Help Fixing Out of Gamut Colors

leadZERO

There are numerous tutorials (including this site and SmugMug) that give plenty of information on how to softproof and preview what a print would look like. And from them I feel pretty comfortable with color management and what different gamuts/devices offer and the terms of the trade.

However, the one thing I can't find a lot of information on is what to do when large important tones in your images are out of the gamut of the printer/paper/ink. For instance, let's examine the following image:




(Feel free to C&C the image, I wished the DOF was a little bigger but I liked the skewer through the leaf and the firey look).

When softproofed in PS3 for the EZ-prints profile we end up with the following (above after Edit/Convert to profile...):




Is there anyway for me to print this image without losing the punch the original image has? What are the best practices when trying to keep on screen representations and printed representations similar? (For the sake of argument let's assume the viewer has a properly calibrated monitor.)

If you'd like to play with a larger version of this image: http://photos.rpsommers.com/photos/445191048_RvTTz-XL.jpg

Thanks in advance!

colourbox

It could come down to creating more color contrast in the available medium, since you can't push the colors out any further on a particular paper. You can try a different rendering intent like Perceptual vs. Relative Colorimetric, using soft-proofing to preview the results and see which is better (since the right rendering intent for one image is often bad for another). You might actually get the best solution from studying some of the ways to create color contrast within the limited available colors, often involving channel manipulations or Lab mode (like steepening the a and b curves). Some of the best techniques seem to come from Dan Margulis' books and his followers on this forum, all of which I am still studying.

arodney

This isn't going to work because you really can't soft proof and edit using this lab, they don't use the output profile they provide for proofing (silly) and you can't use it to convert to the final color space.

Here's the way it should work. Note first, out of gamut colors are out of gamut, unprintable. Nothing will bring them back. The idea is to view the image prior to application of the profile and with a soft proof, then edit one based on the soft proof and convert using that profile. Since you can't convert because the lab demands sRGB, the exercise (or their workflow) is worthless.

The technique is to do the following:

1. Open image you wish to deal with.
2. Make a duplicate of this image ( you will later just throw it away).
3. On the original image, setup a soft proof using the output profile, desired rendering intent etc and have the simulate check boxes ON.
4. Move the windows such that they are side by side and the same zoom ratio (easy to do in CS4, even in CS3, shift key will work with zoom and hand tool to sync up two windows).
5. Make the original the image on the left which you'll edit, make the duplicate sit to the right as your visual reference. Target the original image.

You'll make adjustment layers to the original for doing very small but important tweaks such that the original (on left) more visually matches image on right. The biggest impact is the simulate check boxes which show you the correct dynamic range* of the print process on the display.

Usually a slight curves tweak to the upper tones will improve the "milky" appearance of the lower dynamic range of the paper.

Then a slight saturation adjustment to the positive might be useful (something in the 5-10 range but again, nothing will bring back the out of gamut colors).

Lastly, you might need a slight selective hue shift, again with the Hue/Sat adjustment layer.

You'll end up with two or three adjustment layers that make the original with soft proof closer match the duplicate preview of the "before" (non soft proofed) image. You can throw that away when done. Put the adjustment layers in a layer set, label it! Something like "Epson3800Luster_Percpetual". When ever you print again, turn this layer set on. Make as many layer sets as necessary for each output device and paper of which you have a profile you can actually use. Be sure to turn on and off the appropriate Layer set of course.

This technique will get you from say 90% of what you want to 95% in terms of soft proofing and screen to print matching. Of course, it requires a lab (or you) that has color management properly setup, something the site you mention clearly doesn't.

*Assuming you've got a good, calibrated display, proper lighting for the print and a good output profile for the print process.

arodney

colourbox wrote:

You can try a different rendering intent like Perceptual vs. Relative Colorimetric, using soft-proofing to preview the results and see which is better (since the right rendering intent for one image is often bad for another).

Considering the lab isn't going to use that profile or have any way to (or want to) follow the end users preferred rendering intent, what's the point?

Its silly bashing your head against the wall here, the lab has a half baked idea about how users should be working with profiles. Just send them sRGB and hope for the best (or get them to supply an actual output profile and pass that data through to the printer). Margulis books will not help a lick.

colourbox

I think the scope of my answer was too wide - wasn't thinking specifically about this lab.

leadZERO

arodney wrote:

This isn't going to work because you really can't soft proof and edit using this lab, they don't use the output profile they provide for proofing (silly) and you can't use it to convert to the final color space.

*** snipped ***

Wow, that's a lot of information! I'm not sure I understood it all, but I've bookmarked it to reread a few times. I also enjoyed reading your articles on your web page.

arodney

leadZERO wrote:

Wow, that's a lot of information! I'm not sure I understood it all, but I've bookmarked it to reread a few times. I also enjoyed reading your articles on your web page.

Here's two on soft proofing:
http://www.ppmag.com/reviews/200409_rodneycm.pdf
http://www.ppmag.com/reviews/200411_rodneycm.pdf

Here's an eSeminar (free) from Adobe on soft proofing:
http://seminars.adobe.acrobat.com/p84783897/