Getting the color right (part 1)

Getting the color right – additional steps after calibrating your monitor (part 1)

This is part 1 of a two part article. In part 1 the focus is on camera and printer calibration, part 2 deals with different types of light and setting the white balance in various situations. The goal of the article is to give some practical information and show some examples of the use of profiles to improve color rendering and also shed some light on the choice of WB in situations where non-natural sources of light dominate with different and non-continuous spectra. Both (profiling and WB adjustment) can be regarded as additional steps that may be necessary to obtain acceptable color rendering - after monitor calibration and before you start to edit the image to give it the look you prefer. (NOTE: to view without the need to scroll horizontally, click on the button at the top of the page to the right of the RSS button.)

Contents of part 1

• 1 Introduction
  
  • 1.1 Beautiful is not the same as exactly right
    
  • 1.2 Translation steps
    
  • 1.3 Different languages
  
  
• 2 Adjustment options
  
  • 2.1 Monitor calibration
    
    • 2.1.1 How to calibrate a monitor
    
    
  • 2.2 Camera calibration
    
    • 2.2.1 Do i have to make a camera specific profile?
  
  
• 3 Examples of what can be achieved with a camera-specific profile
  
  • 3.1 Using a dedicated profile to purposely change the rendering
    
  • 3.2 Using a camera profile to mimic the look of another camera
  
  
• 4 Printing profiles
  
  • 4.1 Sending image files with build-in printer profiles to a printing firm?

1 Introduction

Color management can be defined as all the measures taken to get each color represented as correctly as possible in the output - where output can be either paper or screen. Because in digital photography a screen is used to adjust the image as you see fit, the correct display of colors on screen is essential. Correct display of colors in this respect means: technically correct where red is exactly red, green is exactly green and blue is exactly blue. Also colors that result from mixing various amounts of red, green and blue should be represented on screen as precisely as possible, so you can make a reliable judgement whether changes might be necessary. In this article some additional steps are described and examples are shown. Also some remarks are made about getting the right white balance in situations where the human “color constancy system” fails to adapt sufficiently and how results of different lenses and cameras can be matched. As a introduction to the concepts of color management, colorspace and monitor calibration the article written by fellow Dyxumer MichelvA is a good starting point.

This current article is written with both novices and more experienced people in mind, but the description of processes is kept as simple as possible. For the expert that may mean that occasionally things are oversimplified. They may find some more challenging reading in some of the links supplied that point to rather more complex and comprehensive papers. For novices on the other hand, some concepts may yet be somewhat difficult to grasp. For them throughout the article there are also links that point to various sites with additional information.

In this article images are shown within the size restrictions of the forum. Because in most cases several photos are combined into one image, it will not always be possible to see the subtle differences between various types of rendering. Clicking on a particular image in this article will open a new window which will show only that image at a much larger size. Some browsers however may resize the image again to fit within the newly opened window. Clicking on the image in the newly opened window again will then show the image in full size with scrollbars as needed.

1.1 Beautiful is not the same as exactly right
Correct color in the sense of technically correct not necessarily represents the color you like. You may want to change the white balance, contrast, exposure or whatever because you like that better for a particular image. But you can only do so accurately if the screen is really WYSYWIG (What You See Is What You Get) which usually is not the case with an unadjusted monitor. What you want from a monitor is that the colors you see on the monitor resemble as closely as possible as how you will see them when printed.

1.2 Translation steps
The way color is recorded in the camera, the way color is represented on screen, the way the editing program handles the data coming from the image file and the way color is printed on paper all differ. That means that at various stages in the process a translation has to be made. Finetuning the various translations can render much better results then just letting all things pass - and that is the justification for what is called color management: getting exactly the output that you decided you wanted....

1.3 Different languages
If you happen to speak two different languages, you will most likely be aware of the fact that the syntax of languages can differ and that some words or phrases do not have a exact equivalent in the other language. Unfortunately the same is true for the camera, the monitor, your eyes and the printing process. Each device speaks its own language and a translation cannot always be 100% exact for every color. The vocabulary that a particular device is able to understand is described by the colorspace it can use - a three dimensional model of the various colors available. The largest colorspace is used by your eyes; the camera, monitor and printing process all see somewhat differently and in all cases can represent fewer colors then you are able to see when you have more or less standard human vision. You thus cannot expect a screen or print to show colors exactly as you see them in real life with your own eyes, but the various devices working together are trying to get the result as close as possible. Color management consists of seeing to it that the monitor is adjusted as close to WYSYWIG as possible and additionally that all the translation steps involved are carried out in such a way that the output adequately reflects what you saw on your screen.

2 Adjustment options

To obtain the output you want, adjustments need to be made. There are a few things you can do to get better results. The most important step is to adjust your monitor - colors should be rendered as closely to what you would see in real life. That process is called monitor calibration. You can also influence the way your raw-converter and/or editing program handles the data coming from the camera. Every program capable of using those data already has its particular way of doing that, so you need not necessarily bother with it. Sometimes however a program may not do it very well - and then you can make a camera-specific profile that tells the software how to use the incoming data in a better way. The printer also needs to be told how to use the data sent to it (the JPG or TIFF file). Therefore you need a printer/paper combination specific printing profile. Usually the printer software comes with a number of different profiles, also paper manufacturers offer downloadable printer/paper specific profiles. These profiles are quite good and there is not often a need to make up your own. Monitor calibration is a rather straightforward process that can be carried out by anybody who has a basic literacy in using a computer. Making a camera-specific profile that changes the input to the editing program is somewhat more complicated, but also is quite doable for most people. Printer calibration is somewhat more complex – it can be done but may turn out to be rather costly because you need measurement hardware and accompanying software with a pricing not directly aimed at consumers.

2.1 Monitor calibration
A monitor can be adjusted in several ways: brilliance, contrast, color etc. Usually the monitor you have will be designed for general office use. That means that readability is the prime criterion for adjustment and not necessarily the correct representation of luminance, color and contrast of the file which contents is shown on screen. Most monitors are brighter and show more contrast then required and color representation is usually about right but most times not exactly right. Calibration then is the process to get the monitor rendering more or less WYSYWIG - you adjust the way brilliance, contrast, luminance and color is shown. The ideal is to get as close as possible to what you saw in real life with your own eyes. From that moment on you will be able to judge whether the image needs some form of adjustment and you will be sure that the adjustments you make based on what you see on screen, will be adequately carried over to either the file or the print.

2.1.1 How to calibrate a monitor
There are two ways to adjust the monitor. The cheapest way is to compare what you see on screen with some kind of reference. There are various websites that can help you there. It works quite well but lacks the precision needed for critical work. As a start to see whether calibration is needed, it is very useful though. See these links:

http://www.lagom.nl/lcd-test/
http://www.wikihow.com/Calibrate-Your-Monitor
http://www.cambridgeincolour.com/tutorials/monitor-calibration.htm

The second way is to calibrate the monitor with a suitable sensor and accompanying software. The sensor is able to measure the colors that are displayed on the screen and also contrast and luminosity. The process consists of installing the software, linking up the sensor to the computer via a USB port and positioning the sensor as described in the manual. Then you start up the appropriate software. The software guides you through the process - after making a certain number of measurements a monitor profile is generated and the monitor is set to use that particular profile. In the article written by MichelvA you can find a more detailed description of the actual process. It sounds a lot more complicated then it actually is - the software takes out all the guesswork and you end up with a "correct" display. The only thing to remember after that is to not change the monitor settings after calibration and to recheck the calibration at certain intervals after the initial adjustment. How often depends on the type and the amount of use of the monitor - CRT displays will require more frequent checking then flat screens. A professional photolab might start the working day with it, for amateur use every 3 months or so usually will be quite adequate. For more information about the process, the following links may be helpful:

http://spyder.datacolor.com/display-calibration/
http://www.youtube.com/watch?v=P-e0rT_sSuo

2.2 Camera calibration
Unfortunately when you have calibrated your monitor, the story not always ends there. The monitor is showing the colors as they will be printed or visible on screen - but those colors are not nice and are off in some way. You can adjust the white balance and that can improve things quite a bit, but still...some colors are not right, especially skin tones and other more subtle colors may still not be what you want. The reason for that can vary: the camera does not record those specific tones very well, a very different kind of light was used etc. What it comes down to is that for some reason or another the camera is not able to supply the correct data needed for the editing program.

Let's look at a example. Suppose a specific tone is specified by the camera as 25% red, 10% blue and 56% green. In reality the color was 29% red, 15% blue and 40% green. If the file with that data is read by the editing program, it will render the color as 25% red, 10% blue and 56% green - and on a calibrated monitor it will be shown in that color. But that was not the color of the object shot! Normally the raw converter or the editing program will "know" the camera. The raw converter knows about the shortcomings of each camera (hence the software needs to be updated when a new camera hits the market, although the file format remains unchanged!) and tries to compensate for that by using some extra camera-specific information to put things right. The way that is done is designed by the software vendor. And thus it may differ depending on the software you use. Most times it works out quite well, but if you are not happy with it, you can try to improve that process. That is where a camera-specific profile made by yourself comes in. The process of making a camera-specific profile is called camera calibration. That is somewhat misleading - nothing is actually changed in the way the camera works. We only tell the rendering software to apply certain corrections on the data from the camera, as a alternative to the standard settings that we regard as inadequate.

For Adobe products there are two ways you can do it. First you can shoot a color chart where lots of colors are represented and compare those colors with what you see on your calibrated screen. Adobe offers a free program (DNG profiler) that can assist you to generate a profile that, when used, can improve the color rendering. Unfortunately use of that program is rather tedious and time consuming. Another way to do it, is to buy a ColorChecker Passport from X-rite. You then shoot a image of a testchart, feed the data obtained to the software that came with the chart and then the software subsequently generates the profile for you and integrates it into Adobe ACR/Photoshop, ACR/PSE and/or Lightroom. The software thus avoids the use of the DNG profiler and use of it does not require any specific knowledge. It works quite fast – in less then five minutes you can have a tailor made profile! For other raw converters you can also make a camera specific profile. In most cases the procedure is somewhat different and more complicated though. The links below offer more information on the various ways of getting a camera profile:

http://www.gapalmer.co.uk/WBC/DNGProfile_EditorDocumentation.pdf
http://www.youtube.com/watch?v=IrEtS3BnNkU
http://xritephoto.com/ph_product_overview.aspx?action=support&id=1257
http://www.cmp-color.fr/LR_couleur.html (french text)
http://www.cmp-color.fr/raw_lineaire.html (french text)

2.2.1 Do I have to make a camera specific profile?
Not unless you feel the need to. If you are happy with the results you get, there is no reason to. If you are not, it may improve the way colors are rendered. Improvements with normal lighting will usually be minor, but subtle. When shooting under quite different and unusual lighting conditions or when using lenses from various manufacturers it may well be worth the effort. This video shows how it can be done - the hard way.

http://www.youtube.com/watch?v=IrEtS3BnNkU

3 Examples of what can be achieved with a camera-specific profile

Now let us see what we can do with a camera-specific profile. In image #1 various renderings are shown:

Image #1:



The image was shot indoors with mixed lighting: halogen, energy saving lights, tungsten and fluorescent – all in the same room. The shot was taken with a NEX 7 and the 24/2 CZ E-mount lens. The same image is shown five times. All renderings were done with Lightroom with the preset all zeroed and different renderings for the WB and the profile used:

1. Original image, AWB, as shot, profile: Adobe Standard
2. Custom white balance based on a white part of the magazines on the table, acquired with the eyedropper tool, profile: Adobe Standard
3. WB: Lr fluorescent, profile: Adobe Standard
4. WB: Lr tungsten, profile: Adobe Standard
   
5. Custom white balance based on grey patch of Color Checker Passport, profile: dedicated camera profile

AWB does not work well in this situation, and neither does a WB reading based on the white parts in the magazines on the table. The tungsten setting is better. But it can subtly be improved upon when using a camera-specific profile especially made for this particular shot. In the next picture (image #2) shot outside, you can also see some subtle changes. The image was shot with the a99 and the 50/1.4 ZA SSM lens.

Image #2:



Again the same image was rendered all zeroed by Lightroom 5 times, with varying settings for WB and profile:

1. Original image, AWB, as shot, profile: Adobe Standard
2. WB: Lr Auto, profile: Adobe Standard
   
3. WB: Lr Cloudy, profile: Adobe Standard
   
4. WB: Lr Shade, profile: Adobe Standard
   
5. Custom white balance based on gray patch of ColorChecker Passport, profile: dedicated camera profile

In image #3 the original and the final rendering are shown at a larger scale. The image on the right, based on the specific camera profile, looks cleaner, partly because of the changed WB, but also because the color of the foliage and the sky has been changed due to the use of a dedicated profile.

Image #3:



The testchart shot to acquire the profile can be seen in image #4. The differences are quite subtle, but nonetheless noticeable. The original rendering is shown on the left, the rendering with the tailor made profile on the right.

Image #4:

3.1 Using a dedicated profile to purposely change the rendering
One interesting possibility is to make different profiles for different lenses that you use. In image #5 a example is shown for a few 50 mm lenses that are again rendered by Lightroom with all settings zeroed and with the Adobe Standard profile.

Image #5:



All images were shot with the a99, ISO 100, AWB, A-mode set to f/11 and manually set to f/11 on the Summicron. Focusing was done manually with the help of focus peaking. The shots were made within less then a minute from each other from a tripod. Lightroom reported the automatically camera-set WB as: 4500/+2, 5100/-1 and 5400/0 (temperature/tint) for the Minolta, Summicron and Sony/Zeiss ZA.

All three shots are acceptable for a snapshot – the Summicron image to me looks about correct as far as the WB is concerned (cloudy afternoon in late November, about to start to drizzle). Although AWB was set by the camera, the Summicron is rendered notably cooler. That may be because Lr was not able to detect the lens (and thus select a appropriate profile) from the EXIF data, although the camera itself was of course recognized. From experience though I find Leica R lenses in general notably cooler then Minolta, Sony and Sony/Zeiss lenses. The differences between the ancient Minolta and the new Sony/Zeiss are much smaller, although the WB settings reported differed quite a bit. It is clear that there can be quite visible differences. That can be annoying when using different lenses in a photoshoot, if the images later have to be combined into a slide show, screen gallery or when printed.

Image #6:



In such a situation use of dedicated camera profiles (based on the particular camera/lens combination) can help out. In image #6 the results are shown – a Leica Summicron can do Minolta colors!

All images now have the WB based on the light gray patch of the testchart for profiling and for each image a dedicated camera profile was used – actually a camera/lens profile in this case – the profile was made for the specific combination. The results are quite visible: color rendering is much more uniform now. The use of a camera specific profile to synchronize the color rendering of lenses from different manufacturers may well be the most convincing and useful application of a tailor made profile. You may like or not like the final rendering. The WB is based on a reading of a gray card and thus technically should be correct. I personally find it a bit to warm - I did not see it that way when shooting. The colorchart was shot about 40 cm from the camera and the light there may have been somewhat different then the light on the building. It may also have something to do with how human vision works. We will discuss that later in part 2 of this article.

3.2 Using a camera profile to mimic the look of another camera

Cameras do differ in the way they render color, because the underlying software and hardware are different. Even different camera types from the same manufacturer will differ slightly. Thus, the rendering of a particular Canon, Nikon, Leica or Sony camera will differ. Not that much, but enough to get noticed when the same scene is shot. Software vendors tend to reduce the differences, but still shots usually differ. They tend to do the best conversion possible - which may not always be equivalent to doing it in such a way that perfect color rendering is achieved. You can use a camera-specific profile to obtain a particular rendering resembling what you would get when using another camera. DXO has incorporated that feature in their conversion software - you can have your Sony images being rendered as if they were shot with a certain Leica, Canon or Nikon camera if you want to. That can be a useful feature if your images end up in a pool of images shot with varying cameras. Some more information can be found here:

http://www.dxo.com/intl/photography/dxo-optics-pro/features/color-controls/color-rendition-profiles

In image #7 a example is shown. The image was shot with the a99 and the 70-200/2.8 G SSM. Of course the differences are subtle, but differences are notable in the way shadows are rendered and also in the rendering of various primary colors.

Image #7:



Adobe does not offer this feature, but the more recent versions of Lightroom and ACR usually do come with several profiles for more recently introduced cameras. For the a99 for example you have a choice between Adobe Standard (what Adobe thinks to be most suitable) and various other settings that more or less replicate in-camera JPG settings like Camera Clear, Camera Vivid etc. They can be regarded as a easy way to get a specific JPG-look.

4 Printing profiles

When printing, there also has a translation to be made. The data contained in the JPG or TIFF file send to the printer need to be converted into data the printer can handle, because the printer uses a different colorspace and color is not based on mixing red, green and blue but on mixing cyan, magenta, yellow, black and often still more inks. When you buy a printer, it comes with software that gives you the possibility to choose a particular profile based on the printer and the paper used. Those profiles are usually quite adequate. And if you send your files out to a firm that does the printing for you, that firm will take care to use the correct settings. When you are unhappy with the result though, you can try to "roll your own". You can buy sensors that are able to measure the printed output and compare the output to what it should have been and then generate a specific profile. The whole process is a bit more complicated then calibrating a monitor and the hardware and software needed to do it do cost quite a bit - much more then what is needed for monitor calibration. And as said - usually there is not much to be gained. But you can, if you want to.....

You should be aware that the contrast ratio you can achieve in print will be smaller then what you can get on screen. You cannot change that. Adobe software comes with a "proof setting" that gives you a idea how a print will look - somewhat flatter than on screen. It may show clipping in areas where in the on-screen view there is no clipping. You may need to increase saturation and contrast of the file somewhat before sending it to the printer and maybe reduce highlights. On a normal screen the image may look overcooked, but if it looks right when the view is set to "softproof" you get a idea of how the print will look.

4.1 Sending image files with build-in printer profiles to a printing firm?
If you send out a file to a printing firm to be printed, you could consider to also send the home grown profile you developed or a profile obtained otherwise with it. That way you could force the printing firm to use a specific profile. In most cases that is not such a good idea, unless you know exactly what you are doing and have communicated with your printing firm about it first. If you force them to use "your" profile and the results are unsatisfactory, bad luck - your fault. If you let them decide which profile would be appropriate and the results are not good, you can ask them to do it over - at their cost. Usually professional printing firms know a lot more about the correct settings then the average photographer when using a particular combination of printer and paper, so let them do what they know best.

Go to Part 2




Edited by romke - 06 December 2013 at 11:48

Great work Romke, excellent article and a very good addition to the current KB articles.

Thank you very much for this !!
Both parts, are great and time spent on this contribution to the community is highly appreciated !

Romke,
Excellent! I have had trouble understanding most articles on this subject in the past. You have greatly clarified things. The examples are very useful.
I know I will be returning to this well-written article in the future.
Thanks so much,

Mike

GlassEye wrote: Romke, Excellent! I have had trouble understanding most articles on this subject in the past. You have greatly clarified things.

I can't think of better compliment than that
So once again, great job Romke !

quick question on using the ColorChecker Passport... so suppose I"m using it for portraits, I first shot with a model holding the color pallet next to her face... and then enter that profile into LR5, but then in the same shoot (same lighting everything, no change in lighting), but I have a different model with different skin tone, does that make any difference? I ask because in a few YouTube videos I've watched, they emphasized having the color palette be held very close to the model's face... and again... what happens when a different model is involved in the shoot? Can you still use the original custom profile? Or you need to re-shoot another picture with the color palette with 2nd model?

trainerKEN wrote: quick question on using the ColorChecker Passport........ Or you need to re-shoot another picture with the color palette with 2nd model?

The idea is to position the ColorCheckerPassport in such a way that it is lit by the type of light that hits the main object in the frame, so the correction can be made for just that specific type of lighting. Therefore in a portrait the CCP should usually be positioned near the face.

It does not matter how that face looks, there is no difference between a caucasian or a native african or asian face because the lighting would be the same (the exposure might well slightly differ between them).

If however you decide to change the lighting setup (adding flash, reflectors etc to the ambient light or just use flash only in a dark environment, you will however need to make a new reference shot of the CCP.

Whether it is worth the effort depends on the camera and the quality of the software used. Rawconverters do slightly differ in the rendering of colors relative to each other - and they may also improve over time. Each and every rawconverter handles the problem of converting the information in the raw file somewhat differently - what they all do is correct a certain calculated RGB value in terms of R+4%, G-2%, B+12% where the percentages will vary across cameratype and converter. By "feeding" the rawconverter a set of exact known colors that type of calculation can be improved upon - and that is in fact what the CCP does.

The same principle can be applied the other way around. If you find out what a certain cameratype does "wrong" you may be able design a type of conversion that lets other camera's make the same "mistake". In DXO that feature is incorporated. From their starting point (the way they think the colors of a specific camera should be rendered) it is possible to dial in another rendering, making the images shot with a a700 look like they were shot with a specific Leica, Nikon or Canon camera....