Here again, it is a vocabulary that is specific to color management. The conversion is necessary if you want to keep the same colors (L*a*b*) from one device to another. An image is converted when its RGB values are changed to others slightly different in order to take into account the defects of the destination device. If you are new to color management, the most important is to know that Photoshop can take care of everything. But what a satisfaction, if you take the time to learn to do it yourself: your prints, your images on your website will be even more beautiful!
Key points if you are a beginner ...
Here are the key points to remember on image conversion. The rest of this page will be dedicated to those who want to go into more details.
Since you want to keep the same color on your monitor and your print for example, the RGB values displayed on your monitor should be changed to take into account the "defects" of the printer. This modification of values to other RGB values is called a conversion.
This is exactly the same as in an exchange office : you have 100€ in your pocket and you change it for 80$ during the day. You have 100 something then only 80 of something else, but it is the same (the same color in the end !).
This conversion is very complex and fortunately done automatically in Photoshop, if you use the color management tools of the software.
To achieve this, you need two things : a picture WITH its ICC profile and a destination device with its ICC profile (the one of the printer for instance). If one of the two ICC profiles is missing, it is impossible to achieve it. And you will also have to choose the conversion mode which I explain next page...
Since there are larger or smaller color spaces, and since there are also ICC profiles for each device, the problem is now to know how to communicate the right color, the "same color" when it is possible, from one device to another, taking into account their characteristics. The communication of the "right color" and change of the corresponding RGB values is called conversion. Let's see how and why now... knowing that it is kind of similar to a foreign currency exchange office: you have $100 in your possession (as if you were in a given color space, the image's) and you want the equivalent in dollars (to print your file correctly with a given printer and paper) and it is the exchange office and daily exchange rate (the equivalent of the L*a*b* space) that will provide you the equivalence through a conversion... coincidentally, it is also called a conversion when it comes to money.
How to communicate the "right" color?
Let's start with a little bit of history. Over 20 years ago, an international organization, the International Color Consortium - ICC - founded by Adobe, Microsoft, Apple, Agfa, Kodak, Silicon Graphics and Sun invented and installed, first on an Apple computer, a fabulous tool: Colorsync. ICC profiles were thus invented in 1993 - or ICM profiles for Microsoft ® - and conversion tools that necessarily accompany them. But in 1993, the tool Colorsync was only a conversion tool and not an ICC or ICM profile creation tool. They were still to be invented!
In fact, all these ICC creation and conversion tools only show good performances since 2000 for the general public, that is to say since Photoshop version 6 and since companies like MonacoSystems or X-Rite created great profile creation programs at affordable prices and quality sensors. It is thus quite recent!
What is the "right" color?
We now know that the "right" color is in fact a La*b* color and thus the "true" color perceived by a camera for instance. But this "right" color is transcribed in RGB value for obvious practical reasons but since the perfect camera doesn't exist, it is unable to see them correctly directly. So each RGB value supposed to retranscribe these La*b* colors is marked with a little mistake. This mistake is measured during calibration of said camera, and kept available on a little post-it now accompanying this photo and that is called its ICC profile. To display a photo correctly, we saw in the previous page that this ICC profile had to be assigned to this photo. Its RGB values are then correctly interpreted and thus displayed with the "right" La*b* colors.
But we also saw in the previous page that it was possible to face an annoying technical problem. The RGB values that can be measured with the pipette in an image don't match the colors displayed anymore. We took as an example a neutral grey (128, 128, 128) but retranscribed in the image with the RGB values 128, 138, 126. Thanks to the assignment of an ICC profile, we saw that the image then displayed correctly because Photoshop, thanks to the ICC profile, knew that for this camera, a value of 128, 138, 126 should in fact be displayed 128, 128, 128 hence in a neutral grey. But this didn't change the fact that with the pipette, the RGB values showed a predominance. So there's a "logical" problem at this step in color management between what the pipette measures and what the eye sees. The pipette measures a grey marked with a predominance and our eye sees a neutral grey. There's one more operation to go through for everything to be back to "logic". This operation, as you'll probably have guessed, is called a conversion.
What is a conversion and how is it performed?
Convert an image is thus used to change RGB values this time - and not the colors displayed - into other R'G'B' values WITHOUT changing - or as little as possible - its L*a*b* colors so that the colors of a photo are, for instance, printed as accurately as possible, keeping into account the defects of the printer this time.
Let's go back to our example of the previous page - our IT8 test card - dedicated to the assignment of a profile to understand all of this better. We'd opened the image file of this test card (figure a. below) in Photoshop. The image displayed in Photoshop on the left is an image displayed without a profile to understand the whole process. Well not exactly without a profile because Photoshop necessarily needs to assign it one (ideally its own). But without one of its own, it will force the image to use one at all costs and will thus assign it what is called its RGB workspace. If you "just" read the RGB values in the grey parts in figure a (on the left below), they read 98, 91, 87. But in the color space Adobe RGB 98, it does correspond to a reddish dark grey since the R value is predominant. So at this step, our scanner file was assigned an ICC profile by force but since it's not the right one, the image displays weirdly. It is thus time to assign it its own! And when it's assigned the right ICC profile - mon-scanner.icc - everything about the display goes back to normal (figure b. on the right below).
But my image that is now displayed correctly bears a huge handicap. The neutral greys displayed don't match at all the RGB values that could legitimately be expected in the info palette when I place my pipette on it. I see a perfectly neutral grey but the RGB values match a reddish gre (98, 91, 87). I thus have to perform this famous second operation that is called a conversion this time, in order to match the colors displayed with "realistic" RGB values. For that, I need to convert RGB values into other R'G'B' values but this time without changing the La*b* colors displayed.
In figure c below on the left, identical to figure b above on the right, I see my image displayed correctly but still with my scanner's ICC profile. Via the menu Image/Mode/convert into profile in Photoshop until PS CS2 or via the Menu Edition/Convert into profile in the next versions of Photoshop, I'm going to convert the RGB values of this image in a rather broad color space like the well-known Adobe RGB 1998 (figure d.) This time, the display of L*a*b* colors doesn't change BUT the corresponding RGB values change (R'G'B') to become "almost" neutral 109, 110, 110. It is called balancing my image. In the color space Adobe RGB 1998, the RGB values 109, 110, 110 are expressed as a neutral grey since it is a neutral space.
Once converted into a neutral space, I have a visual match between my image and the RGB values.
Now that my image is displayed correctly and in a neutral and broad color space, I can edit it as much as I want - possibly with the pipette.
In fact, each device, each image, etc. does have its own ICC profile that a program like Photoshop is able to interpret. It can assign a profile to an image or "translate" colors, which is called converting an image, from one device to another. Devices are thus able to communicate through the CMM - color conversion engine - and their ICC profile. Since Photoshop version 6, it is called: ACE color engine. It didn't exist before this version of Photoshop (hence a long time ago!).
To work well, the CMM needs to know which ICC profile is assigned to an image - source ICC profile - to know which L*a*b* colors it's dealing with for given RGB signals and to which device it should send it - destination ICC profile -, hence convert it in R'G'B' signals.
The CMM is the central element and is based on the L*a*b* colors (the universal standard) and not on RGB or CMYK signals because, as we've seen several times before, L*a*b* colors are absolute. It knows which or which other L*a*b* color it is supposed to transmit as information (hence absolute) and not each or other RGB value (relative) thanks to the ICC profiles and know which RGB or CMYK signal it matches for this one and this one only. It will thus translate this L*a*b* color, a different RGB value for each device - even if they're quite close, obviously -, into another R'G'B' or C'M'Y'K' signal for the destination device to reproduce the same L*a*b* color. And the best part is, if it can't do it directly, it will replace it with a different one without betraying the perceived visual impression. It's where its main strength lies! This operation of "translation" is called a conversion and there are four different ways to perform it depending on the rendering you're expecting. We'll see that in details in the next page.
When must a photo be converted?
It is necessary to convert an image at least a first time in its history to match what the Photoshop pipette measures and what is given to see. Then, we'll perform or the image will undergo automatically other conversions to change the RGB values of our photo - hence still without changing its colors or as little as possible - to broadcast this image on the internet (sRGB color space), to be displayed correctly on our calibrated screen (this is done on the fly in the graphic card) hence keeping into account its own display defects, or at each print.
Important! When you convert an image into a neutral color space, you need to choose it a bit broader than the camera you used's ICC profile. It's why there are many of them, from the smallest sRGB to the broadest ProPhoto RGB.
What happens during a conversion?
We just saw in the paragraph above the different roles of conversion. When you choose to perform a conversion, what does really happen to the RGB values of my image? How does the conversion tool, the CMM, keep L*a*b* colors as much as possible, that is to say, the visual sensation of the picture, if the target device does not allow it as it is unfortunately often the case with printers? A conversion is a kind of translation which aim is not to alter the original...
Large or smalcolor spaces
As we already saw in the page dedicated to color spaces, devices color spaces or workspaces are more or less wide. Some are so large that they cover all the others (ProPhotoRGB or certain digital cameras). Some are so small that they are included in all the others as for certain printers on matt paper. But sometimes, some are a little bit larger towards a specific color and vice versa for another. This can clearly be seen in the illustration on the right. Globally, the color space contained in the ICC profile of this printer/paper pair is smaller than the gamut of the monitor except towards blue-green colors. These colors are printable but not visible on the monitor (not displayable originally hence displayed with less saturation). This is more often the reverse case you will face, by the way.
In other words, some colors can be contained in the original file, but not be printed. So two scenarios can be observed:
Either original colors are printable (in the destination gamut),
Or original colors are not printable (out of gamut).
In the first case, if the colors in the photo are printable, conversion is only used to change the RGB values of the image to get the same colors on the print. So far, so good! But what to do in the second case or if the gamut of the image is much larger than the gamut of the printer or... the opposite?
Of course not! In fact, they're "replaced" by the closest color(s) - see the next page about conversion modes - hence the most saturated color(s) of this screen. On a print, colors don't disappear either! They're just printed with the closest and most saturated colors that this printer and its inks can manage to produce. So where there was originally a color gradient on screen (for instance), you might end up with a color block, and a less saturated one moreover, on the print. In fact, with glossy papers, it is more often the opposite!
What to do with non printable or out of gamut colors?
Suppose the profile of your image is profile 1 (in red) and your printer's is 2 (in white). Your image has greens and yellows that the printer won't be able to reproduce because they are said to be out of its gamut. No combination of CMYK can exactly reproduce this L*a*b* color in my image. What should I do then? The conversion consists into forceing, these greens into the printer's color space so that they're printed anyway, even if normally the printer wouldn't be able to, as with a shoehorn! But the visual impression must remain as close as possible to the visual sensations of the original image. To complete this shoehorn task, color management and conversion tools use four conversion rules - among which only two are used by photographers:
They are discussed in the next page: relative mode and perceptual mode There will inevitably be losses, but a good rendering engine will reduce them to a minimum (the losses!), without altering the visual sensations of the original image. Photoshop's is particularly powerful if you choose the correct conversion mode. Obviously some image editing programs, much cheaper, also feature edition tools like the clone tool and others of high quality, but none has such a good color management. Unfortunately, this comes at a price...
Notes about color spaces
It is perfectly imaginable to think that a one and only neutral color space such as Adobe RGB 98 could be sufficient to make conversion from the ICC profile of a device towards a neutral color space, but then why have so many been created, more or less broad, like ProPhotoRGB, sRGB? Well, the answer stands in two main and simple reasons :
On the one hand, it is only justified by the fact that digital devices not only do not reproduce all L*a*b* colors but additionally do it in very different ways. However, when a color of a photo - so one seen by your camera - is unprintable, conversion should find a way to recreate the color on your print so that it looks like your original. But, the more differences there will be between the original and the capabilities of the printer, the bigger the risks of having the original colors changed. Color management tools do not deliver miracles! A printer on matte paper does not have the same gamut than on glossy paper.
Secondly, the larger the color space, the more colors it contains, therefore the more the computer work is important. But computers did not use to have the computing power they have today. Do not forget that Colorsync was created in 1995 ...
... And concretely?
Conversion can be performed in three different ways:
Directly in your camera body (in JPEG),
In your demosaicing program (Camera Raw, Lightroom, DXO, Capture One, etc.) in RAW,
In Photoshop (all types of image files).
1 - In your camera body (in JPEG)
The assignment of the ICC profile of your body must have happened automatically in your camera body. Conversion towards a neutral color space like the famous sRGB will also be performed directly in your body, to the confition that you're shooting in JPEG. For that, you'll have to go into the menus of your camera and choose your color space from two options (very seldom, three):
sRGB (by default),
Adobe RGB (or Adobe RGB 98 but it's the same).
Caution! This choice only applies to JPEG files, even if you're shooting in RAW + JPEG.
2 - In your demosaicing program (Camera Raw, DXO, etc.) for RAW files
The assignment of the ICC profile of your body applies automatically at the opening of your RAW file in your program. You have nothing to do and you can't do anything at this stage! If your body isn't recognized by your software because it is very recent or you haven't updated the database of your program, you just won't be able to open your RAW file.
Once opened with the right ICC profile, you need to choose your destination color space, just like on your body (see above), but with an important difference: you can choose from more options among which the very interesting and useful ProPhoto, in certain situations. Example with Photoshop's Camera Raw below:
Right at the bottom of the window, you can choose the destination color space. In Camera Raw, you can choose between the classic sRGB, the bigger Adobe RGB 98 and Colormatch RGB and the very broad ProPhoto.
Important update! Since the update of Photoshop CC to 14.1, it is now possible to choose your destination color space from all these profiles! If you installed them on your computer, you can thus choose DonRGB, Best RGB, Melissa RGB, etc. and not just between 4 possibilities and moreover, you can now choose the conversion mode. (Until then, the default choice was relative, which is quite logical). The situation is thus perfect now and finally leaves us the complete choice of color management in Camera Raw.
When you'll click the button "Open an image", your photo will necessarily have this color space.
A right color (L*a*b*) can be "translated" by a great variety of RGB values, depending on the device. An RGB color is thus not a "right" color but only a color for a given device hence in a given color space.
To be sure to communicate the same color (meaning the same L*a*b* color), two devices must exchange RGB values thanks to a conversion: the RGB values of a color are converted hence translated into R'G'B' values for another device hence for this same L*a*b* color.
So this conversion tries to keep the look and colors of a photo even if some of the colors are unprintable, for instance.
There are two "ways" to solve these issues of unprintable colors: relative and perceptual conversion modes, studied in the next page.
This conversion can also be performed first in your camera if you're working in JPEG, in your demosaicing program if you're working in RAW, and then in Photoshop or your photo editing program.
This site is entirely free of charge and yet almost free of commercial ads. Why? Well because they're ugly! So apart from this banner, you won't find any other commercial spaces. But you have to admit that advertising is pretty profitable and that it enables many websites to keep on living.
So I asked myself the question: what can I do since this website can only continue its growth generating regular incomes, like all other websites. How can I keep on giving access to all this information for free, which I've historically been doing since April 2002 (in french)?
Well thanks to affiliate programs. Meaning that every time you click links from this site towards my partners', Amazon.com, B&H, etc, they know you're referred by me and give me a small percentage on the sale. It doesn't cost you anything! Everyone wins: you, them, me. And also, feel free to contribute here: