Untitled Document

I'm often in a situation where I need to repurpose an image for different printing conditions.

I'll receive an image in CMYK and I know nothing about how it was separated. I start by

checking the Total Ink Limit. If I get a reading of, say, 340-360% and my printer specifies no

more than 320%, then I know I have to reseparate.

If I use the "Convert to Profile Method", nothing seems to happen (the numbers don't

change). Under Destination Space (Custom CMYK), I specify Black Ink LImit at 90% and Total

Ink Limit at 320%. I'm not quite sure what the Source Space should be as I didn't originally

separate the image. Anyhow, the bottom line is nothing happens. Can anyone shed any light

here?

Much thanks

Michael Fedecky

___________________________________________________________________________

A conversion requires two different profiles, a source and a destination. In this case you'll need to tag your previously untagged image with a profile that roughly approximates the way it was separated. *Then* you can convert to your destination profile. Just make sure you're not trying to convert from Profile A to Profile A - that's probably what is happening.

--

J Walton

___________________________________________________________________________

On 3/14/07 8:42 AM, "mfedecky" wrote:

You sure you're not confusing this with Assign profile? That doesn't change the numbers, only the definition of the numbers (the color space the numbers are associated with). Convert to profile always changes the numbers.

The source space is whatever the color space is being defined by the embedded profile. No embedded profile? CMYK mystery meat. You NEED an embedded profile to define the scale of the CMYK numbers or you have to use the Assign profile command and make a big guess. Either way, a bad situation!

Andrew Rodney

___________________________________________________________________________

Michael,

Convert to Profile changes the numbers so that the source image can be printed under the different press conditions specified by your new profile. In other words it recognizes that image colors will probably will print differently under different press conditions, so it changes the numbers to compensate for this. Its purpose is to maintain the same colors. It will not change ink settings. To do that you will need to convert to Lab, then convert back to your target profile. Once a LAB or RGB file is converted to CMYK, the ink limits can only be changed by starting over with a non-CMYK file and then converting to CMYK. You can change the colors with Curves or other editing tools, but not black ink limit or total ink.

Let us know how you handle the problem, would you?

Howard Smith

___________________________________________________________________________

Michael I agree with your inspection for untagged CMYK images, info palette and total ink is a good start. Also note the CMYK colour build of the deep shadows while you are there. Next take LAB readings via a fixed sampler as you assign different profiles to the CMYK image, one can sometimes presume that this should be neutral and the LAB or RGB converted numbers in the info palette should reflect this. If there are presumed neutral grays in the file, then they should have a sampler placed for neutral LAB value readings too. This can help with the visual process, by giving numerical feedback in various modes.

That being said, I do not agree with your conclusion about *having* to reseparate if the only concern is the deep shadows, one can edit the excess ink to conform to the required limit without leaving CMYK. Both have good and bad points. The profile transform is the quickest and easiest method, but it relies on having a good profile to describe the CMYK numbers. Manually editing the build and or total ink limit of shadows takes more work but it preserves everything else.

Toward the end of the following post I briefly mention two methods of manually editing the total ink to conform to the new desired target:

http: //tech.groups.yahoo.com/group/colortheory/message/17014

A third approach, one could also convert a duped file and then layer the converted file over the original (preserve colour numbers/values, no conversion to maintain appearance) and only blend in the deep shadows of the upper layer.

The more different the new target condition is, the more need for a conversion. If the condition is the same or similar and you just need to adjust the total ink, then edit the file manually so that only the deep shadows and their transitions are smoothly adjusted.

Sometimes knowing the separation profile is important, but other times it is more important to know the output condition, which may or may not have a good profile - and not how the image came to be CMYK. One may use a different or 'false' CMYK profile of a similar or very different condition for use in another condition, as the 'foreign' profile does a better job, one then assigns or assumes the target condition and the conversion profile does not matter.

The assumed or assigned CMYK profile in use for the image makes a big difference to the preview and conversion. Do you have a hardcopy reference image to compare against, that you attempt to match by assigning?

Regards,

Stephen Marsh.

___________________________________________________________________________

Such a nice answer I gave. But it's wrong. It seems that Convert to Profile does, indeed, change total ink and black limits. Which leaves me wondering why it didn't seem to work for Michael Fedecky. Maybe we're both doing something wrong here. Dan seems to frown on CMYK to CMYK conversions, preferring to convert back to RGB or to Lab and then reconvert. Now I wonder more than ever why that should be so.

Howard Smith

___________________________________________________________________________

Howard Smith writes,

I don't know which Dan you're referring to, but it isn't this one.

If you're planning to make some kind of change in LAB, then of course you'd have to go CMYK>LAB>CMYK. But if all you were trying to do was change the ink limit or the dot gain compensation, converting out of CMYK would be a wasted step.

Dan Margulis

___________________________________________________________________________

CMYK to CMYK conversions are very problematic unless you have a device link profile which does a direct conversion without going though LAB. You hose the black generation that way where as doing a direct conversion using a device link can honor the black generation depending on how the device link was built. Ideally, go back to the original RGB document and re-convert. Anyone doing a good deal of CMYK to CMYK conversions should look into a product that can build device link profiles. The bad news is you can©ˆt do this in Photoshop without a 3rd party plug-in.

Andrew Rodney

___________________________________________________________________________

One of the biggest reasons we do CMYK to CMYK conversions is to change the black generation of a CMYK image supplied to us, and we do not have the RGB original to go back to. The problem of CMYK to CMYK conversion not changing the total ink or black generation is most likely from having the same profile as the source and destination. Going to RGB then to CMYK allows the conversion to have an effect.

--

Bret Hesler

L.P. Thebault Company

Parsippany, New Jersey

___________________________________________________________________________

The above resolves my problem. Thank you Mr Walton. I hadn't assigned a source profile because I was dealing with "CMYK mystery meat". I now realize that I have to do some guess work here if I am to use the "Convert to Profile" method.

My further problem is that I am dealing with hundreds of images that weren't necessarily supplied from the same source. I'm looking for a quick, real-world solution here, that will get my Total Ink Limits down to 320% and thus avoid annoying my printer. I'm finding it easiest to just convert back to Lab/RGB and reconvert to CMYK. This just seems to be the easiest way to achieve this.

Michael Fedecky

___________________________________________________________________________

On 3/16/07 8:48 AM, "Bret Hesler" wrote:

I understand the need. But direct CMYK to CMYK conversions have to go thought the PCS which means you©ˆre going CMYK to LAB (usually) back to CMYK. The original black info is gone in the process. So if you have say black type, you©ˆre now going to end up with CMYK type as an example. A device link profile can honor the black, ensuring you only use K to reproduce that element. The same would be true for say a drop shadow. If you©ˆre doing a good deal of CMYK to CMYK work, you©ˆre going to have a lot more control using a device link plus, many RIPs support this on the fly.

Andrew Rodney

___________________________________________________________________________

Michael Fedecky writes,

To rephrase what J Walton said, the problem is that Photoshop isn't understanding what you are asking it to do, which is to convert it to 320% total ink. However, your CMYK workspace is presumably already set to 320% total ink, so Photoshop thinks that the file is already where you want it to be and sees no need to reconvert it. It doesn't realize that the file itself may not really have that limit.

The easier solution is to make a very modest change. Do the Convert to Profile to a 319% total ink. That will fool Photoshop into thinking that you are asking for a new file (because 319% is not the same as the 320% that it thinks the file currently is) and the conversion will take place as planned.

Dan Margulis

___________________________________________________________________________

We don©ˆt usually deal with type in photoshop, but a K-only drop shadow would benefit from device links.

I agree, in a RIP, device links are the way to go.

--

Bret Hesler

L.P. Thebault Company

Parsippany, New Jersey

___________________________________________________________________________

I have a Photoshop action that uses an image calculation trick to select areas above a pre-specified TAC. This might be useful in your case to check whether a particular image is an offender or not, so that you can decide whether to convert.

If you, or anyone here, is interested, I'll make this action available on line.

Mike Russell

www.curvemeister.com/forum/

___________________________________________________________________________

Sounds interesting. Let's see it!

Thanks,

_andrew webb

___________________________________________________________________________

Dear Mike, I would also appreciate having this action made available at your www.curvemeister.com website!

Thank you for the generous offer.

Sincerely,

Stephen Marsh.

___________________________________________________________________________

Michael Fedecky wrote:

Yes, Michael, this is so. For untagged or "non-colour managed" files, the colour management will be assuming that particular colour mode's working (editing) space in Colour Settings (no avoiding CM). If you are not going to assume a "one size fits all" setting for converting incoming CMYK files, then you have some detective work to do and even then it is often a guessing game/crapshoot as to the intended condition or separation method in use in the untagged file. As has been debated many times on this list, tags are good when one wishes to make a conversion, but untagged is also a valid insurance approach so that files are not automatically re-separated. This is most often to ensure that the K channel data is preserved and that K only elements with no underlying CMY component remain that way.

If you have a reference print, then you can assign various profiles to the image and pick the one with the best match to the hardcopy sample, before converting directly to another CMYK profile or perhaps using an intermediate step of a RGB or LAB conversion and edit to boost deep shadows before going to CMYK again etc. But choosing the "correct" profile is not so easy to do. Should it be the presumed conversion profile that was used to go from RGB>CMYK, or should it be the "output condition" profile describing the process that the separated image is targeted to?

That being said, there are some general guesses that one can make. One may inspect the total ink, the channel build for the total ink, the weight and length of the K plate, the composition of CMY to K of presumed neutral tones and the relative dot gain (TVI) of the file in a presumed output condition. You can also make a guess as to the general condition that the image is prepared for, if in the UK and receiving data from Europe, a Euroscale/FOGRA profile would appear to be a more obvious choice than say SWOP (total ink limits can be a quick guide here).

It will often come down to a personally subjective, "artistic" call on your part, as to which CMYK profile you choose to assign to the mystery meat image, before converting to the new profile that describes the new output condition.

TIP: One is not restrained to only one profile, one may combine and average duplicate images that have had different CMYK profiles assigned and then converted to RGB, with the averaging done in RGB (using either layers or apply image). This is also the case for RGB mystery meat too. If one can't decide on one profile, the image may be a compromise of many multiple profiles that appear valid. It will never be 100% "correct", as it is an average of multiple possible "renderings", but it may be more "correct" in final preferred appearance than a single best guess. Your mileage may vary though.

I hear you Michael, I used to have to do this for all the old untagged Crosfield drum scans, going down from 360+ to 320. I could have assigned the house custom proofing profile, it only changed the raw scans colours slightly and was a quicker method...but I preferred to edit the CMYK file as is, so as not to introduce colour changes via the assigned profile. The CMYK numbers represented the image better than the custom profile did, although I did convert when I was not happy with using the semi-automated action to edit.

If there is a preferred black generation and GCR and or dot gain or inkset/stock variables that you deliver with the chosen CMYK destination profile, then converting is a good idea – if this is more important than preserving the original K plate. But you live and die by the profile chosen for the source. If these are simply editorial images, then it may not be too much of a concern, as the art director or somebody will be the one to satisfy. If the images are advertising, then this can become costly and expensive.

If you have good softproofing or hard copy proofing, you can see how the untagged CMYK files numbers will reproduce in the expected output conditions and then if it is only the total ink that is a concern, then one may instead edit the deep shadows while keeping the file in CMYK. Although it takes more time than a P2P conversion, it does preserve the K channel and the existing separation in the majority of the image. This would be my preferred option, unless the new target condition was quite different from the source. This does not have to take a lot of time, it can be actioned and perhaps even automated.

Regards,

Stephen Marsh.

___________________________________________________________________________

Hi Stephen,

I'm working on a new web page for it now. I'll let the group know when it is available.

Mike Russell

www.curvemeister.com/forum/

___________________________________________________________________________

The total ink limit actions I discussed earlier are available for download:

http: //www.curvemeister.com/downloads/cmyk_tac/index.htm

For those of you who just tuned in, these actions will create a selection containing any areas whose total ink is above a specified limit. The action set contains actions for TIL values from 210 to 340, in 10 percent increments.

It is easy to modify an existing action to reflect any specified TIL value. A "TIL Target" image, which contains a labelled CMYK gradient, is included for testing the actions, and making new actions.

Please let me know of any errors or other problems. I welcome any comments from each of you.

Mike Russell

www.curvemeister.com/forum/

___________________________________________________________________________

Dan Margulis wrote:

Thanks Dan, this works for me. Let me just restate my approach to this to clarify.

To begin, I click on my deep shadows to determine what my Total Ink Limit (TIL) is. (I will try Mike Russell's Photoshop action "trick" soon, thank-you Mike). Some images are between 340-360% TIL and some images are around 300% TIL. They ALL need to be around 320% TIL to get a consistant result on a sheetfed press.

All my images are untagged CMYK. I use the "Convert to Profile" method to get me to the 320% TIL target. I use 2 Source Profiles, one that assumes a 300% TIL and one that assumes a 350% TIL. I leave my dot gain at 20%. My Destination Profile states 320% and dot gain 20%.

My understanding is that I don't need to Assign Profiles to my untagged images. The Source Profile "picks up" whatever my Working Space is. So I change my Working Space for each of my two TIL assumptions. Once the conversion is made, I click my eyedropper on the deep shadows to see if it gets me in the TIL ballpark, at which point I might adjust my black channel in Curves to get me closer to 320%.

If my original image needs major contrast fixes, I convert to Lab and fix it there and reconvert to CMYK. My understanding is that I risk the possibility of color shifts by doing this. But the images always look the same to me after reconverting, so I leave it at that (I have a deadline to meet and another 200 images to go).

There it is. I think this is important to do for the sake of uniformity of shadow density across images. Speaking as a graphic designer, I don't know any of my colleagues who bother with this, as it seems impossibly technical and there seems to be an attitude that "the printer will take care of it".

___________________________________________________________________________

Michael Fedecky wrote:

Michael - every time you convert to another CMYK profile, your CMYK colour settings are being assumed as the descriptor of how these colours are transformed.

You are betting everything on your default profile. A good bet if it is the same as what separated the file or if this is a good description of the output conditions that the file is preped for.

You *may* be dramatically changing the colour appearance from what was originally intended - and you don't realise it or see it! (it may not be that critical, to you or the client etc).

Again, as stated earlier in the thread by both myself and Dan, you can choose to *not convert* out of CMYK and simply edit the offending ink down lower, thus ensuring that the majority of the separation remains as supplied - or you can choose to convert with all the pros/cons that come with such a decision.

Have you tried a single profile? (I don't think that the TIL matters in the source profile when going out of CMYK to the profiles PCS, that is more going into CMYK)

You don't "need" to, but you are making a blanket, one size suits all guess (which may be "wrong") - if there are no complaints, then this may be good enough for your work and workflow.

Photoshop ASSUMES that the working CMYK defines the numbers correctly, as there is no profile assigned (a profile is always there in use).

This is why tagging is there (this method was common in v5), to stop one making a potentially dangerous colour settings change that one may later forget about and hose a different image. If you are used to constantly changing the editing space in colour settings and this extra complexity is not a real issue for you, then it would do the same as just tagging.

Why does it need such "fixes"? How does the file proof/print with no edits or conversions introduced by you?

This is probably because for these images, your ASSUMED working space CMYK profile is not a good description of the numbers...assign or assume a different profile and the result may be better - or worse!

No, that happens when you convert to another mode out of CMYK, Photoshop has to define the numbers somehow, so it does it via the assumed working space for the colour mode that is not colour managed. Choose the wrong profile and the colour will shift. What profile do you then choose? See my earlier posts in this thread!

They look the same becuase the first time you "eyeball" them, they are assumed to be working CMYK and your new conversion is to a similar sapce just with less TIL. Colorimetrically and appearance wise - the stock, inkset, dot gain etc is the same. So no surprise that you don't see any change or problems here!

But if you saw the clients tagged CMYK or the RGB or their prepress proof that indicates what the wanted colour appearnce actually is - you may be very shocked to how your assumption compares to their original intent.

It sounds like you don't get negative feedback for changed colour.

If it is only yourself involved in "signing off" on colour, then fair enough - you are happy. Same for an art director, you make them happy. If a customer, then one would hope they supply a reference print.

Sincerely,

Stephen Marsh.

___________________________________________________________________________

Thanks Stephen, I think I'm getting a handle on this although it hasn't been easy for me to understand this. I think I finally get what "Convert to Profile" is doing and it wasn't until I actually did some tests by comparing neutral values and CYMK builds (before and after converting) that I was fully aware of the colour shifts that take place. My conversions had the effect of introducing a cyan color cast to the mid-tones and lowering magenta in the shadows. The overall contast looked the same, but when I looked more closely at the colors on-screen there was a shift. The amount of detective work involved is just too much work for a large project and images coming from a number of different sources.

As for simply editing the inks up or down to reach the TIL target (maintaining the proportion of CMY to each other) is very difficult especially if the TIL is off by a 15-20%. I'm able to maintain the colors and achieve the 320% TIL but the contrast of the image changes dramatically.

Michael Fedecky

___________________________________________________________________________

On 3/20/07, mfedecky wrote:

Why don't you do the Convert to Profile and then put your original

image on top set to "Color?"

--

J Walton

___________________________________________________________________________

Michael Fedecky wrote:

OK Michael, now we are getting somewhere, I am glad that you are getting a better understanding of the process.

It all comes down to the (assigned/assumed) source profile if you are going to convert. I am of the firm opinion that one should make very careful considerations about converting any existing CMYK file.

When I had to reduce TIL by 40%, I had a choice of a profile conversion that only changed the appearance of the file by say "5-10%" a very minimal colour shift - in a closed loop drum scanning and inkjet press proofing workflow. The majority of the time I did not convert, but manually edited. Less commonly, I did assign and convert the house proofing profile to the house drum scans (that were targeted at the proofing). For foreign supplied CMYK that we did not scan, I did not convert so as to preserve the original seps, I just edited the TIL down.

I mentioned some approaches in the past replies, see below.

One should mask the correction into the deep shadows only, so that lesser shadows and shadow transitions are not adversly affected. Fixed colour sampers can inform you of the ink build and total ink as you edit.

http: //tech.groups.yahoo.com/group/colortheory/message/17070

http: //tech.groups.yahoo.com/group/colortheory/message/17056

http: //tech.groups.yahoo.com/group/colortheory/message/17014

About 5 years ago there was a good thread on reducing TIL too.

Let me know if more explanation is required on targeting and editing

the deep shadows, you may also find the action from Mike to be of some

help too.

Regards,

Stephen Marsh.

___________________________________________________________________________

Michael, the following may help illustrate the point.

There is an sRGB image at this Photo link:

http: //tech.ph.groups.yahoo.com/group/colortheory/photos/browse/be4e

(Low Cyan in Skintones)

sRGB_Reference.jpg (it was tagged before uploading to Yahoo...Try a different browser if you can't see the image).

This sRGB image is to indicate the authors intention for colour (as questionable as it may be). Normally this intent is unknown, you have to guess at the best profile for the mystery meat image.

In the Files section of this Yahoo group, one can find the untagged CMYK file:

4C_Weak_C.jpg

Open both jpeg files into Photoshop, the RGB image is sRGB and the CMYK image is unknown (hint, it was prepared in North America). Your mission, should you choose to accept it - is assume your standard working CMYK colour settings and compare the image to the sRGB reference image. How is the tonal/colour shift? Now you can assign other profiles to come closer to the reference image. Now when you convert out of the current CMYK to another space the colour appearance will be translated.

You should see quite a lot of variation when assigning different CMYK profiles to the mystery meat image.

This is one major reason why many are wary of converting untagged CMYK, when there are other alternatives.

Hope this helps,

Stephen Marsh.