I'll try to summarize what I've learnt about soft-proofing in this forum over the last few days (and again, many thanks for your help!). What I’ve written here is based on my current understanding and may not be correct – if there are errors please let me know and I’ll correct this summary.
Converting the Document for Printing (or viewing on the web)- Conversion from working space to working space: This will always be Relative Colorimetric (even if Perceptual, Saturation and Absolute can be selected). In fact, conversions using a profile that is matrix-based (as are the working spaces) always use Relative Colorimetric.
- Perceptual: The entire gamut is remapped so that even if all the colors in the document are within the destination gamut, the colors will all be shifted. The objective of this transform is to maintain the relation between colors, at the expense of color accuracy.
- Relative: Out-of-gamut colors are remapped to the nearest in-gamut color (on the gamut envelope). This may involve hue, saturation and or lightness changes, depending on the profile algorithm used (one of the things that distinguishes a good profile from a bad one is how well this mapping is done). The black and white points are mapped to the new colour space, so that the greys appear to remain grey even if this leads to some shifting of colours within the rest of the image. The profile attempts to preserve the relationship between colors given the white-black point shift. Colors nearer the grey line are shifted more, colors further from the grey line shifted less.
- Absolute: Exactly the same mapping is used as in Relative (the two share the same mapping tables) so that both gamut mapping and adjustment for the D50-normalized paper color occurs. However, all the colors are shifted so that they will retain their absolute value when viewed under the target illuminant (in fact, what happens is that the grey line is not adjusted, so that if the target illuminant temperature is warmer than D50, the print greys (and other colors) will have a blue tint). At any rate, the effect is that if the target illuminant has a very warm temperature of say 2800K and the print is viewed under a 5000K light, it will appear very blue. If the target illuminant is 5000K and the print is viewed under the 2800K light, it will appear very yellow. The idea is that when viewed under the target illuminant, the colors will retain their PCS Lab values. This rendering intent is rarely used in photography and for this reason it isn’t available in Lightroom.
- Saturation: This attempts to preserve highly saturated colors. It’s normally of no interest in photography unless graphic-like effects are desired.
- Black Point Compensation: This rounds the transition to the destination black. If BPC is not used, the transition will be abrupt, so that colors will suddenly get clipped to black as they get near the output black. With BPC on, the colors will gradually get squashed to black as they get near the output black. This avoids areas of flat black in the print.
- Paper Black: This is the blackest black that can be printed on this paper with the inks used. The paper black is also measured during profiling. Colors darker than this will be clipped to black. Black point compensation uses the paper black figure.
- Paper White: The paper white is measured during profiling and normalised to D50. The paper white is both the tint in the paper and the brightness of the paper (when viewed under the standard illuminant). The intent mapping tables are built to map the image to this white.
- Illuminant: The default illuminant is D50. Unless the profile takes the illuminant into account, D50 is assumed … in other words it’s assumed that we will view the print with a lamp which has a D50 spectrum (this is a daylight spectrum which can’t be achieved with artificial lamps). D50 has quite a flat spectral power distribution, so that all colors are equally well excited or represented. Lamps like 2800K tungsten are weak in the blues, a bit weak in the greens and very strong in the reds: with lamps like these, without compensation the blues will look darker and the reds brighter with greens a bit darker. Colors with combinations of the colors will be affected proportionally, so yellows , which are a combination of greens and reds will be brighter, and cyans, which are a combination of greens and blues will be darker. 6500K fluorescent lights have more power in the blues and less in the reds and they also have spikes along the spectrum due to the phosphors used. With this sort of lamp the blues will be brighter and the reds darker, and certain colors along the spectrum will be bumped up (by the spikes in the spectrum). Programs like ArgyllCMS have an option to measure the target illuminant and to incorporate the illuminant’s spectrum in the transform. Assuming that this is done well, it will improve the accuracy of the colors in the print for the target illuminant (in other words, it allows us to target the print to different lighting conditions like incandescent, fluorescent etc.). However, if we do not know what type of light the print will be viewed under it’s better not to use illuminant compensation as it could make things worse, for example if we target warm tungsten lamps and the print is actually viewed under 6500K fluorescents.
- Optical Brighteners: Optical brighteners are chemicals added to the paper and sometimes to the ink. These shift ultra-violet light to blue light, so the print looks ‘whiter’ … or bluer, in reality. As with illuminant compensation, programs like ArgyllCMS can attempt to adjust the print so as to counteract this effect … which is not something one would normally want to do as it would be defeating the ‘brightening’ effect. (In proofing, we normally want to be able to compare the print and the proof side by side, and so we may wish to eliminate the effect of the optical brighteners in both print and proof).
- Chromatic Adaptation: The eye has the ability to compensate for the illuminant when viewing paper. So we see the paper white as white, even though it may have a yellowish tint and even though the lamp may have a very warm color. This means that on the whole prints will look OK with very different lighting. However, it’s important to check the print under the possible lighting conditions that it will be viewed under because different lights, with different spectra, can cause colors to shift relative to one another, and the eye cannot compensate for that.
Soft-ProofingSoft-proofing uses whatever intent is chosen (Perceptual, Relative …) and also Absolute when Simulate Paper White is selected.
- Converting for Print: The image is transformed (to a temporary soft-proofed image) using the chosen intent, exactly as above, so that the out-of-gamut colors are clipped and in-gamut colors are re-mapped.
- Black Point Compensation: Again, the soft-proofed image rounds the black transition if BPC is selected.
- Simulate Black Ink: This effectively lightens colors which are below the Paper Black to give an idea of what the print blacks will look like. It works with Black Point Compensation.
- Paper White: The soft-proofed image will be compensated for the paper white as for a normal conversion.
- Illuminant: If we know precisely what light the print will eventually be viewed under and if we can view the print with the exact same light, then we should use a profile that has illumination compensation and we should soft-proof with a profile that does not (same profile for same paper, one with illumination compensation, one without). If we do not know what light the print will be viewed under, then we should do the opposite: print without illumination compensation, soft-proof with. In either case the print and soft-proof should match. The reason why it’s better not to print with illumination compensation is that the compensation is only for a particular lighting condition: if the print is then viewed under different light, the compensation could make things worse, not better.
- Optical Brighteners: Same as for the illuminant. Unless we know the target illuminant and can view the print under it, we should print without compensation and soft-proof with.
- Simulate Paper Color: The soft-proofed image is converted back to the working space using the Absolute Colorimetric reverse table (that is the table that converts back from the destination colors to the PCS, or Profile Connection Space). The effect of this is to simulate the paper white because all the colors are shifted from D50 to the paper white under the target illuminant (so if the target illuminant is warmer than D50, the soft-proof will also be warmer). Since out-of-gamut colors have already been mapped in the first transform (image to destination) there will be no out-of-gamut colors in the Absolute conversion, so the colors should accurately reflect the printed colors, viewed under the specified illuminant. However, unless the white point is changed in the output profile, the paper white is normalised to D50, which means that the soft-proofed image will only match the print if the viewing light is very close to D50 (or unless the monitor white is changed … see below).
- Monitor Gamut: It’s very important to bear in mind that we are viewing the image on a monitor and that this has a different (and possibly smaller) gamut than both the working-space we are using and the destination (print or web) space. So there may be colors in the image that we cannot see, some which can be printed, and some which will be clipped on output. Lightroom has a useful feature in soft-proofing that allows out-of-gamut colors to be highlighted both for the monitor and for the destination. In Photoshop, it’s necessary to soft-proof to the monitor profile to see monitor out-of-gamut colors.
Viewing a Soft-Proofed ImageThe whole objective of soft-proofing is to enable us to cut out the test print and go straight from the image viewed on the monitor to the final print, with considerable confidence that the print will look the same as the image on the monitor. It should allow us to see what happens to out-of-gamut colors, how in-gamut colors are shifted … and ideally what the print will look like under specific illuminants, like a tungsten lamp or a fluorescent lamp. This is a big objective, and to see if we come close to meet it, the best thing is to make some test prints and then compare them to the image on the monitor. We can do this in two main ways:
- Viewing separately: The soft-proof is looked at, some time allowed looking away from it, and then the print is looked at. The monitor should be at or near 6500K as this is the optimal viewing temperature for monitors. Simulate Paper Color should be left off, otherwise the image will look very yellowish (assuming that the viewing light is well below D50). What we are depending on here for comparing the soft-proof and the print is visual memory; allowing some time between viewing the soft-proof and the print gives time for the eye to adjust to the two different lighting types and temperature. In this case we are not attempting to see what the print will actually look like under a particular illuminant.
- Viewing Side-by-side: This is the ideal. However it is only ideal if the soft-proof and the print look the same when viewed side-by-side! It is possible to get this to work properly, but as it isn’t so easy I’ll cover it under its own heading below.
Side-by-side Viewing of a Soft-Proofed ImageWhen we look at a print, we see the paper white as being white, even if it is actually an off-white. The color of the paper depends on the illuminant, of course, so this compensation is for both the paper color and for the color of the light. It is called chromatic adaptation. Unfortunately the same adaptation doesn’t occur with a monitor display, so if the ‘white’ on the monitor is yellowish, we see it as yellowish. This means that an image on the monitor which is in absolute terms identical to a print may not look the same to us. An instrument may see the two as being identical, but we do not.
However, it is possible to fool the eye into thinking that the monitor image is actually a print, and then chromatic adaptation seems to work. If we view a print side by side with the monitor image so that they can be seen by the eyes
at the same time, then the eye thinks they are both prints.
Furthermore, if a sheet of plain paper (the same paper that will be used to print on) is viewed side by side with the monitor image, under the target illuminant, then, again, the eye sees the monitor image as though it were a print being viewed under this target illuminant.
This means that if we set up our viewing environment carefully, that we can soft-proof an image on our monitor and see the soft-proof as though it were a print. This is the subject of many research papers and it isn’t cast-in-concrete science as there is as much psychology as biology involved, but it does seem to work for me.
There are two ways to set this up.
- Adjust the Monitor: This requires software like the NEC Spectraview or Eizo ColorNavigator. What we do is to view the paper beside the monitor, and visually adjust the display white until it matches the paper white. Again, it’s important to stress that the two must be viewed side-by-side while doing this. The calibration/profiling software then generates a new profile for the monitor based on this white point. To view the print side by side with the monitor, the profile is then changed from the normal 6500K profile to this adjusted profile; Simulate Paper Color is not used.
- Adjust the Destination Profile: As explained in Simulate Paper Color above, the white point in the destination profile is used by Photoshop/Lightroom to simulate the paper color. This white point is in a tag called ‘wtpt’. To adjust it to the print paper white, the following procedure can be used:
- With the illuminated (white) paper beside the monitor, create a new document in Photoshop and (in full-screen mode) adjust the Lab values so that there is a visual match to the paper white.
- Convert these Lab values to XYZ using http://www.brucelindbloom.com/index.html?WorkingSpaceInfo.html. The Ref. White must be D50. This gives the XYZ values for the paper white.
- Convert the paper profile (which should be made to D50) to xml using IccXML: http://sourceforge.net/projects/iccxml/
- Copy and paste the XYZ values from Bruce Lindbloom's calculator into the 'wtpt' tag in the xml version of the paper profile
- Change the ‘desc’ tag to append the illuminant condition (say ‘–Tungsten 2800K’)
- Save the modified xml file to a new filename with the illuminant condition appended (say ‘–Tungsten 2800K’)
- Convert the xml version of the modified profile to icc, again using IccXML
Ideally, the black point should also be adjusted, but this can’t be done visually (the colors are too dark to compare visually). However as an approximation, the values in the ‘bkpt’ tag in the xml file can be adjusted in proportion to the adjustment in the ‘wtpt’ tag: XBN = XBO*(XWN/XWO) where XBO is the Old Black X Value, XBN is the New Black X value, XWN the New White X value, XWO the Old White X value … and the same for the Y and Z values.
The modified profile can now be used to soft-proof: Simulate Paper will give the correct paper white for the lighting conditions used.
We need to go back to the original profile if we want to make further adjustments to the white point or brightness. What I do is to save the Lab adjusted image so that I can tweak the original Lab values. I also copy the values into a spreadsheet which computes the modified black point and I also save this spreadsheet for reference.
I personally prefer the second method as it can be used on monitors that don’t have profiling software which allows visual adjustment of the white point, and because once it’s set up the soft-proofing Simulate Paper White works properly as it is designed to work, and there is no need to change monitor profiles. However, the first method works just as well, so it’s a matter of preference and what monitor software you have.
Other FactorsThere are other factors to take into account when soft-proofing images. For example, the images should be the same size and viewed from the same distance away and at the same viewing angle; the brightness of the paper and (soft-proofed) monitor white should be the same (or very close to each other); the soft-proof should be viewed in full-screen mode with a dark background (not possible in Lightroom at present … may be a reason to prefer adjusting the monitor white to adjusting the destination profile); there should be no bright colors in the room; ideally the only thing that should be lit is the print (the reflected light is normally bright enough to work in); … and no doubt other things!
I hope this will be of use to you. Again, many thanks for your help!
Robert