Untitled Document

Hi all I have been in the Adobe Color Management forum and a great guy -

Lou Dina - recommended me this one, so this is me first post.

1- I recently started to work on a newspaper, dealing now with digital cameras from several photographers free-lancers. It seems that compared to scanners the Cyan channel is missing detail or has less "hi-fi" definition. Tried several cmyk color profiles, the renderings with and without the black point compensation. Is this a normal behaviour? Is possible to regain the cyan detail?

2- And another question... the advertising... Some came with 300, 350 and even 400% of Ink limit. We are now converting to the euroscale uncoated v2 cmyk profile that has a limit of 260%. Problem: the K ink is converted to the 4 inks, and we have lots of work and selections to preserve the K based colors. Any ideas?

Thanks,

Diogo Gonçalves

___________________________________________________________________________

This doesn't make a lot of sense to me; There's nothing peculiar to digital cameras that hoses the cyan channel *before* converting to CMYK.

You'll need to post some examples of a "good" cyan image and a "bad" cyan image so we can tell the difference. Just remember not to attach it to an email. To the right of this text should be a link (I hope) that says "Visit Your Group". You'll need your yahoo login to visit the files section and upload the images.

If that's too confusing you can send them *directly* to me (not the list). Make sure they're less than a megabyte or so, in jpeg format.

Typically newspapers I deal with will kick back images that are over their specs. 300, 350 and 400 dmax images are ridiculously high. I'd recommend giving offenders your profile (but CERTAINLY your specs) to make sure they have things converted correctly. Otherwise you can make no guarantees about how the conversion is going to look if you do it for them.

Newsprint conversions always have some sacrifice involved - and the advertiser is the only one who can determine what they are willing to sacrifice in an image and what they are not.

If you have a lot of black-only images (maybe type done in Photoshop?) I suppose you could look into doing a device-link conversion. I think there's a Photoshop plug-in that allows you to do that.

But I personally wouldn't go that direction. Clearly communicate your specs and let offenders know when they are not meeting these. It's their job to make the files right in the first place. If they don't care that much you can do a straight conversion (no masking) and let the ad look bad.

--

J Walton

___________________________________________________________________________

Diogo,

Regarding question 1 – what sort of subject matter are you experiencing this with or is it something you’re seeing right across the board regardless of subject matter?

Darren Bernaerdt

___________________________________________________________________________

Darren, it is my guess that Diogo will reply - skintones (in particular)!

I believe this is a symptom of:

i) Prepress skills being lost and colour work being done by others that traditionally did not do this work.

ii) Over-reliance on the monitor for colour correction over the info palette.

I will be interested to see what Diogo reports!

Diogo, I will try to reply deeper later in my day (I suspect the RBG file is the issue more than the CMYK target).

Regards,

Stephen Marsh.

___________________________________________________________________________

Darren,

I not sure of what you are asking, so I will define better my post.

Most of my previous experience with RGB files are done with scanners. Converted to CMYK the results are normally 4 "well defined" channels. The conversion was made in Photoshop, not in scanner softwares.

Now I deal with digital cameras captures and I started to create a mental pattern: when converting to CMYK I usually have a "blurred" cyan channel, with loss of the detail.

To regain some information I use Channel Mixer in CMYK or a copy/paste of the Red in the Cyan file, using the fade command and, sometimes, the blend modes.

After this explanation, we also have a weekly magazine in the newspaper. I am present when printing both products and I really need a good C channel (specially in the magazine) to achieve a fine balance in the final print. Hard to do with a "blurred" channel.

Regards,

Diogo Gonçalves

___________________________________________________________________________

Hi, Walton

I am posting a doubt and I will post a good and a bad image. No problem.

After the conversion, the C channel seems "blurred" when compared to the others, and the loss of detail compared to the red channel is clear. The "generic" Adobe profiles, like euroscale..., U.S..., FOGRA and others produces this kind of results.

The "offenders" have sometimes difficulties understanding the Ink Limit concept. A good K area is the one made with lots and lots of ink. The device-link... too many input profiles. And who will guess the original profile in a eps with vector based text above an image?

I will now try to post the image.

Regards,

Diogo Gonçalves

___________________________________________________________________________

In "PHOTOS" I posted 2 examples (good and bad) in an album called "CMYK conversion".

Regards

Diogo Gonçalves

___________________________________________________________________________

On 3/13/07, J Walton wrote:

Strike that last statement! What I meant to say was that after looking at his Cyan plate it seems that his Black plate is probably a little too strong. It's hard to tell from just one channel but that's my guess. A strong black (Heavy GCR) + saturated colors + newsprint press = OUCH.

I guess that serves me right for trying to write an intelligent email after midnight. (Come to think of it, maybe the "after midnight" qualifier isn't necessary :-)

--

J Walton

___________________________________________________________________________ 1i.

Stephen/Diogo,

My inclination was that the main issue was with skintones. From the files I deal with from other photographers, it seems that the concept of shooting to keep detail in the red channel is beyond some "professional's" skill level or control. I'm not sure why this is so difficult, however...

When I'm capturing an image of a person, I take care to ensure the red channel does not exceed 245 and often keep it a little lower than that. If it does exceed 245, then I am in the position of having to rebuild that detail in Photoshop. Why not just shoot it properly to start with?

In the case of receiving files later on, there are a few courses of action that can help fix this. 1 - Channel blend to re-build the detail. 2 - Get the RAW files (if avaialable) and use the highlight recovery feature of Camera Raw to get the detail back in the red channel.

As Dan illustrates in the PP books, the cyan channel is the one to give shape and depth to a face. A vacant cyan channel is going to leave a face as a puddle of magenta and yellow.

Darren Bernaerdt

___________________________________________________________________________

Those captures help, Diogo. Thanks.

At first glance I'd say the behavior you are seeing is expected. You are comparing the Red channel of RGB to the Cyan channel of CMYK and noticing less contrast in the Cyan - that's how the process works.

Total Ink Limits (which you mentioned earlier) affect the way the Cyan looks in relation to the Red. In RGB the three channels combine together to describe dark shadows, whereas in CMYK much of the dark shadow weight is carried by the black channel. The C, M, and Y add some weight and color to dark shadows but are reduced in darker blacks because of your Ink Limit.

That being said, I'd say your CMYK profile is a little stronger than I'd expect. If that's true, that can cause you some problems on press - particularly with saturated colors on newsprint. But overall the Cyan plate is what I'd expect. Things that are very blue will have a strong Cyan, but everything else will look a bit muddy.

Here's what's throwing you off, and why I think you are targeting the Cyan channel in particular: The areas you are focusing on are basically shades of Red, which means that they are primarily made of Magenta and Yellow and use Cyan only for shape.

The orange (or grapefruit?) will never have a strong Cyan plate, the bread is a reddish neutral, and the background area is a pale neutral. None of these kinds of colors will have dark Cyan plates.

The image of the man in the blue suit actually will have a dark Cyan (in his suit), but you chose to zoom in on his face, which is basically Red. Pale faces (like mine) have very little Cyan in them, and his face is no exception.

It seems to me that you are serious about color correction, but have some learning to do. Who doesn't? I'd recommend buying Dan's new book, Professional Photoshop something-or-the-other. I'm sure someone will give you a more precise title than that, but from what I can gather of your willingness to analyze images learning Dan's techniques will be of great help to you.

--

J Walton

___________________________________________________________________________ \\

Diogo,

On the first point (channel detail):

I think this is more a difference in the channel structure of RGB vs. CMYK (4C will be lighter in the CMY than the RGB and the K may carry the neutral detail more than expected and thus it is not in the CMY).

That being said, different profiles or third party non-ICC separation tables may engineer more detail into the CMYK channels than others.

How does the composite appear upon conversion, is the detail taking a hit, or is it only in the individual CMY channels that this is so?

On the second point (total ink limits):

Some places reject the ad if not to spec, not all can afford to do this though and may have to do others work for a fee or often free. It is probably separated for better conditions than newsprint, do the colours proof/print too dark (not enough dot gain/tvi) and perhaps a bit impure? If this was separated for newsprint, then I doubt it would have total ink limits from 300-350+.

Advertising has always been 'hands off' for alteration in my experience, one does not alter the colour and the ad meets spec for total ink!

But newsprint is a different game to magazine work, so if the images are prepared for magazines and then sent to a newspaper instead one may get a better result with reseparating. The differences being greater between magazines to newsprint, than magazines to magazines.

Why are you converting/reseparating the entire file, when it is only th deep shadows that are a concern being too heavy in ink? This can introduce lots of fun in the colour management and separation of files. Is it because more than just the total ink needs fixing and a profile transorm is the best place to do this over manual editing?

If the file will print OK except for the deep shadows, one can edit the deep shadows only to conform and leave the rest as supplied. There are two basic ways to do this, one using selective colour-shadows and blend if sliders to isolate the deep tones and create smooth transitions in the edit. Another method uses channel mixer, using the K channel as a 'lever' to reduce the CMY values, again layer blend if sliders are used. One can also use an inverted selection off the K channel as a layer mask (perhaps with more edits to the mask) in addition to the blend if sliders in both these methods, to isolate the deep shadows.

The Euroscale profile is still presuming "purer" print conditions for the separation than is ideal, but if it works for you? Have you looked at results with the free profiles from groups such as:

http://www.naa.org/ (look for SNAP)

http://www.swop.org/ (look for QUIZ)

Hope this helps,

Stephen Marsh.

___________________________________________________________________________

Is this JPEG, TIFF, or raw? You might be looking at the effect of aggressive chroma downsampling in the camera.

Andrew Haley.

___________________________________________________________________________

Darren, the shots from other photographers, are they unedited files straight from camera or have they been 'corrected' first?

When you shoot digital, you keep this new medium in mind. When you edit, you keep the red channel in mind.

Others may not.

My thinking is that with film and scans, the resulting red channel was not always great, but perhaps had more detail than is commonly found today with digital capture. It is likely a mixture of both technology and technique and the fact that more people with less education and experience are taking pictures and also editing/separating pictures than in days gone by when there was only room for pros.

Regards,

Stephen Marsh.

___________________________________________________________________________

Hi all

In our work flow with the images:

1- We try to have as many raw captures as possible. The posted image (the bread) is a raw, with the option "as shot". The image is not edited.

2- The images are only edited by our prepress pros and they are experienced people. We do not rely on monitor colors; instead we use color values since the images are targeted for 3 kinds of media: newspaper white, newspaper salmon and a magazine with bright paper.

3- We are using IFRA profiles for newspapers and a "modified" euroscale coated with a ink limit of 290%, since the magazine is printed on a "thin" 65 gms. paper. This seems to preserve the detail in shadows with a good appearance of the black colors.

The skin tones case: if in a newspaper the K + M are used for detail, in the magazine we need the C ink for good detail. This will give "volume" to a face.

Darren said: "take care to ensure the red channel does not exceed 245 and often keep it a little lower than that. If it does exceed 245, ...". Since I am not an expert, can you be more specific? In the example (bread image) the dark tones are "blurred".

Walton said: "your CMYK profile is a little stronger than I'd expect. If that's true, that can cause you some problems on press"

I posted a cmyk conversion to a default "generic" profile. This way the experiment will be more clear. But I also tried with others and the results seems to be identical.

Walton said: "The areas you are focusing on are basically shades of Red, which means that they are primarily made of Magenta and Yellow and use Cyan only for shape."

True. In the bread, were is the "shape" detail? Walton, I started with Dan in 1995; "color correction, retouching and image manipulation with adobe Photoshop" (john wiley and sons, Inc edition). If you have the book, be kind enough to look at some pages like 148, 188, 224, 246, etc.. There you will find a good C channel example. And I am always trying to learn; that's why I am here. :)

Marsh said: "Why are you converting/reseparating the entire file, when it is only th deep shadows that are a concern being too heavy in ink? This can introduce lots of fun in the colour management and separation of files. Is it because more than just the total ink needs fixing and aprofile transorm is the best place to do this over manual editing?"

I am trying to preserve the color appearance of the add, not the color numbers. But in some parts, like in logos and other corporation colors and in the K text area I preserve the color numbers. A red logo with a M: 100 and Y: 100 can turn to something like C:7, M:95, Y: 90, K:5. Thanks for the great links!

Last but not least: The all point is: the C channel is losing importance, detail or whatever? Are we doing something wrong? The digital machines have a "bug" or this is the best/normal they can do?

I know that euroscale coated has been replaced by the based FOGRA with 350% ink limit. In this new color profile the K ink is well above the others. Is the C channel a looser?

And guys, thanks to all

Diogo Gonçalves

___________________________________________________________________________

Diogo,

When converting from RAW to RGB (I'm referring specifically to ACR), especially with light skin tones, I find it is easy to blow out the red channel. If the white balance and tint sliders are too far off (usually set too far to the right), the brightness too high, or the saturation slider is set too high, the red channel can be very thin. I'm not sure if this is contributing to the problem or not. It might be worth trying another conversion, leaving a lot more "headroom" in the file, so you have a better red channel to start with. Then you will have more visibility as to what is happening in the red channel when working on it in Photoshop. You could even try adjusting the calibration sliders. I used Lee Varis' technique (from his book, Skin) for creating a skin tone calibration and it worked nicely.

If it isn't in the red channel before you convert to CMYK, your cyan channel will be very weak. Perhaps a lower GCR separation will help preserve some of the cyan channel detail.

Lou Dina

___________________________________________________________________________

Diogo,

I forgot to include one more comment. As mentioned by others, you can use Apply Image, Channel Mixer and other techniques on a luminosity layer to beef up the red channel. One other technique that sometimes works well for me (as long as there is sufficient detail and tonal difference already in the red channel) is to add a luminosity layer, then apply a fairly steep curve just to the red channel to add density and more tonal variation. This provides a little more countour and definition to faces. To prevent unwanted extra density throughout the image, I use the Blend If sliders on the underlying red channel to limit the effect. Of course, if you are starting with a very deficient red channel, you may have to resort to blending from other channels, or reconverting the RAW file from scratch to preserve detail.

Lou Dina

___________________________________________________________________________

"Stephen Marsh" wrote:

Stephen -

Both scenarios. If I have the raw and there is no xmp file or ACR settings in a DNG, then I'll start out with the exposure slider at "0". That is my starting point and I'll adjust from there. I really don't understand how a professional photographer can send out a file with blown highlights or blown individual channels. Seems like there is a quality control problem IMHO.

The shoulder of the sensitometric curve with film seems to have a different nature compared to digital. I use one raw converter that appears to have some sort of voodoo under the hood that causes the highlight to roll off very slowly when pushing the exposure brighter. If it worked like film (transparency), adding additional exposure to a correctly exposed image would result in clear film. Digital is not working quite the same way. The net result is a weak red channel in the highlights (assuming we're still discussing a red subject like a face).

The easiest fix to all this (from a photographer's perspective) is to expose properly and keep detail in the red channel's highlight.

Darren Bernaerdt

___________________________________________________________________________

"Louis Dina"wrote:

Hi Lou

Good tip. In fact, the curves + luminosity blend seems to preserve more the C channel.

Meanwhile, K channel also has some improvements. I must try/test this method for a while. That's a start for improvements.

Cheers

Diogo Gonçalves

___________________________________________________________________________

Stephen, I want to be sure I understand exactly what you are doing here with channel mixer to reduce total ink in the deep shadows. If you wanted to remove, say 30 TIL in the deep shadows, would you simply remove 15 points of C from the C channel, remove 15 M from the M channel, remove 15 Y from the Y channel (a removal of 45 points total) and then add 15 K to the K channel? I understand the blend if move to limit it to the shadows, but I wasn't certain of the channel mixer part.

Thanks,

Lou Dina

___________________________________________________________________________

Dear Lou, briefly put:

1. Add a 5x5 fixed sampler in the deep shadows, then perhaps another one in the same spot (set elsewhere then drag over the first sampler) reading as total ink...then you can inspect the edit for colour build and total ink at the same time while editing.

2. Add a new channel mixer adjustment layer, we will only be editing the CMY channels and NOT the black with this edit.

For example, with a 317% file that is being reduced to 300%, I may go to each CMY drop down use the K field at -10. Cyan Output Channel: C=100, M=100, Y=100, K=-10 (negative)...then repeat for M and Y. After this edit, the total ink is at 299% and not 317%.

3. One can then limit the effect to the deep shadows only, so that lesser rich black shadow tones are not affected, or transitions into the shadows are not adversly affected etc.

The limiting is first being done via channel mixer, using the K channel structure to affect the CMY component. Further limiting is done with a mask off the K channel and or layer blend if sliders.

4. K channel edits are usually best done with care and curves for greater control, one may or may not wish/need to adjust the K.

Hope this helps,

Stephen Marsh.

___________________________________________________________________________

Stephen,

Yes, that cleared it up completely. I now understand what you are doing and why. Nice move. It never occured to me to place two samplers over the same point, but reading different values. Pretty cool.

Thanks, Lou Dina

___________________________________________________________________________

Dear Darren, I think the original post was slightly different to our discussion of skintone. But back to our discussion of skintone. The CMYK file was custom tagged with the printers/prepress profile, but has been stripped to protect their identity for this posted example.

I have made a directory in the Photos section of this site titled: Low Cyan in Skintones (sRGB correct colour appearance, browser viewable)

The 4C file can be found in the files section titled:

4C_Weak_C.jpg (unedited CMYK, tag stripped, non browser viewable)

This is what has become of my trade, sigh.

Stephen Marsh.