Jan 14, 2008

Camera Raw and the shadow output levels


Some image editing habits become so ingrained that it is hard to unlearn them when newer ways come along and make the old favorite methods redundant. An example of this is the subject of setting the output Levels in Photoshop for the print output. For many years Photoshop users were taught to set the output levels for the shadows at a slightly higher value than 0,0,0, even though (as you will read here) it has not been necessary to do so for quite some years now. Then came along Camera Raw and Lightroom and some photographers have been agonizing over how to set the output levels for an image when there is no output levels control in Camera Raw or Lightroom. Hopefully the following article will help shed some light on how Photoshop is still able to manage the output levels for you and why the solution is really a lot simpler than you would think.

Conflicting advice
You will sometimes come across advice that the output levels for the black point in an RGB image should be set to something like 20, 20, 20 (for the Red, Green, Blue RGB values) The usual reason given for this is because anything darker than say, a 20, 20, 20 shadow value will reproduce in print as a solid black. Just to add to the confusion, some people may suggest different numbers for the output levels: someone may suggest using 10,10,10, while another advises you use 25,25,25. In all this you are probably left wondering how to set the Blacks in Camera Raw, since you can only use the Blacks slider to clip the Black input levels and are offered no means to set a black levels output value using of the above suggested output settings.

This is one of those areas where the advice given is more complex than it needs to be. It is well known that because of factors such as dot gain, it has always been necessary to make the blacks in a digital image slightly lighter than the blackest black (0,0,0,) before outputting it to print. As a result of this, in the early days of digital imaging, the only way to get a digital image to print correctly was to manually adjust the output levels so that the black clipping point matched the print device. Back then, if you set the levels to 0,0,0, RGB, the blacks would print too dark and lose detail in the shadows. The solution therefore was to set the Output levels point to a value higher than this (such as 20,20,20 RGB), so that the blacks in the image matched the blackest black for the print device. These are the background reasons for such advice, because the black levels had to be adjusted differently for each type of print output including CMYK prepress files.

For the last 10 years or so, Photoshop has had a built-in automated colour management system that is designed to take care of the black clipping at the print stage. The advice these days is therefore really quite simple: you decide where you want the blackest blacks to be in the picture and clip them to 0,0,0, RGB (as discussed on the previous page). When you save the image out to Photoshop as a pixel image and send the image data to the printer, the Photoshop or print driver software will automatically calculate the precise amount of black clipping adjustment that is required for each and every print/paper combination. In the Figure 3.xx example you can see how the black clipping points for different print papers are automatically compensated when converting the data from the edited image to the profile space for the printing paper. But don’t just take my word, it is easy to prove this for yourself. Open an image, set the Channel display in the Histogram panel to Luminosity and refresh the histogram to show the most up-to-date histogram view. Then go to the Edit menu, choose Convert to Profile, select a CMYK or RGB print space and compare the before and after histograms.


Figure 1. Here is the image that was used to generate the histogram panel views shown in the Figures below.


Figure 2. This Histogram panel view shows the original histogram for this ProPhoto RGB image. Note that this histogram and the others shown here all used the Luminosity mode since this channel view mode accurately portrays the composite luminance levels in each version of the image.


Figure 3. This histogram shows a comparison of the image histogram after converting the ProPhoto RGB data to a print profile space for Innova Fibraprint glossy paper printed to an Epson 4800 printer. The print output histogram is overlaid here in green and you can see that the black levels clipping point has been automatically indented.

Figure 4. This example shows the print profile for a Somerset Velvet matte paper printed to an Epson 9600 printer, again colored green so that you can compare it more easily with the before Histogram. The black clipping point is moved inwards even more here because the matte paper needs a higher black clipping point to avoid clogging up the shadow detail. The same thing also happens when you make a CMYK conversion, but the black point will possibly shift inwards even more.

So is it wrong to set levels manually?
Not really. All l I am suggesting here is that it is an unnecessary extra step to use Photoshop to set the black output levels to anything higher than the zero black after you have already set the black clipping to the desired setting at the Camera Raw editing stage (or done so in Photoshop). If you do set the black output levels manually to a setting that is higher than zero you won’t necessarily get inferior print outputs, providing that is, you set the black levels accurately and don’t set them any higher than is needed. And there’s the rub: how do you know how much to set the output levels, and what if you want to output a photo to more than one type of print paper? You see, it’s easier to let Photoshop work this out for you.

Some picture libraries are quite specific about how you set the output levels, but their suggested settings are usually very conservative and unlikely to result in weak shadows when printed to most devices. It is therefore probably better to oblige the libraries with what they ask for rather than fight them over the logic of their arguments.

The only time when you may need to give special consideration to setting the shadows to anything other than zero, is when you are required to edit an already converted CMYK or grayscale file that is destined to go to a printing press, where the black output levels have been incorrectly set. But as I say, if you use Photoshop color management properly you won’t likely encounter such problems.

In conclusion I would say that you shouldn’t need to worry about what the output levels should be. If you are editing a photo in Camera Raw or Lightroom, all you need to concern yourself with is applying the black point clipping that you feel is most appropriate for each individual image. Then let Photoshop or Lightroom take care of the black point output levels conversion at the point where you send the image data to the printer.

About Martin Evening
Martin Evening is the author of Adobe Photoshop CS3 for Photographers as well as the Adobe Photoshop Lightroom Book. Martin is a London based fashion and beauty photographer as well as an alpha 7 beta tester for Adobe. He is also a founding member of Pixel Genius and was the product manager for PhotoKit Color 2.

11 Responses to “Camera Raw and the shadow output levels”

  1. toddv Says:

    While this sounds logical in theory, I find in practice it is still necessary to adjust the black levels of the print. I have tested this conclusion with my particular printer (Epson 2200) using a stepped grayscale chart I created in Photoshop specifically for this purpose. I have tested this with CS2 and found that anything with a value under about 17 (RGB values 17,17,17) to 20 prints completely black with no detail. I would guess that this value would differ slightly from printer to printer, which would account for the differences of opinion mentioned at the outset of your article. The purpose of printer calibration is to account for these differences as well as variations in batches of ink/paper. If Photoshop is supposed to compensate for this I would suggest it is not doing so effectively. I would recommend users test their own paper/printer combination if they are concerned about getting the most detail in their shadows.

  2. Martin Evening Says:

    It could be that you have Black Point Compensation (BPC) disabled using a custom profile. With such profiles, disabling BPC will mean that the shadow tones get clipped in the printing (unless you set the Black output levels point to a higher value and disable BPC) and then you will get tonal separation.

    Another way to look at this is to ask what is the color management meant to do if not to bring about more consistency when converting from one gamut space to another? A profile conversion from the document RGB space to the print destination space should be mapping all the colors from the source to the measured target space including the blacks. Which means we shouldn’t have to add any extra manual tweaks. If we do have to do this, that suggests we are using the wrong print settings or there is something not right with the profile or the calibration of the print device.

    It is interesting to note that in the Lightroom Print module the only options you have are to select the print profile and the rendering intent. I have not asked this question of the Lightroom team, but I have always assumed that the print profile conversions default to using BPC since this is the recommended way to convert from the source space to a print profile. The reason there is no output levels adjustment in Lightroom is that you shouldn’t need one.

  3. Andrew Rodney Says:

    LR uses BPC as it should (there’s NO reason not to use it). And only Adobe Color Engine (ACE) provides this useful functionality.

  4. arne Says:

    Martin, very interesting article! Maybe you have the answer for the following scenario:

    I send my “mass production” photos to a lab which prints them out on a calibrated Fuji Frontier. The lab provided me with a profile which is supposed to work only for softproofing at home, not for converting. They expect sRGB images. For the profiled DP-mode they take a ridiculous amount of extra charge, which I am not willing to pay since this job is not color critical at all… but I want at least some decent shadow tones.

    So I took the provided profile and converted my ProPhoto images to their profile with black point compensation. After that, I converted the image to sRGB without black point compensation. One sees that the black clipping point was moved from 0 to 17.

    So my idea was to batch process my print images with a levels correction in luminosity mode with a output level black point of 17 as a first step for soft proofing after converting to sRGB.

    Can you tell me if that workflow makes any sense or how one would normally set manually the black point to a proper value?

    Thanks, Arne

  5. Andrew Rodney Says:

    >The lab provided me with a profile which is supposed to work only for softproofing at home, not for converting

    So lame. They are trying to convince you they have a clue about color management, they don’t. IF you can’t use the profile to actually convert to the output color space, what’s the use? You can’t soft proof and edit based on whatever rendering intent you’ll pick (which is image specific). You can’t be sure what color engine they will use. In fact, they probably are not using the profile at all anyway. Many Frontiers use a front end processing system that only accepts sRGB and does the conversion, on the fly to the printer without an ICC profile. That makes it fast for them, far less accurate in terms of knowing what you’ll get. There ARE labs that have a color managed front end. This is taken directly from the Fuji Frontier manual:

    When the images are sent to the Frontier for printing from the PIC Pro, the files can be converted to the Frontier output color space (using a color profile or Look Up Table) for that specific paper type, using a setup in the output spooler of PIC Pro. Therefore, when the image is resampled and a raw file is created in the PIC Pro software, it is also converted to the custom color space. If this feature is not utilized, then the images are sent to the Frontier printer in sRGB color space.

    If you have interest in controlling the print process, soft proof and so forth, find another lab.

  6. Martin Evening Says:


    As Andrew says, this is confusing and over-complicated advice that you have been given. Change labs! Soft proofing is valid when working on your system and you can knowingly match the rendering intent you are actually going to convert with. From what I have read of the Frontiera system printers, they have been optimised to work with the sRGB gamut. The gamut of the printer may be greater than this, but the system is designed to accept sRGB as the default and if no profile is present, then assume an sRGB profile.

    I don’t know the supplied profile, but if it is a print output profile then there should be no surprise that it indents the shadow output point during the conversion. All print devices I have seen will do this to a lesser or greater extent, even the LED Fuji Pictrograph device that I used to use. If you analyze your workflow you should see that you didn’t need to tinker with the output levels because the supplied profile was doing this for you. If you are happy to send an sRGB converted RGB image, then don’t adjust. The printer software should be doing the necessary profile conversion and adjusting the black output levels as you have seen happen when you did your profile conversion in Photoshop.


  7. Andrew Rodney Says:

    I’ve built profiles for Frontiers, it can be done and like all output devices, its useful. The issue here is the lab, the front end they use and their unwillingness to supply a profile that a user can use to convert the data, then have it passed directly to the printer without further conversions. Many labs just want to crank out prints as fast as possible. They don’t want to be bothered with profiles or differing color spaces so they ask you to funnel everything into sRGB. It can work, but its not ideal for the customer who cares about controlling the process. And again, the printer doesn’t output sRGB (there’s only one device that can do this, its a CRT display). What’s happening is, the front end is assuming sRGB as the source color space for all conversions to the output color space. But you don’t have to work this way. Labs can and do bypass this as described by Fuji above, using the PIC Pro front end.

  8. KVSSetty Says:

    Being a lab owner I completely agree with what Andrew Rodney says the present generation photofinishing equipment from both Fuji and Noritsu and its print drivers are not completely ICC aware like Photoshop SW.They always assume all the incoming files are in sRGB space and even if you embed a profile it doesn’t care about it and simply assumes the image is in sRGB and converts on the fly to Printers space and prints it.
    To test this follow this simple procedure:
    # open some sRGB image in PS,Assign Pro photo space(please note ‘Assign’ not ‘Convert’ save a copy with this profile embedded.
    #Submit both the original sRGB embedded image and the second ProPhoto assigned and embedded imaged for the lab and ask them to make ‘all normal’ prints (with no manual corrections for color and density).
    You will be surprised to see both prints look exactly same on the print where as when you open the files in PS they look entirely different the Prophoto image radiantly brilliant highly saturated and sRGB image very dull.Why? because PS is ICC aware where as lab printer driver is not ICC aware and assumes all the files are in sRGB.
    So we many labs, always suggest customers to submit files in sRGB though the printers are in fact has much wider gamut than sRGB, labs are not able to take advantage of it.And using a external application (usually third party) like PicPro is the only solution.And from the point of business revenue logistics it is not a viable solution at least in our country.
    The other point is that when machine scans film and save files they automatically convert the files to sRGB and save as 8bit jpegs or 8bit tiffs though they scan at 12/14bits per channel.and the most silly part is that the it doesn’t embed the profile.

  9. arne Says:

    Thanks for all your numerous clarifications!

    Andrew, I’m not shure what you meant with the PIC Pro front end. I cannot find a single website located in Germany that talks about PIC Pro, let alone finding a minilab offering it. So either Germany is a CM-wasteland or you are referring to the “PD/No convert” mode, also described in the third illustration (Camera ColorspaceAdobeRGB) on page 16 in this Fuji manual (1.6MB):

    I also think that my lamelab describes the second workflow illustrated on page 16.

    Martin, as I said before and you also annotated: The color accuracy is not critical at all for this job; it’s a mass print for giving away… though the colors obviously shouldn’t be weird.
    What bothered me were the constantly muddy shadows – no shadow detail at all. Like cut off. That’s why I came up with this maybe strange idea with the levels command.

    KVSSetty, I’ll follow your advice and send different versions of the same image to the minilab… just to see how it will come out.

    Though I thought I had a little clue about CM I obviously need more experience and further reading on this topic. Especially with this unpredictable Frontiers!

  10. Martin Evening Says:

    What bothered me were the constantly muddy shadows – no shadow detail at all. Like cut off. That’s why I came up with this maybe strange idea with the levels command.

    Could it be that the Black Point compensation was switched off when using this measured profile? I have noticed that canned printer profiles are generally immune to leaving the BPC option unchecked when doing a mode conversion in Photoshop. But with other, measured profiles, it is critical to have the BPC checked otherwise the black levels will all clip to solid black. I am guessing that something similar to this could be happening with the printer’s colour management settings.

  11. Andrew Rodney Says:

    >Andrew, I’m not shure what you meant with the PIC Pro front end. I cannot find a single website located in Germany that talks about PIC Pro, let alone finding a minilab offering it.

    Its an optional front end for the Frontier. You’ll have to ask the lab if they spent the money for this option.

Leave a Reply

You must be logged in to post a comment.