Color Management Myths 29-33

From ColorWiki

Revision as of 22:24, 21 August 2014 by Patrick (Talk | contribs)
(diff) ←Older revision | view current revision (diff) | Newer revision→ (diff)
Jump to: navigation, search
Reserved Article

This page is a
Reserved Article.
For more details see
Reserved ColorWiki Articles

ColorNews

This reserved article originally appeared in CHROMiX ColorNews Issue 20 on September 7, 2005.

Click here to see the original in its original context.
Email
colornews(at)chromix.com to subscribe to the ColorNews newsletter.

by Steve Upton


Contents

Myth #29: The more patches on the printer target, the better the resulting profile

From the "I can't believe I haven't covered this yet" department. This one is a classic. I think it perpetuates because, well, sometimes it's true. A lot of the time, actually. But not all the time - that's why blanket statements like this should be avoided and we should all fall back on that old, worn color management favorite, "It depends".

There are a few things to keep in mind when choosing / using a profiling target.

  1. You are trying to sample the behavior of your printer as effectively as possible. You are printing about a thousand patches which are going to be used to create a profile with 15-25 thousand color entries, which is going to be used to process images that contain 50-500 thousand colors. To say there is interpolation going on is a major understatement. "Effective" means the RIGHT patches, not just the right number of patches.
  2. More patches is better IF they are spaced well in your printer's space AND if your printer has non-linear behavior that needs to be mapped. Your printer may obey RGB or CMYK, but it speaks Lab. Many profiling targets evenly space patches according to RGB & CMYK numbers. If you are profiling an inkjet, chances are that it changes density and hue in strange ways when the ink gets heavier. An effective target will have more patches in the problem areas and less in others.
  3. More patches is BAD if your printer is very linear (like a well-behaved printing press). In this case more patches introduces waves into the system which are caused by abnormal behavior of the press. Abnormal as in "here today and gone tomorrow". Fewer patches will make smoother profiles and overcome transient problems.
  4. As much as you might like to be democratic, you don't care about all colors and all color shifts the same. We see shifts in neutrals before we see them in other colors. Our visual system has evolved to watch out for sick people and bad food. This means that we might need to sample neutrals, flesh and food colors more than others, and ensure that hue shifts are never seen.

I am sorry to say that it is not easy to determine which targets are best for your system without trying them. Nor is it easy to create targets without some experience in the field and the right tools. But it can help. If you need help, let us know and we'll see what we can do.

Myth #30: If your RIP doesn't allow total ink limiting, just limit each channel.

It is surprising and disappointing to find RIPs being sold today that only offer ink limiting by the channel.

A bit of a review.

When you set up a new RIP, the first thing you discover is that it can be a bit wild with the ink until you get it under control. If you are trying a 3rd party paper with a new RIP, you might be startled to see ink puddling, curdling, and pouring down the paper when you print a profiling target.

The good RIPs have a number of ink testing charts you can print to find the right balance of ink for your paper. The first typically has inking from 0 to 100% in each channel. 100% on a wide-open RIP can mean way more ink than is recommended, but it is fairly easy to determine which percentage you should enter into the RIP software to stop the ink. This then becomes 100% whenever you ask for maximum ink and you can stop worrying about single ink channels over-inking.

The problem is that some RIPs stop there. This is woefully inadequate, as you soon discover when you reprint your profiling target and find over-inking as the inks get above 200-300%. A good RIP will then lead you through the process of printing a total inking chart which will layer inks in a few different orders so you can clearly see that your printer "loses it" above 280%, for example. Then you simply key this number into the RIP and it won't let ink get above that level.

If you try to limit the total inking by limiting the maximum amount in each channel, then you are forced to reduce this example RIP to about 70% per channel. Have you seen 70% Cyan on your printer? Lame. Watered down.

Do it in each channel and you have squeezed the gamut of your printer down to something that is probably half the size it could be. Yes, you could print the target at full inking and then set the total ink limit in the profiling software to 280%, but then you will have a profiling target that is over-inked on many of it's patches - which is a waste of patches, ink and accuracy. (see previous myth)

So if you are RIP shopping (and I pity you if you are, as there are literally hundreds of them out there), then avoid the ones that only allow per channel ink limiting. They will be limiting your control and are a good indicator of a company that doesn't understand all they should to build good RIPs.

Myth #31: When setting up a RIP for proofing, you should match density and dot gain of the inkjet to the press

If you somehow manage to get the same inks into your proofing system that you use on press (or at least the same hues) and it is producing dots exactly the same way as your press, then this MIGHT be a good idea. Otherwise, don't bother.

Inkjets and their inks are very different from press inks.

When it comes to inkjets matching a press, they can do a remarkably good job. The trick is to learn effective RIP configuration and profiling techniques, and rely on profiles to do the conversion work.

Myth #32: Relative Colorimetric intent means no color shift

This confusion tends to arise when people graph colors in ColorThink and then apply profiles to those colors, creating vectors. When the relative colorimetric intent is selected, vectors appear that surprise people who've been led to believe that rel col means no shift.

Remember that colorimetric rendering intents come in two flavors: relative and absolute.

Relative means that, as colors are converted to the printer (for example), they are altered to appear correct RELATIVE to the new paper white point. Normally this is exactly what you want, as your eye will see the paper white and expect colors to be balanced to it. The perceptual and saturation intents also convert colors relative to the new paper color. Absolute is the intent that overrides the new paper color with an attempt to simulate the source white point (a press paper color simulated on an inkjet, for example).

If you don't want to see color shifts in your graphs, choose absolute colorimetric and you see graphs with the shortest vectors possible.

I should note that the shifts produced by rel col are typically not a bad thing when done properly. As mentioned above, the eye expects it and anything else would look odd.

Myth #33: The new SWOP guide (version 10) contains an effective summary of color management.

OK, so I'm getting on the soap box here, but please!

I'm sure the updated guide is effective at describing how to print according to SWOP guidelines (I really am sure, we are using it today) but the description of color management is contradictory and, in fact, violates at least one color management myth I have described before (#2, that there is some internal color reference that output is compared to)

I have a feeling it is a repetition of descriptions from previous SWOP editions but I shudder to think about how far back it was written.

Please, helpful SWOP people, rewrite the color management portion of the guide. Expand it to include updated soft and hard proofing information and perhaps even some information regarding printing using colorimetry in addition to densitometry. If you need help please let us know, we are happy to be of assistance.

Thanks for reading,

Steve Upton


List of all color management myths

Personal tools
Namespaces
Variants
Actions
Navigation
Toolbox