I was reading that dSLRs can now capture around 8 stops of dynamic range. I assume that the DR of paper is still around 6 stops. That implies that somewhere there is a compression of the DR taking place when printing.

I'm wondering if something can be done to make sure that this compression is done optimally or is it not an issue to be concerned about?

Knowing about that compression, among other things, is what differentiates point-and-shooters from dedicated photographer/printmakers -- knowing about it, and how to use it well to present an image. I can't define an optimal mathematical transform for all images; I wing it, go with what looks right. In the end, my images look like mine and, often, like no one else's.

Does that mean you process your images before printing to compress the dynamic range?

Well, that would be the point of the zone system. Back in the days of film, this quantified a method of compressing the dynamic range of the real world down to the smaller dynamic range of black & white negative film, exposing and developing for the values that mattered to the image; and then compressing the dynamic range of the developed negative down to the yet smaller dynamic range of printing paper. Nothing has changed in principle; today you compress the real world dynamic range down to the range of the sensor, taking care not to blow out the highlights yet exposing to the right to maximize shadow detail. Then you work the image in Photoshop to optimally compress the dynamic range down to what your inkjet paper/ink combination can handle. There is no "formula" for doing this; it depends on the nature of the image, the limitations of your chosen paper, and your personal artistic interpretation. Enjoy!

I believe good color management and good profiles will handle most of the "compression" as you call it, and choice of rendering intent, whether perceptual or relative colorimetric is an important part of that. Soft proofing using the output profile is one way to make adjustments based on what is going to happen on the paper vs what you see on the screen, and I usually have a set of adjustment layers that I only enable when printing. Some of those adjustments are also made after having viewed an actual print (what can I say ... I'm not very good at soft proofing yet). In fact I often have two sets of adjustment layers, for different paper types.

Actually, unless you know how to properly soft proof in Photoshop, you are leaving the dynamic range compression from the display which could be as high as 500-600/1 on LCDs to maybe 150-200/1 on a print pretty much to chance. Color management is designed to move the color gamuts of the image to the color gamut of the printer/paper profile. And that's pretty much it. Where the tone curve falls is up to you and only soft proofing can really show you that.

It will be a strange print if color information was compressed and the tonal range not.

Like with color the grey component is compressed in perceptual rendering and CM will shift the densities within that compressed range so they are separated best for the eye. Along the greyscale Lab curve. The same curve that is used in the QuadTone Rip profile creator + profiles to shift the possible densities of a paper + ink for a better perceptual B&W print. That doesn't correct a bad tonal separation in an image file but will correctly compress an image that showed correct tonal separation on a calibrated monitor (where it actually could have been compressed along the same Lab curve when the screen has a limited range too). For relative colormetric BPC will do something alike for shadows but that's more application CM specific and can vary per developer's taste. Compression happens, nice or less nice. That editing based on the softproof (a highly overrated feature in my opinion) can improve a printed image to your eye is nice but doesn't proof that there's no compression or perceptual tone rendering done in CM.

Ernst Dinkla

try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/

Does a 16 bit image have to be converted to 8 bits before it is printed?

I have to disagree to some extent with the esteemed biker. Printer profiles for a given paper do include a DMax, which describes the darkest black that the paper/printer combination can produce, and the CMS (color manage system) can take this into account.

For a good introduction to the degree of compression required for printing, I would recommend the excellent white paper by
Kark Lange posted on Adobe's web site. Another good reference is Adobe's white paper on black point compensation (BPC):
AdobeBPC. This is an extension to ICC profiles that has been implemented by Adobe to further improve reproduction of the dark tones in the print.

The OP might find the following thread helpful. The topic was about the blacks printing too dark on prints from Costco:
Prints Too Dark

Shown below is a graphic produced by Gamutvision, which demonstrates compression of the shadow values in the problematic print. The shadows are considerably lifted by the profile as shown by the black lines at the bottom of the graphic. These indicate lifting of the shadows from the values in the original image to those handed off to the printer. One can try both perceptual and relative colorimetric rendering to see which gives the best result in soft proofing. Glossy paper has a higher DMax than matte and can reproduce the shadows better.

Click to view attachment

Click to view attachment
Of course, one should not expect an automated process to substitute for the artistic intent of the photographer and soft proofing is invaluable as Jeff indicates. For less advanced photographers, the profiling process does offer some help.

Bill

Well, which rendering intent?

RelCol tries to preserve the original luminance values while Perceptual will adjust the the saturation and luminance values to fit into the output gamut. But, how are you gonna tell which intent will be optimal for a given image?

Also, have you ever seen a ICC profile with a brain? I haven't...which means that a profile (and color management in general) treats everything the same in a given pipeline. So, you may have a high contrast scene, or low contrast or high key or low key...color management treat them all the same.

No, color management was _NOT_ designed to deal with dynamic ranges...it was designed to deal with color appearance. The fact that it does an "ok" job of dealing with paper white and the Adobe ACE of dealing with blackpoint is incidental to it's main role of color management.

Which is why I say to soft proof to determine the optimal tone curve for the output...

ICC profiles don't do that good of a job with out of gamut colors either. RelCol merely clips them, while Perceptual compresses them by a given arbitrary amount without even considering if the image has any out of gamut colors or not. It may apply compression when none is needed, and it may not apply sufficient compression when more of it is needed. We need a smart CMS that actually looks at the tonal range and color gamut in order to do a better perceptual rendering.

Of course, it is doubtful that a CMS will exercise artistic judgement, but the process can increase the likelihood of obtaining a usable image.

If color management does not deal with dynamic range how would soft proofing help you determine the optimal curve for output? Are you considering the profiling of the printer and paper to be a separate function from color management?

Soft proofing saves you money. In the bad old days you'd spend an hour in the darkroom dodging and burning, and throw away the print because you goofed it up yet again. Now you can get it right before you print. This doesn't mean you get to "let the machine decide" which is what happens when you just jam an image through an ICC profile and pray.

Biggest advance in technology for printing, as far as I'm concerned, is the out-of-gamut warning display in Photoshop and other programs. One big splash of magenta tells you "hey, that area ain't gonna print right." And now you get to figure out how to deal with it.

Too much math, not enough art makes me sad.

There's no real substitute for proof prints. Softproof and out of gamut warnings depend on the softproof part of the printer profile (that can be edited separately) and what the application's CM does with it. Compared to proof prints I consider it an overrated feature. Like I have doubts about the printer gamut competitions based on profiles in profile viewers without actual prints compared.

Relative Colormetric can not clip gamut, the gamut of the printer will clip gamut, the four rendering choices are there to deal with the distribution of the image color on that more limited gamut. The less perceptual the more compression at the gamut extremes if the file has colors beyond the gamut of the printer.

With an intelligent CM you still have to make decisions. If that intelligent CM stretches or compresses a high contrast scene, or low contrast or high key or low key to the available print tone range or similar color contrasts to the available printer gamut you may have lost what makes them different. In that sense treating them all the same in a dumb way may be as good as an intelligent method. The CM isn't there to replace image editing.


Ernst Dinkla

try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/

Sounds like somebody who doesn't USE soft proofing (or at least doesn't use it the way I do).

While it's a given that soft proofing DEPENDS on the accuracy of the profile (printing will too) and the accuracy of the display profile, assuming you have good profiles for both, Photoshop's soft proofing is about 90% (or more depending on your display gamut) accurate with the Epson printers I've used...and if you turn on the Display Options and check both Simulate Paper Color and Simulate Black Ink, you'll get a pretty accurate preview of the print's dynamic range, which at this point is probably more important that merely simulating the color renderings on print since the gamut of todays printers is so large relative to that of the display.

Keep in mind that the dynamic range of the print can also depend a lot on the viewing lighting, esp. for gloss prints.

One reason for that is in the work I do most of the time. There's a piece of art to be reproduced and the prints have to match it. Original and replica can be put next to one another. In the translation from say 12 pigments through RGB, displayed as RGB, in softproof represented as RGB and at the end represented in 3-6 hues that are not related to the 12 mentioned first I find the softproof not so much proof. So while I appreciate the ICC system to fit the image into the prints gamut including the tonal range (less a problem in my case) there's color tweaking needed that is not visible in the softproof. Maybe opinions differ because the work to be done differs. I share this experience with another art print shop.


Ernst Dinkla

try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/

Largely it's out of your control. This is one of the joys of digital printing. What happens is your camera takes a subject brightness range (SBR) of eight stops (say) and converts that to a range of 0-255 (if 8 bits), 0-65535 (if 16 bits), etc.

Your printer looks at this same range of values and prints them out as the range of 0-255 also.

The trick here, if you call it a trick, is that when you linearize and profile your printer you define the reflectance range for the printer/ink/substrate. That is, you define how ink black (and that gets assigned "0"), paper white (and that gets assigned "255") and everything in between. The printer knows how to assign your file's numbers to tonal values on the print by using the ICC profile. Nowhere was the reflectance range of the printer/ink/substrate defined explicitly for you. That's what I mean when I say it's largely out of your control.

What this means is that you get a perfect match between your digital file and your printer/ink/substrate.

Compare this to the Zone System for exposing and processing film for darkroom printing. In the Zone System you are taught how to use the film as an intermediary. Its job is to capture the SBR of the scene and convert it to a density range on the film such that an enlarger light shining through the density range of the film created a similar density range from the paper. In other words, the film was used to shoe-horn the SBR into the (generally smaller) reflectance range of the paper.

With digital, there's no shoe-horning. The digital file is an exact match to the printer/ink/substrate, by definition. This puts a premium on your linearizing the printer (especially for B&W work) and making excellent ICC profiles of course. But if you do that well, you get an exact match where the file's "0" gives you ink black and the file's "255" gives you paper white.