Color of Green | Photoshop Training with Dan Margulis

Dan Margulis Applied Color Theory

The Color of Green

Date: Sat, 19 Jun 2004 13:27:52 -0700

From: Lee Varis

Subject: The Color of Green?

Hi all,

I've taken on a color correction project for a book about palm trees (Dan, you know what I'm talking about). Before I get too far into this I thought I'd solicit suggestions from the group regarding the situation I find myself in with the printing conditions.

There are 133 images of palm trees with every discernible shade of green and in some cases these trees are more of a teal color in the leaves.

All of the images are scans from film from various sources but all of the scans were done by the Korean printer. Almost all of the scans open with an embedded profile that is reported as "SWOP (coated) 20%"... curiously, I'm seeing TIL as 360 and, in some cases, as high as 376!!! Black seems to be limited to 85% in just about all of the scans I've looked at. If I look at these scans in the standard "SWOP coated v2" they don't look radically different but most of the time the shadows do plug up as I would expect. A typical deep black shadow reads (cmyk) 95, 90, 90, 85. The black plates look like UCR - fairly light except for the deepest shadows but I'm only seeing the slightest color removal in the yellow plate. Otherwise the cyan and magenta plates look very rich into the shadows - definitely not GCR!

The printer has requested that the images should not be "converted" to another profile – presumably, that means don't change the character of the seps, the black generation or the general balance of the plates.

However - just about all of these images look like they'll be too dark with plugged up shadow detail ( as the numbers would suggest), even when viewed with their "profile" . Color correction is in order but I need to careful about how aggressive I am and what kind of assumptions I make about the printer. My experience with Asian printers suggests that I'm not going to get much, if any useful information from the printer - we will get press proofs but from a different press than the one that will actually run the job - they don't do matchprints and there will be no press check! I do have some preliminary press proofs printed on the stock that they will use for the book and they match the screen previews reasonably well, though, of course they look a bit dark - not as bad as I would have expected.

SO, for all the color detectives out there... what are your suggestions? Anything in particular I should look out for. If it looks like the black limit for their setup is 85% - how far can I deviate, how much cmy can I subtract / how much K can I add. Any suggestions on green values, channel mixer tricks for elevating dark greens in CMYK files. Can I even assume that their scans reflect their print condition?

Please no "profiling" suggestions, thats not an option in this case. I'm looking for general color correction guidance from anyone with experience with this sort of subject matter / printing conditions.

regards,

Lee Varis

http://www.varis.com

888-964-0024

___________________________________________________________________________

Date: Sun, 20 Jun 2004 00:28:11 -0000

From: Stephen Marsh

Subject: Re: The Color of Green?

Lee Varis wrote:

All of the images are scans from film from various sources but all of

the scans were done by the Korean printer. Almost all of the scans open

with an embedded profile that is reported as "SWOP (coated) 20%"...

curiously, I'm seeing TIL as 360 and, in some cases, as high as 376!!!

Hi Lee, although it is perhaps possible they scanned as RGB and separated in APS using custom CMYK set to UCR/Lt GCR with an ink limit that was indeed this high...I would guess that these are drum separated scans (probably a prop. sep. table).

I don't know what I can tell you Lee, from what I have picked up from your posts over time you seem to be very clued in to the post camera process.

I would perhaps perform two initial tests, to gauge the profiles/files. Obviously if the tagged profile plays no critical part in the process from this point on, you will need to find a profile which does reflect how a known image will generally print in the expected conditions.

Test 1: Look for presumed neutrals and or deep shadow areas which are presumed to be neutral (the CMYK numbers will help guide you to this). View these areas as LAB info, do the CMYK numbers when linked with the profile which describes the image indicate neutrals? If the original was neutral and no edits have thrown this off...then the profile would presumably indicate neutrality for these areas. This can sometimes show that the tag is indeed the separation tag and not one which has been embedded which may not be the sep profile but an indication of the more general but similar output conditions or it could be a mistake and the profile is a wild goose chase.

If it does appear that this profile does accurately describe the numbers in the file (the separation profile) then one may feel safer about using it for mode conversions or other tasks.

Test 2: Proofing test. I guess this could be done on-screen, but I usually do this as a hard copy digi proof for inspection in a light box.

Produce a two up proof, in-house if possible on a typical image or a montage of a small average range of images. The upper image is unchanged, the CMYK values are output as is and the source profile does not come into play for this test...only the numbers matter. The proofer can assume whatever source it likes to convert into it's native space for proofing purposes but a SWOP like space would obviously be a good choice.

The lower image will be converted from the tagged source profile to whatever profile you decide to use for re-separating and editing in, if you go down this path. No edits are performed, just the mode conversion from the tagged CMYK to your preferred CMYK. Again the profile plays no part in the output, we wish the see how these new numbers compare with the old numbers.

If the tagged profile from the original image/s is an accurate reflection of the separation point - then the lower image should not look radically different to the upper image. If this is the case, you could have more confidence in using the original tagged profile for your later use.

Black seems to be limited to 85% in just about all of the scans I've

looked at.

The SWOP v2 profile is set to a 90% limit, but the custom CMYK can be altered from the default 100% to others - and it obviously depends on the LAB value of the original shadow region, only the 0L would get mapped to the darkest value of the output profile.

If I look at these scans in the standard "SWOP coated v2"

they don't look radically different but most of the time the

shadows do

plug up as I would expect.

In my experience, there is quite a bit of difference (OK, not huge -

but far from subtle) of incorrectly assuming a SWOP v2 profile as the

profile for a legacy SWOP engine conversion. There is hue variation

(in all areas) and very much so in important areas like skintones and

neutrals, greens are duller...contrast is flatter too.

Do these images have the legacy profile tagged and you are just

seeing how another preview of 'SWOP' compares, or are these files

untagged and you are attempting to see what would suit them as a

description?

A typical deep black shadow reads (cmyk) 95,

90, 90, 85. The black plates look like UCR - fairly light except for

the deepest shadows but I'm only seeing the slightest color removal in

the yellow plate. Otherwise the cyan and magenta plates look very rich

into the shadows - definitely not GCR!

Sounds even more like a 'raw drum cmyk scan'...or a short K and UCR.

The printer has requested that the images should not be "converted" to

another profile – presumably, that means don't change the character of

the seps, the black generation or the general balance of the plates.

However - just about all of these images look like they'll be too dark

with plugged up shadow detail ( as the numbers would suggest), even

when viewed with their "profile" . Color correction is in order but I

need to careful about how aggressive I am and what kind of assumptions

I make about the printer.

It does sound like their are conflicting briefs here. You have taken on a colour correction job...so either the client or printer understands that some work will be needed on the files, either to depart from the current colour - or if this is 'fine', then to massage the overweight total inks into a safer range but without altering the feel of the non deep shadow areas?

If the task is the later, then things are fairly easy to do manually but this will not be fun for over over 100 files. One can write an action and batch the TIL correction if needed, presuming all the images in the batch have similar K to CMY ratios in the shadows.

After the batch, a quick inspection/correction may be needed but the

batch can speed up this work over doing things totally by hand.

Good tools are the selective colour command or channel mixer, using blend if sliders and or masks derived from the K plate to isolate the edit to only the deepest portion of the shadow and to not overly affect any crossover transitional shadow tones into the three quarters which may create a hue alteration or other unwanted visual result from the edit.

This has come up in the past on the list, please feel free to contact me off-list if you would like to go into this in more detail Lee.

My experience with Asian printers suggests

that I'm not going to get much, if any useful information from the

printer - we will get press proofs but from a different press than the

one that will actually run the job - they don't do matchprints and

there will be no press check! I do have some preliminary press proofs

printed on the stock that they will use for the book and they match the

screen previews reasonably well, though, of course they look a bit dark

- not as bad as I would have expected.

What seems a safe TIL and K to you, since you know the stock and can see a general aim print? Is 300 too light and 330 a good middle ground to edit or re-sep to? This has often worked well for me in coated magazine publication quality type conditions where I knew that I could use more than 300 but did not wish to go to 360.

Regards,

Stephen Marsh.

___________________________________________________________________________

Date: Sat, 19 Jun 2004 22:45:46 -0700

From: Rick Gordon

Subject: RE: The Color of Green?

For what it's worth, we recently published a book that was printed in China. They requested that we separate to a 360 TIL with 90% max black. We profiled their wet proofer output, and I generated profiles to their specs. The final results were very good, with detail held through the range, so I can vouch that a 360% TIL can be managed under the right circumstances.

The seps were generally created using black compensation, which seems to lighten the TIL a bit from the requested amount. And they requested a much weaker limit for flat rich blacks -- as I recall, about 220%.

Rick Gordon

___________________________________________________________________________

Date: Sun, 20 Jun 2004 14:08:53 EDT

From: Dan Margulis

Subject: Re: The Color of Green?

Lee Varis writes,

>>All of the images are scans from film from various sources but all of

the scans were done by the Korean printer. Almost all of the scans open

with an embedded profile that is reported as "SWOP (coated) 20%"...

curiously, I'm seeing TIL as 360 and, in some cases, as high as 376!!!

Black seems to be limited to 85% in just about all of the scans I've

looked at.>>

Par for the course. The files were drum scanned; the numbers are consistent with 1980s views on sheetfed printing. The files are obviously untagged when scanned, then the printer opens them up in Photoshop, clones out any dust, hairs, etc., and resaves them using a color setting that unbeknownst to him embeds an irrelevant profile.

>Can I even assume that their scans reflect their print condition?>>

It's probably not horrendously far off, particularly if this is sheetfed printing on a good stock. But the paper quality is critical. If it's not good, the chances of them being able to hold shadow detail with such high numbers are slim and none. But in any case, I would be more conservative, because if you assume the best and it turns out not to be so, the job will crash and burn, whereas if you assume the worst and it turns out better than what you had hoped, you're not badly off.

>>My experience with Asian printers suggests

that I'm not going to get much, if any useful information from the

printer - we will get press proofs but from a different press than the

one that will actually run the job - they don't do matchprints and

there will be no press check! I do have some preliminary press proofs

printed on the stock that they will use for the book and they match the

screen previews reasonably well, though, of course they look a bit dark

- not as bad as I would have expected.>>

Well, that's a whole lot better than nothing. If everything looks a bit dark, then you lighten everything a bit. The printer is going to rely on those proofs, so if the proofs are too dark, so will the final job be.

>>SO, for all the color detectives out there... what are your

suggestions? Anything in particular I should look out for. If it looks

like the black limit for their setup is 85% - how far can I deviate,

how much cmy can I subtract / how much K can I add. Any suggestions on

green values, channel mixer tricks for elevating dark greens in CMYK

files.>>

The more important shadow detail is to the picture, the more you add K and subtract CMY.

As for the greens of palm trees, the very bluest ones will have C=Y. The very yellowest will have (Y-C)-(C-(M+K))=0.

To increase the saturation of greens in a CMYK file, try Channel Mixer>Magenta=Magenta 130%, Yellow -30%. You may need to do this on

a separate layer in Color mode.

As drum scans, these have presumably been sharpened already. However, palm trees are a difficult sharpening case because their narrow leaves often aren't fully resolved. Wide semi-haloes are often a good solution. So, try (in CMY only, not K) sharpening at Amount 60%, Radius 10.0 pixel, and see if you see any improvement, understanding of course that you can probably find better numbers if you work at it.

Dan Margulis

___________________________________________________________________________

Date: Mon, 21 Jun 2004 09:09:01 -0700

From: Lee Varis

Subject: Re: Re: The Color of Green?

Steven Writes:

it is perhaps possible they scanned as RGB and

separated in APS using custom CMYK set to UCR/Lt GCR with an ink

limit that was indeed this high...I would guess that these are drum

separated scans (probably a prop. sep. table).

Yes, this is my guess as well – I also think Dan has it it right in the head below –

Dan writes:

The files are obviously untagged when

scanned, then the printer opens them up in Photoshop, clones out any dust,

hairs, etc., and resaves them using a color setting that unbeknownst

to him embeds an irrelevant profile.

Steve writes:

Do these images have the legacy profile tagged and you are just

seeing how another preview of 'SWOP' compares, or are these files

untagged and you are attempting to see what would suit them as a

description?

They mostly have this mystery "SWOP (coated) 20%" tag though some have no tag which suggests to me, as Dan says above, that they were probably Drum scans into CMYK and originally didn't have any embedded profile. The wrong profile ended up there after spot knocking, the files that didn't need any dust removal didn't receive the bogus profile. Come to think of it, the printed press proofs supplied as patch color look a bit better than the screen previews through the profile supplied – that makes sense as these files are destined for a sheetfed press not a web press!

Again Steve writes:

What seems a safe TIL and K to you, since you know the stock and can

see a general aim print? Is 300 too light and 330 a good middle

ground to edit or re-sep to? This has often worked well for me in

coated magazine publication quality type conditions where I knew that

I could use more than 300 but did not wish to go to 360.

This is kinda what I was thinking. and as Dan writes:

But in any case, I would be more conservative, because if you

assume the best and it turns out not to be so, the job will crash and burn,

whereas if you assume the worst and it turns out better than what you had hoped,

you're not badly off.

So...

The more important shadow detail is to the picture, the more you add K and

subtract CMY.

As for the greens of palm trees, the very bluest ones will have C=Y.

The very yellowest will have (Y-C)-(C-(M+K))=0.

LOL, Dan, I was hoping I wouldn't need a calculator for number suggestions – can that last bit be expressed in a simple ratio? I was thinking something like 1.5:1, Y:C or can I let Y go higher?

To increase the saturation of greens in a CMYK file, try Channel

Mixer>Magenta=Magenta 130%, Yellow -30%. You may need to do this on

a separate layer in Color mode.

That's exactly what I was looking for!!! I'm used to doing this in RGB with +G and -B. It works like a charm in CMYK! Excellent!

And finally, Dan writes again:

As drum scans, these have presumably been sharpened already. However, palm

trees are a difficult sharpening case because their narrow leaves often aren't

fully resolved. Wide semi-haloes are often a good solution. So, try (in CMY

only, not K) sharpening at Amount 60%, Radius 10.0 pixel, and see if you see any

improvement, understanding of course that you can probably find better

numbers if you work at it.

They have been sharpened – some more than others and some even too much! I'l give your suggestion a try on the less aggressively sharpened ones though I'm inclined to leave well enough alone on this.

Thanks so much for this advice! I have one more question for you all. Should I leave the profile that is embedded with these files (since that's the way they come) or would it be safer, since this is probably not a real profile, that I just return the finished files "untagged"?

Please bear in mind that this IS NOT an opportunity to educate a printer on the benefits of proper profiling.

I'm just wondering if I should leave the existing profile on the theory that it may be there for a reason even though the profile is "named" SWOP (coated) 20% , we are printing sheetfed and the actual numbers in the file suggest that they are prepared with the higher TIL of the sheetfed press in mind.

regards,

Lee Varis

http://www.varis.com

888-964-0024

___________________________________________________________________________

Date: Mon, 21 Jun 2004 22:11:57 EDT

From: Dan Margulis

Subject: Re: The Color of Green?

When working in CMYK, the best way to understand what color the numbers are suggesting is as follows. The three CMY inks will obviously be in some order from darkness to lightness. If the middle ink is closer to the top one than it is to the bottom, the color being portrayed is red, green, or blue. If it's closer to the bottom, the color is yellow, magenta, or cyan. Therefore, 60c20m90y is a yellowish green, suitable for trees, but 60c50m90y is a greenish yellow, and a nauseating one at that. You don't want to use that color for trees.

Dan Margulis

___________________________________________________________________________

Date: Mon, 21 Jun 2004 22:08:27 EDT

From: Dan Margulis

Subject: Re: Re: The Color of Green?

Lee Varis writes,

>>LOL, Dan, I was hoping I wouldn't need a calculator for number

suggestions – can that last bit be expressed in a simple ratio? I was

thinking something like 1.5:1, Y:C or can I let Y go higher?>>

It can go a bit higher provided that there's more distance between cyan and magenta than between cyan and yellow.

I'm just wondering if I should leave the existing profile on the theory

that it may be there for a reason even though the profile is "named"

SWOP (coated) 20% , we are printing sheetfed and the actual numbers in

the file suggest that they are prepared with the higher TIL of the

sheetfed press in mind.

Good grief. Now that you've confirmed (by the fact that some files don't have a profile, and that the profile previews darker than the proof) what I suggested in the first place, that these particular profiles got there by accident, couldn't possibly have been used for the separation, and have no relation to the actual printing condition, you want to know whether to send them on to the next person.

Granted that if anybody uses these incorrect profiles for any purpose other than wasting disk space, they'll get an overly dark result, the second best method would be to take half of the files and embed other CMYK profiles in them, making sure that there are at least half a dozen different random CMYK profiles scattered throughout the job. That way, there's a better chance that the next person will realize that the profiles on the job are random, meaningless, and shouldn't be used. If you send them as they came to you, with no profile attached to some and only one variety of wrong one attached to the rest, somebody might carelessly use the profile on the incorrect theory that you couldn't possibly have been crazy enough to deliberately supply something with the wrong tag. Therefore, the defensive measure of supplying at least six different profiles, to make sure that the next user will be so confused that he won't use any of them.

Of course, that way, in addition to being the second best method of handling it, also takes a bit longer than the best way.

Dan Margulis.

___________________________________________________________________________

Date: Tue, 22 Jun 2004 13:59:25 -0700

From: Lee Varis

Subject: Re: Re: The Color of Green?

On Jun 21, 2004, at 7:08 PM, Dan Margulis wrote:

Therefore, the defensive measure of supplying at least six different

profiles, to make sure that the next user will be so confused that he

won't use any of them.

LOL... of course!

I've already started stripping out the profiles as I go through these, I guess I was just fishing for confirmation and perhaps to see if Profile crusaders would jump in with any kind of reasonable defense ( I'm still trying to keep an open mind, wink, wink...)

regards,

Lee Varis

http://www.varis.com

888-964-0024

___________________________________________________________________________

Date: Tue, 22 Jun 2004 16:40:02 -0600

From: jim donovan

Subject: Re: Re: The Color of Green?

Outstanding answer Lee!!!! Now lets hear the " It's not possible to

determine a color by looking at the combination and relationship of

cmyk in the file" that has been stated on this forum by some profile

crusaders!! LOL!!! of course it doesn't!!! LOL!!! regards, Jim