom is also cadmium red with a hint of scarlet. They are all reds.I don't shoot color and can't tell you how to fix your problem, but fing it interesting because in certain other mediums a good red is also difficult. In ceramics you have to use a gas fired kiln, often illegal. Otherwise red comes out brown. If you paint with gouache red is also the problem. It is never red enough. I wonder if the problem is with the structure of our eyes.
     
  40. John, I'm with the guys that say, "Looks red to me." Maybe a tad on the orange side but only slightly. Also the calibration of you monitor should be checked. There's a nice before/after check at the end of the Spyder2 software. You might see the difference plainly there.
     
  41. Thanks Bernie. I think I am seeing what you are seeing, more or less. I have an NEC calibrated with Spectraview and a Spyder2 puck. As with walking from a tungsten-lit room out into the sunlight, our eyes adjust to color changes. You only notice the difference when things are side-by side. The only real test is when you hold the object up next to the monitor. The Adobe color chart I posted, with some lightness/contrast adjstments, comes pretty close to my Macbeth card, so I think I have to look to the camera more than the monitor.
    Nice red, Jordan. About as good as it gets on a monitor. It seems like you can get a better dark red than lighter red. Does that make sense? And yes, the red channel is only part of it.
    Bill T: That dpreview raw converter comparison is eye-opening. The differences are stark. Another seemingly wild variable in the world of digital photography.
    Indraneel: Thanks; the Dry Creek color space viewer is another great tool. Now I am going to have to find the time to experiment with all these great suggestions...
    Bruce: I know what you mean. My wife is a fine art painter. When it came to painting the house we bought she was frustrated that she couldn't even approximate her artist's paint colors with house paint--especially reds. And I can rarely find a red shirt that I like. They are almost always too orange. But now and then I find one.
     
  42. Bill Tuthill,
    I have the same suspicion as you concerning the prominent green in digital camera's Bayer sensor arrays as the cause for the orangy reds. Also I've seen 255,0,0 RGB red on a display look different between display models and whether it's an LCD or CRT depending on their brightness and color temp appearance. My old laptop has the worst chalky orange looking pure RGB red I've ever seen. And when you think about it the only way you can get a vibrant, bright, intense red without it looking orange is to pull back the green channel which of course makes the red appear darker reducing its brightness.
    The image of the peppers was shot quickly under low spot light and is a png format screengrab resaved as a jpeg off my 2004 sRGB gamut iMac so I don't doubt there's posterization. It was meant as a quick demonstration. Another reason for the posterization is I had the red saturation slider in ACR's HSL panel at 100% so I'm pretty sure I was maxing out the gamut of my display.
    Gave another go at shooting those red peppers and found a better HSL setting where all I had to do is apply +40 red saturation and -40 red luminance after setting a more accurate contrast appearance. This time I saved out as jpeg and resized in CS2 so EXIF data is intact. You'll see what I mean by the green having to be pulled all the way to zero converting to sRGB to get a less orangy red. In ProPhoto RGB their was no clipping in any of the channels.
    00SgBd-113785584.jpg
     
  43. I heard a similar complaint from a Design Director at work - that they just cant get the juicy reds and yellows in the product prints they used to since they went digital.
    Nothing beats slide film in that department!
     
  44. There are at least two things that might be going on.
    On my calibrated LCD monitor, the original picture looks orange unless I open it in a program that is color management aware. For example, in Internet Explorer I see orange, but in Adobe Lightroom or Firefox 3 I see what I would call red. Without having seen the original subject I can't tell if it is the right shade. My calibration only loads the tone curves into the video card hardware and handles the color correction in software. Even though Photoshop probably handles it automatically, you might check Photoshop's preferences for display profiles (I don't have Photoshop so I'm not sure where to look).
    The most common color space used on the web and assumed for uncalibrated monitors is sRGB, and it doesn't have a ruby red color in it. It's possible that you are trying to get a redder red than most monitors can display or most printers can print.
    Basically when you say that (255,0,0) pure red in Photoshop looks orange that seems like a red flag, pardon the pun. I recommend figuring out why first before you modify the images.
     
  45. You guys sound like those people who sell paint at Home Depot. "No sir, you don't want arctic snowflake, you would be much better off with glacial moonbeam." Anyway, when I started scanning film about 4 years ago, I soon realized that if you got the skintones right, the rest didn't matter too much. If a flower is off color, it has a 50% chance of being an improvement. But if Aunt Mildred's complexion is off, it ruins the whole picture.
    With that said, I have had very good luck scanning. Here is a scan of the old Kodak Gold 200 made with a Nikon Coolscan LS-2000. This was scanned before I knew what color correction was.
     
  46. You guys sound like those people who sell paint at Home Depot. "No sir, you don't want arctic snowflake, you would be much better off with glacial moonbeam." Anyway, when I started scanning film about 4 years ago, I soon realized that if you got the skintones right, the rest didn't matter too much. If a flower is off color, it has a 50% chance of being an improvement. But if Aunt Mildred's complexion is off, it ruins the whole picture.
    With that said, I have had very good luck scanning. Here is a scan of the old Kodak Gold 200 made with a Nikon Coolscan LS-2000. This was scanned before I knew what color correction was.
     
  47. Here it is ...........
    00SgN5-113873784.jpg
     
  48. I wouldn't say you can't get decent reds with digital cameras.
    00SgO3-113887584.jpg
     
  49. Neither the vegas airplane nor the ginger are really red. When you want to see red, think Van Gogh. Poor fellow got that right, although his first image (potato eaters) was a tad low in brightness. Now it's time for me to chop off my ears..
     
  50. John, I am personally in the camp of people that thinks that reds on monitors, televisions and pretty much any digital display looks like crap. This is not just a fly-by-night opinion, but one that is garnered from working in front of computer monitors doing digital editing and painting since those things became possible at home in the late 1980s. I have also spent more than a decade painting with high quality artists oil paints. I have spent many many hours attempting to get colors in Photoshop to look like the colors in my paintings with great disappointment.
    One of the great falsities of not just digital photography, but ALL photography is that there is no film on the planet or digital sensor on the planet... or print or display for that matter, that can accurately represent ALL colors. As red is the least common color we encounter and our eyes are not very sensitive to it, I honestly beleive that red is where designers let out the slack. The reason that a sensor would have twice as many photosites for red as for green is because our eyes are more sensitive to green and the assumption is that if you get the greens close to right no-one will notice. The people who WILL notice are people who spend their lives looking at REAL colors in the form of dyes and pigments. And no, color photographic or digital prints are not what I'm talking about. For that person, who knows about all of the different pigments and what they mean in terms of visual and emotional response, color photography and especially digital displays will consistently be a let-down, probably for some time to come.
     
  51. Because we see less of it, red is the most important color. Vegetation has so many green shades that accuracy is not hugely important, provided there is differentiation. And we are used to off-cyan off-blue skies due to film, most emulsions of which add purple. I wonder how well the Foveon sensor performs for red?
    <p>
    "sRGB, doesn't have a ruby red color."
    <br>
    sRGB lacks nothing in red gamut compared to Adobe RGB.
     
  52. I find there's nothing wrong with the way display's render any color, even sRGB gamut displays. The issue really is about how light with regards to quality and quantity used to light the scene and then processed with a digital camera when shooting intensely colored object is most influential.
    You can't fight physics. Folks want blazing bright images with dramatic contrast the way their eye perceives the scene as captured and expect it to render faithfully with the limited amount of brightness and contrast inherent within a display. It's all about surrounding density, contrast and color temp appearance rendered within an image that influences our perception of red. The user must edit the image for the desired intensity, contrast and brightness relationship within a scene combined with the limits of a display which in my experience can render reds quite well. It just takes some effort getting there.
    To show this color relationship below are four different red squares grouped together created in sRGB of varying density so as not to induce an orange cast while at the same time retain brilliance/high luminance. By checking the RGB readouts the first upper left small square is pure 255 red and the rest are variances toward getting a more redder looking red while keeping density/luminance as high as possible.
    The warm and cool colored surrounds are to show how perceived intensity and color cast (orangy's) are affected by this surround. The second and third rows are to show how various gamuts assigned to sRGB can show how intensity, color cast and density affect red's appearance. The PhotoGamutRGB version shows how average minilab printers will render such reds.
    You'll note on the first sRGB row of red squares look different in intensity depending on their warm/dark cool surround color. Out of all the small red square shown which looks the most orange? The 255 red and the Wide Gamut RGB version looks the most orange on my calibrated iMac display
    Those with wide gamut displays need to view this color chart in a color managed app.
    00Sgrx-114167784.jpg
     
  53. Note how the color swatches are rendered in this page of some of the finest and most expensive paints on the planet:
    http://www.winsornewton.com/products/water-colours/artists-water-colour/colour-chart/
    They look pretty accurate in hue and density but the intensity is diminished compared to how I remember these paints viewed in direct full spectrum sunlight. Cobalt blue is the hardest color to render on an inkjet because all you have is cyan and bit of magenta and black ink to render this deep, rich and vibrant color. In painting class I was told black was a big "NO-NO" when wanting to render such richness and density because it absorbs light and thusly the reflected color pretty much dulling these rich, dark colors considerably.
    A display has no problem rendering Cobalt blue as shown here:
    http://www.donnahickmanpottery.com/Cobalt%20blue.jpg
     
  54. I haven't had problems with reds since I switched to the j holmes DCAM work spaces. I'm using DCAM 4 right now because of intense oranes and bright reds.
     
  55. If I remember correctly from Tony Couch's Watercolor books, he always stresses using complementary colors side by side to enhance apparent intensity. Watercolor has the limitation that the brightest white is paper white (unlike pigment white in oil or gouache); kind of same as a transmissive display (LCD, CRT). Personally, I feel no amount of bright/true colors will make an (computer monitor) image bright/true unless color contrast/design is carefully monitored. I have absolutely no idea how slide film pigments solve this problem. Does anyone know?
     
  56. Indraneel,
    My guess based on observation is that film records luminance in intense colors that is detached from the overall luminance/contrast of the overall scene. This is quite inaccurate where as DSLR's generally keep these color relationships balanced at the sacrifice of hue shifts (orange reds) when it records bright daylight shots.

    This Kodachrome scan:
    http://www.flickr.com/photos/23585735@N06/3173277378/in/set-72157612226326832/

    from this site:
    http://www.flickr.com/photos/23585735@N06/sets/72157612226326832/

    that's been linked on this forum before shows a noticeably bright red coat that is abnormally much brighter and intense than the overall dark surround. My color chart was an attempt to show how this changes the perception of colors.
     
  57. Geez! I can't believe this.
    I conducted a google image search on the color red and every image I found whether I entered crimson, scarlet or deep red came up looking orange. Flowers, sports cars, dinnerware, fabric you name it, it all looked orange no matter how deep, rich, dark, saturated or intense.
    So I'll make it official here. This is the most intense, bright red you can get viewed on a color managed RGB display system without it looking orange and with no clipping of any channel. The numbers I came up with in my iMac monitor space and NOT sRGB is 250, 5, 70 in Photoshop. This swatch needs to be viewed in a color managed app. Not one image I found on the web had these color numbers. Not even film images.
    00ShX4-114431584.jpg
     
  58. Thanks Tim. I guess it never occured to me that everything other than an LCD/CRT is a 1bit/color image. I guess dithering would be the right word to describe how all the hues and shades are generated (albeit randomly for film but with precise algorithms for a printer). That would surely mean that film has many many times better resolution/dynamic range than digital, and that this resolution would vary with the hues/luminosity being recorded. I'm wondering that given the 12MP image from today's DSLRS, can the quality be improved if one changes the dithering/bayer interpolation to obtain only a 1MP image? Can this be done from the already interpolated RAW file? Taking this one step further, LCD displays nowadays provide 100ppi (an improvement from 72ppi). Can one display a 1bit/color dithered image on the screen that will look more saturated when viewed across the room? How about using the same pigments in slide film for display screens (with additional per pixel light switching mechanisms, and then we can buy a kodachrome/velvia/provia screen to suit us)? Considering that even CRT/plasma don't really have a large gamut, maybe the problem isn't even solvable (today) by any transmissive display technology. From wikipedia (on Phosphor):
    Red: Yttrium oxide -sulfide activated with europium is used as the red phosphor in color CRTs. The development of color TVs took a long time due to the long search for a red phosphor. The first red emitting rare earth phosphor, YVO4,Eu3, was introduced by Levine and Palilla as a primary color in television in 1964.
    Now let's see how much the new laser TVs cost...
     
  59. Tim, regarding your 250, 5, 70 swatch above, what background should I paste it on, and should it be viewed with squinted eyes? And if so which one, I heard somewhere most people see more yellow with the right eye ;)
     
  60. Indraneel,
    About the red swatch. It's suppose to look red viewed by itself on a white background in a color managed app and hardware calibrated display. This red swatch number combination can stand an increase in the green channel by quite a bit, but it will start leaning toward a chalky coral and lose richness. Too much blue and it leans toward purplish red or lipstick color.
    You can assign a number of color spaces including your monitor, sRGB and others and see how orange it gets. If it looks orange in this browser then you need to drag and drop to the desktop and retain the embedded profile when prompted opening in Photoshop. It should not look orange in Photoshop. If it does look orange in Photoshop, your display profile is messed up and you need to recalibrate using a hardware calibrator.
    Not sure what you mean about everything other than an LCD/CRT being a 1bit color image. As for film having more dynamic range than a DSLR image, I have no evidence for that but would just have to say they are just different in the way they record color.
    The Bayer interpolation/demosaicing algorithm used to process raw images and manufacturer's jpeg is subjective and interpretive according to those that write the conversion software so quality will differ accordingly. The internal source profile used while editing in the raw converter will also play a big part in quality of color rendering.
    Any color be it of any bit depth will increase somewhat in perceived saturation the farther away you view it. The gamut of current displays have no problem showing a wide range of colors correctly. It's how image capture devices and their hardware map the color recorded to look correct.
    That red swatch posted shows the max luminance (around 56 Lab Luminance) most displays can render this color without it turning orange. So a digital capture of a bright red object's brightest highlight other than it's white spectral highlight would be around 56 Lab luminance which is pretty dark for a highly saturated red. Luminance and saturation and accuracy of hue are the guides at determining a display's gamut capabilities.
    I have yet to determine how much of these hue, saturation and luminance color errors like orange reds is attributed to the camera's hardware and software as opposed to the limitations of the display gamut. But as you can see a good bright orangeless red can be displayed. It's just getting the camera and the software to map this red into the image is what seems to be the difficult part.
     
  61. "...it all looked orange no matter how deep, rich, dark, saturated or intense."
    YES. And your patch looks orange. The only way you can approach red is to reduce luminescence in the HSL panel, as you suggest earlier. Bright red does not exist on monitors. Dark red sort of does. In real life we know what bright red is. Somehow it becomes orange or pink on a monitor. I think this is what Indraneel is suggesting by mentioning squinting. Your patch approaches an acceptable red if you squint enough. Or if you reduce luminescence as shown.
    00ShkV-114535584.jpg
     
  62. John,
    Your patch appears as a dark but rich red brown. It doesn't look red at all.
    I toyed with making the No Orange Red patch darker the same as yours in Photoshop's Color Picker where the patch was originally created and noticed different results using the Lab Luminance vertical slider compared to the Brightness vertical slider in the HSB section and just reducing the Red channel all to make each version dark from the original 56 Luminance to 46. See the results below.
    This is further indication that software plays a larger roll in mapping correct looking colors than does the display though I do suspect those with wide gamut monitors may see extra orange in their more brighter and intense reds than what I get on my iMac which is a tad larger than sRGB especially with regards to having more brightness which LCD's have over sRGB. sRGB was created from averaging measurements taken of a wide range of 1995 CRT's which weren't all that bright back then.
    Also after further study of ACR's HSL behavior, I find it mimics more of the Lab version and maybe a little toward HSB. I still get too much yellow in HSL using the Luminance slider. I find I have to reduce saturation more to get less yellow in reds.
    00Si7L-114729584.jpg
     
  63. Correction on the Lab Luminance readout on that third patch. Made a mistake on the luminance which should read 46 when it's showing 49. For some reason when clicking out of the Color Picker I get incorrect readings and have to make it darker in the CP than to get it to read 46. Another software oddity.
    00Si81-114733584.jpg
     
  64. Decided creating a nonclipping bright red in my monitor space wasn't a good idea so came up with this RGB combination in AdobeRGB=215,12,73 and converted to sRGB which yielded 251,5,73 which gives a completely different look compared to the iMac created version even though the numbers are close to being the same in either iMac or sRGB. I can barely make out the iMac center square but I can see more orange in it which distinguishes it from the Adobe version.
    What's intrigued me about this subject is that there have been countless complaints over at the Adobe forums blaming them for orange looking reds in their raw converter. Adobe over the years has had to bend over backward coming up with newer profiles for mostly Canon and Nikon DSLR's in ACR addressing this issue. I never knew how hard it really is to get a bright looking red to look RED without it looking orange.
    00SiAL-114747584.jpg
     
  65. What Photoshop calls pure red (255, 0, 0) looks like orange to me.​
    This color is out of gamut on my monitor, so the color that is displayed on the laptop I'm currently using, is actually not 255-0-0 but something that is clipped or converted. To check if this is your case, too, In Photoshop go to View -> Proof Setup and set the soft proof to your monitor profile. Then set both Proof Colors and Gamut Warning on. You should see red if the color is in gamut or grey if it is out of gamut. You will see that 255-0-0 will be out of gamut as well, so no matter what monitor are your customers using, as long as you give them the file in sRGB or show it on the web, it will be off.
    So, here is what I would do as a starting point:
    • Start with calibrating your ACR using the DNG profile generated using your GMB color checker. Make sure that your GBM checker is very evenly lit. The tools you need for that is your camera, ACR and the rest should be available at http://labs.adobe.com/wiki/index.php/DNG_Profiles
    • Use ProPhoto RGB color space when converting the image from RAW in ACR as opposed to sRGB or Adobe RGB. Set the output from ACR as 16 bit as opposed to 8 bit
    • After you are done with editing, use the soft proof feature to make sure your colors are not out of the target gamut (sRGB for web, printer profile for print etc.)
    • If the red is out of gamut for your target color space, tweak the red to your liking using new adjustments layer of type Selective Color or Hue/Saturation or something like that.
     
  66. The end of the first paragraph shoud read:
    You will see that if you soft proof using sRGB , 255-0-0 will be out of gamut as well, so no matter what monitor are your customers using, as long as you give them the file in sRGB or show it on the web, it will be off.
     
  67. John Dole got me to rethink my adjustments with ACR's HSL sliders to get a redder red with those plastic decorative peppers I shot under artificial light. I reshot outdoors using direct sunlight and got better results. Also instead of darkening using the red Luminance slider, I decided to desaturate the red and boost red luminance. I think the problem may be red is over saturated by default in digital captures and/or ACR default settings and profiles. The CCchart based ACR profile made the peppers VERY orange.
    Here's the pepper's shot under direct sunlight with one on the left without HSL using Adobe Standard profile and the one on the right readjusted in ACR using HSL. Both don't match the real color as my eyes see it which is an indication the color of these peppers are beyond the gamut of both the camera and display:
    00SiP5-114853884.jpg
     
  68. There are many "reds".
    The more "red" you can see on a monitor is the red primary in the monitor profile.
    If you have an accurate profile, you can get the primaries (XYZ or xyY).
    Starting from that values it is possible to compute the RGB values in any space (or L*a*b* or LCH).
    That may be interesting if you want to know what "red" you can have on monitor.
    But this may be different from the red you can have from a camera.
    If you have a "red" sample, flower or panel or anything else, then you can shot it.
    dcraw can generate a XYZ output and this file may be investigated to find the red values.
     
  69. What Photoshop calls pure red (255, 0, 0) looks like orange to me.​
    In which color space?
    If red turn orange, I suspect a bad profile.
     
  70. John
    it is a long shot (and peraps unlikley), but have you considered that you may have a slightly different vision system? Three colour model presupposes a few things about colour vision, this can perhaps explain your different perception of reds. On Wikipedia you can find this:
    The OPN1LW gene, which codes for the pigment that responds to yellowish light, is highly polymorphic (a recent study by Verrelli and Tishkoff found 85 variants in a sample of 236 men), so up to ten percent of women have an extra type of color receptor, and thus a degree of tetrachromatic color vision. Variations in OPN1MW, which codes for the bluish-green pigment, appear to be rare, and the observed variants have no effect on spectral sensitivity.​
    Try reading this too which describes how some people have 4 colour vision sensors. This would result in real pigments being percieved similarly by you but 'false' colour created by three colour models (RGB or CMYK) as being unmatching others vision.
     

Share This Page