Image conversion : perceptual and relative colorimetric conversion
It is the heart of color management when you want to print photos. those are therefore very important concepts. Behind these words hide two printing options. You will have to choose one of those two possibilities, before you print your picture because the rendering may be significantly different BUT it depends a lot on your photo. In most cases, the differences will be imperceptible. Let's have a look at this ...
Key points if you are a beginner ...
Here are the key points to remember on conversion modes. The rest of this page will be dedicated to those who want to go into more details.
How many conversion modes exist ?
There are four different conversion modes, but only two of them interest us :
What is a conversion mode for ?
During the conversion of RGB values of a photo to R'G'B' values of the destination device (eg Photoshop to the printer), it may exist colors from the original image which can not, for example, be reproduced. The printer does not have enough saturated ink. What should be done with these colors? Whereby change ? The two modes of conversion will take care of this but in two different ways.
In relative colorimetric conversion mode
In this mode, all original unprintable colors, because too saturated in the source image, are replaced, by the printer, by the nearest and most saturated ink color printable. All other colors do not change.
What is the consequence ? Imagine you have photographed a beautiful red dress using a flash. The red will be very saturated. The camera has photographed all the nuances of the drape. However, the printer does not know how to print all these so saturated nuances. In relative colorimetric, all these reds will unfortunately printed with the same red. The dress is going to lose a big chunk of its nuances, so its drape. The photo will be flat.
In perceptual conversion mode
This conversion mode tries to keep all the drape of the dress, so will print all shades of red but with less saturated colors, at least the more saturated of the printer.
What is the consequence ? always with our dress, the drape is kept here, but all colors are changed. The general perception of the image is preserved but some other perfectly printable colors will also be desaturated or slightly modified. That's why it is called perceptual. This mode keeps the drape, but at the color truth cost.
Then, which one is to be chosen ?
Firstly, there is no ideal conversion. Each image has its most suitable conversion mode. If you know how to do it (and I explain it on my page dedicated to print with Photoshop), you know that you can choose the conversion mode for your picture BEFORE printing. If you hesitate and as it works very well in most cases, I recommend the relative colorimetric mode. This is the conversion mode by default of Photoshop and that is fine with me.
Next key point : what is calibration ?
We have seen on the previous page the different roles of conversion. When we choose to perform a conversion, what is really happening at the RGB values of my image ? How the conversion tool, CMM, offers the best La*b* colors, that is to say, the best visual sensation of the picture, if the targeted device does not allow it as it is unfortunately often the case with printers ? The conversion is a kind of translation which aims not to distort the original ...
Why is there a need for two conversion modes ?
As seen on the page dedicated to color spaces, color spaces of devices are more or less wide. Some are so large that they cover all the others (ProPhotoRGB or DonRGB). Some are so small that they are included in all the others. But sometimes, some are a little bit larger for a specific color and vice versa for another. This is clearly seen in the illustration on the right.
Here, the color space contained in the ICC profile of this printer/paper set is smaller than the gamut of the monitor except in the blue-green colors. These colors are printable but not visible on the monitor. This is more often the reverse case you will face.
In other words, some colors may be contained in the original file, but be not printable. So two scenarios are 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 serves only 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 ?
What should be done of non printable or out of gamut colors ?
Suppose the profile of your image is the profile 1 (in red) and that of your printer the profile 2 (in white). Your image has greens and yellows than the printer can not reproduce because they are said to be out of gamut. No combination of CMYK can exactly reproduce the color Lab of my image. What should I do then ? The conversion is to force, as with a shoehorn, the greens into the printer color space so that they will be printed, even if normally the printer would not know how to do so ! But the visual impression must remain as close as possible to the visual sensations of the original image. To achieve this task, the color management and conversion tools use four conversion rules - only two of them are used for photographers :
They are discussed below. There will inevitably be losses, but a good rendering engine will reduce them, without altering the visual sensations of the original image. The Photoshop one is particularly powerful if you choose the correct conversion mode. Obviously, some image editing softwares, much cheaper, are also having edition tools, such as clone tool and others of high quality, but none has such a good color management. Unfortunately, this comes at a price ...
Rendering modes : relative colorimetric or perceptual
When a conversion is performed in Photoshop, you need :
An image with an ICC profile;
To Choose the destination device ICC profile - your printer and your paper -
But also, as seen, the rendering mode (Rendering intent).
What is it ? Just how colors are managed out of gamut when converted. Are we deleting colors out of gamut - destructive but logic - are they replaced by other colors but which one ? What happens to the colors used for the replacement ?
Note on conversion
The conversion takes care of changing the RGB values - origin to destination - and the rendering mode (Rendering intent) handles to know what we're doing of out of gamut values - preserve, remove, replace or change -.
Let's have a look at it now ... According to the rendering mode selected, the conversion rules will change. There are four of them : perceptual, relative colorimetric, absolute colorimetric and saturation. Best suited to photography is very often, in my opinion, the relative colorimetric mode, but the most "logical" because supposedly more linked to photography is the perceptual mode. I will now describe them.
An example may illustrate the problem : Imagine that I want to print an image. It has the profile of my workspace - red line on the illustration on the right - and I know that my printer is much smaller for the greens and yellows - white line -. Colors of my original, around the point A should be unprintable - Lab color associated is not printable. Never the less, I want to try to print these colors, so I have to convert my image into another space, the destination - inside the white line - to read the right RGB values BUT HERE THE RIGHT RGB VALUES DO NOT EXIST AS NOT PRINTABLE. In the case of the point B which is common to both areas, and the Lab color associated belongs to both spaces - the LAB color can be converted into other RGB values.
Under these conditions, how to print those yellows and greens of my image that are outside the reproducible area of the printer and not completely distort my image when printed ? What is the printer doing of colors out of destination gamut ? Are they just deleted ? In addition, on my original image, the distance between A and B points gives me a certain visual "sansation", how to keep it after the conversion ?
But there are several problems :
1 - If greens out of the color space, so normally unprintable, are placed in the space of the printer - inside the white line - many colors will actually be "overlayed". Perfectly printable colors that belong to my original image, would be replaced by initially unprintable colors. One might think then that the only option open to us is to keep the color B and eliminate A. In other words what does one decide to keep and eliminate?
2 - Also, if I put the color A on the color B and red line colors on those of the white line, it will be impossible to differentiate them, although they were different in the original space.
3 - Finally, I remind you that the yellow-green point A was at a distance of yellow-green point B in my original image, which gave me some visual sensation.
To solve these problems, mathematicians have invented conversion rules they called rendering intents.
In perceptual rendering intent
The ACE conversion engine of Photoshop will work hard ! It will indeed be placing the greens out of space, so not to completely remove the nuances and keep my visual impression (my visual perception). To make a conversion respecting the visual sensations of the original image, the ACE conversion engine of Photoshop will proceed in two steps. It will first of all "set" the colors of the red line (the limit of the space of the image) on the white line (space limit of the printer) as the colors of the white line are the closest visually of the red line colors. The colors of the red line are the most saturated colors of my original image and the white line colors are the most saturated colors of my printer. Then, to keep a close visual sensation of the original image, it will slightly and gradually "shift" colors within the space of the printer as shown in the figure above. B will shift to B' and so on ... B being "liberated", Photoshop can put an out of gamut color in its place. As this cannot be done indefinitely, without having colors disappearing on the "other side" (red colors side), it is wiping out a few nuances of the original image in the gamut as it "pushes" colors inwards, while trying to retain the same visual perception. That is why this mode is called perceptual. The image will lose some of its original colors but the visual sensation remains very close to the original image.
Summary : The red line goes to the white line. Color A disappears because unprintable and is replaced by A', the nearest printable color. A' is replaced by A''. From time to time out of gamut colors really disappear because the volume of color of the red line can not physically go into the white volume. The relative distance between A and B - A 'and B' therefore being smaller - is conserved as much as possible. That is why an image should be worked as long as possible in the original color space. This is the only way not to lose information. Conversions, in this rendering mode, are destructive because all colors are changed, even those that were printable. Conversions need to be limited in number and done at the very last moment. Finally, the converted image will appear slightly desaturated and slightly less contrasted, if the color space of the printer is really too small for these colors. Then simply apply a level layer or contrast layer to find - almost - the Lab color from your original image.
In relative colorimetric rendering intent
The ACE conversion engine of Photoshop works much less because it is then simply a matter of removing all colors out of gamut ! The Lab colors shared by the two color spaces remain unchanged. So, if the colors out of gamut disappear, other colors are not changed at all, which was not the case in the perceptual rendering intent. This mode is therefore perfect when the original image contains very few colors out of gamut or when a printable color should not move at all.
How to choose between perceptual and relative colorimetric ?
The most important thing is that there is not a better way of rendering than the other, it depends on the photo to convert and more precisely of Lab colors contained in the file. Reminder : in nature, the colors are not always very saturated and that is why the choice of working only in sRGB is not "catastrophic". While it is fashionable to work in larger color spaces to show that you know a lot about color management, I find even smarter to stay in sRGB in most cases because the sRGB color space already contains enough colors, to be only rarely forced to work in a larger color space. Each picture taken does not contain millions of colors and all photographers do not take pictures of the turquoise seas or dresses of designers with intense reds !
So if your original image has little or no out of gamut colors for your printer, they will not be "lost" by a relative colorimetric rendering intent ! It is then the most appropriated mode because it does not change - so do not deform - the colors within the gamut. Coincidentally, this is the one that works best in most cases, which is perfect ! Even if you work in ProPhotoRGB, it is surprising how this rendering mode often works very well. Your image was not necessarily containing a lot of colors out of gamut for the printer, whose gamut is much smaller.
If your original image contains a large amount of unprintable colors, then it may be wise to use the perceptual rendering intent. Certainly it changes all the values of your file but it has a much better preservation of some materials, of some relative distances between two colors as explained above.
How does it get actually materialized ?
Where the relative colorimetric rendering intent could transform unprintable colors into a solid color without any nuances since all colors out of gamut are "deleted" or in other words replaced by a single color, the closest printable color, the perceptual rendering intent tries to keep those nuances, even if it has to replace them with others, very similar visually. I'll try to show you, I hope, with the illustrations below !
One point is very interesting : if you look at the whole picture EXCEPT the circled area, you will see that the picture does not "move" between the original, the perceptual and relative colorimetric rendering intents. The "movement of colors" will take place for some colors only !
In this example, you should have a look at the yellow color of the wall on the left, the twilight colors of the sky near the horizon and finally, the colors in the circled area. On the original (1) the sky is saturated with shades of blue, more or less dark and more or less saturated.
The converted image with the relative colorimetric rendering intent (2) show the disappearance of many nuances in the sky - circled area - while the rest of the image remains unchanged. The blue out of gamut color of the sky in the circled area cannot be kept the way they were. And where there were nuances in terms of saturation and brightness, there is a flat area !
On the image converted with the perceptual rendering intent, we are less moving away from the original. While the blue sky in the circled area is also less saturated than the original, it retains brightness nuances as on the original. The result is closer to the original image. The rest of the photo in the common gamut is preserved.
- There are two color conversion modes. The Relative colorimetric and the Perceptual rendering intents.
- There is no ideal mode. Each picture has its best conversion mode.
- If your photo have really saturated colors (for instance the red dress drape) it will be a good idea to verify before printing that the perceptual rendering will deliver better results than the classical relative colorimetric rendering.
- Photoshop and Adobe Suite, by default, use the relative colorimetric rendering intent and in most cases, it is just perfect that way. You do not always photography highly saturated colors.
Finally, the order in which we will calibrate our color workflow is very important. You have to start with your monitor calibration. Then you will deal with your scanner (if needed) and your printer calibration. But before going any further, I would like to spend some time on defining some more words : calibration and characterization