Topics

Carnival results


Dan Margulis
 

I’ve posted the first pass of our Carnival study as described below.

Reviewing: we are given a silhouetted shot of a person wearing a brilliantly red costume. The file is in Adobe RGB, but unfortunately it is to be printed, specification being FOGRA27. An ice cream cone in a blast furnace would have a better chance of survival than we have of matching that bright red in CMYK. The question is how can we make it as red as possible under the circumstances, while holding enough detail to be convincing, yet not purtting in so much that it seriously detracts from the redness.

We have 19 entrants, some people submitting more than one version. Everyone also submitted a list of their steps, thanks very much. I haven’t read these, because I’d rather get a sense of who was successful and who wasn’t before investigating why. 

The files don’t have people’s names on them, and were random-generator numbered from 201 to 219. There is also version 200, which is a straight default conversion. There is also a “par” version, #220. To get it, I quickly chose what I thought looked like the five best entrants, and I averaged them, each one weighted 20%. This often creates a version that is superior to most if not all of its parents; I haven’t checked to see whether that’s true in this case.

In viewing these CMYK files, remember that you must honor the FOGRA27 profile. I’m not sure that they will view accurately within groups.io, you may need to download them. For convenience, then, I also made a folder containing the files converted back to Adobe RGB without Black Point Compensation. You may find this easier to browse. Also you can get a direct idea of how much has to be lost even in the best scenario by comparing the original to the par version with both in Adobe RGB.

People can comment on these right now. I haven’t yet formed firm opinions. When I do I’ll have something to say. Meanwhile, if you’d like to know how your own version stacked up, download the par version and compare the two directly. Do you think you got the same kind of quality? If not, I hope you’ll find further discussion useful.

Both folders are in the group Photos section, named Case Study: Carnival. The CMYK folder is

and the Adobe RGB folder is

Dan Margulis


Roberto Tartaglione
 

First of all I would like to thank anyone who shared their pictures in this case study.
It was hard for me to afford this conversion, almost for 2 reasons:
1) the brilliant and saturated color (red and yellow) to be converted
2) the kind of textile: it is glossy and the conversion should retain this aspect; what I would achieve are deep and neutral shadows + brilliant highlights.
IMHO one critic area is the red gag under the black mask: the glossy is “softer” related to the whole garment ( maybe is another kind of fabric) and it is the more penalized in the conversion in CMYK
Any help from you has therefore been most welcome.

If I should assume, the five entrants that contributed to the 220 par are: 210 - 201 - 203 - 206 - 207
Probably the 220par is the best representation of the original file, the 210 has a very good outline too but I prefer the color of 220par or 203 and 206.
In other words, if I were allowed to mix two pictures only, my choice would be 210 (for the structure) +203 (for the color)

At a glance I think the main difference  between 220par /210/203 is in the structure of the C channel, and more,
I see that in the 220par not any channel (M and Y) raises 100% - I ask: is this an indispensable condition to achieve a good result?-
I prefer to leave this deep investigation to more experienced members in CMYK.

All the best

Roberto Tartaglione



Dan Margulis via groups.io <dmargulis@...> wrote:

I’ve posted the first pass of our Carnival study as described below.
…………………..
Both folders are in the group Photos section, named Case Study: Carnival. The CMYK folder is

and the Adobe RGB folder is

Dan Margulis


ROBIN MARK D'ROZARIO
 

Hi,

Re the gag below the black mask, a suggestion.
1.Duplicate the layer and desaturate.
2.Change the blend made of desaturated layer to linear light/hard light and apply a black mask to layer.
3.Paint with white at opacity and flow of your taste onto the mask in the area of the gag.Adjust the opacity of layer to your taste.
It adds some more contrast and apparent glossiness/shine to the gag.

Regards,
Robin


ROBIN MARK D'ROZARIO
 

On further trial using Screen mode iat 20-25% opacity nstead of Linear/Hard Light gives more shine.
Robin

On Sun, May 17, 2020 at 3:51 PM rdrozario via groups.io <rdrozario=gmail.com@groups.io> wrote:
Hi,

Re the gag below the black mask, a suggestion.
1.Duplicate the layer and desaturate.
2.Change the blend made of desaturated layer to linear light/hard light and apply a black mask to layer.
3.Paint with white at opacity and flow of your taste onto the mask in the area of the gag.Adjust the opacity of layer to your taste.
It adds some more contrast and apparent glossiness/shine to the gag.

Regards,
Robin


George Machen
 

I didn't do the exercise, but want to comment on just one thing. When I looked at the CMYK contributions, many stood-out to me with cyan-ish cast folds of the robe, presumably to give the appearance of gloss. The cyan casts there really bothered me. At first I thought the casts were simultaneous contrast from less saturation at the folds, but when I inspected the RGB counterparts, the folds were lighter, but still red, yet still suggested gloss, so I thought they looked much better. I wonder whether the cyan-ish casts could be avoided in the CMYK versions if the default conversion cyan plate were left alone, then a separate beefed-up cyan plate were blended into the composite in Luminosity mode.

-- 
George Machen
 


Dan Margulis
 



On May 17, 2020, at 7:53 AM, George Machen <gmachen@...> wrote:

I didn't do the exercise, but want to comment on just one thing. When I looked at the CMYK contributions, many stood-out to me with cyan-ish cast folds of the robe, presumably to give the appearance of gloss. The cyan casts there really bothered me. At first I thought the casts were simultaneous contrast from less saturation at the folds, but when I inspected the RGB counterparts, the folds were lighter, but still red, yet still suggested gloss, so I thought they looked much better. I wonder whether the cyan-ish casts could be avoided in the CMYK versions if the default conversion cyan plate were left alone, then a separate beefed-up cyan plate were blended into the composite in Luminosity mode.

A suggestion to all: in commenting on a certain defect it’s best to cite one or more particular numbered version that shows it. Nobody will be offended because the entries are all anonymous.

This also helps clarify the intent of the criticism. This gown is so overwhelmingly red that a cyan cast as such isn’t possible—the cyan would desaturate the red first. The black also serves that purpose and I don’t favor one more than the other.

Somewhere in this garment there has to be 0c100m100y0k or something very close. Pretty much everybody appreciated this point, somewhat to my surprise.

The problem is that we have to make it look like a real garment and not a red blob. To do that we have to make some folds darker, by adding cyan and/or black, and/or make other folds lighter, more reflective, by reducing magenta and yellow. Unfortunately both of these things diminish redness, so they have to be confined to small areas. Otherwise, even if the maximum red is there someplace, overall the image seems grayer. So we get things like #204 (great detailing but so much darkening that the overall red impression is reduced) and #206 (great shape by lightening certain folds, but again not seeming as red as it should).
Both of these corrections would be judged spectacularly successful in many other images, but this one is a special case.

Dan


Gerald Bakker
 

I find it very hard to choose a favorite. All versions are good to very good, none really stands out. I had to do a one-by-one comparison to pick out my own submission. I agree with Roberto, the par version is very good (as expected, because it's the average of five of the best ones).
 
The main distinguishing factor is the amount of detail brought into the reds and yellows. Not enough detail, and texture is lost. Too much detail, and color is lost.
The question then is, do we want a version that has so much detail that the structure of the fabric leaves no doubt at all (e.g. #206)? Or should we strive for a version that has stronger color, and just enough detail to discern what we are looking at (e.g. #218)? I prefer the latter, and it's how I understood the exercise: as colorful as possible, without losing believability of the subject. Looking back at the original, I notice that it doesn't show that much detail either.

So my top-5 is 218, 214, 211, 205 and 203. Interestingly, of these five, only 203 is also in Roberto's list.
--
Gerald Bakker
http://geraldbakker.nl


Roberto Tartaglione
 

Thank you Robin,
this technique does really help in lightening and adding contrast to folds. Never done before.

Regards
Roberto
roberto@tartaglione.com

ROBIN MARK D'ROZARIO <rdrozario@gmail.com> wrote:

On further trial using Screen mode iat 20-25% opacity nstead of Linear/Hard Light gives more shine.
Robin


Dan Margulis
 



On May 17, 2020, at 10:08 AM, Gerald Bakker <gc.bakker@...> wrote:

I find it very hard to choose a favorite. All versions are good to very good, none really stands out. I had to do a one-by-one comparison to pick out my own submission. I agree with Roberto, the par version is very good (as expected, because it's the average of five of the best ones).
 
The main distinguishing factor is the amount of detail brought into the reds and yellows. Not enough detail, and texture is lost. Too much detail, and color is lost. 
The question then is, do we want a version that has so much detail that the structure of the fabric leaves no doubt at all (e.g. #206)? Or should we strive for a version that has stronger color, and just enough detail to discern what we are looking at (e.g. #218)

With an unusual image like this one it’s hard to answer that question in advance; it becomes clear only when the choices are seen. I agree with all the above points

So my top-5 is 218, 214, 211, 205 and 203. Interestingly, of these five, only 203 is also in Roberto's list. 

It goes to show the truth of your statement that all are good and none is an obvious standout, although I think some could be eliminated without much discussion.

On my list of five that created the par image, all three of us agree on #203. I also agree with Roberto about #210, and I agree with you about #211 and #214. Thus I am two for five matches with Roberto and three for five with you. My fifth pick, not shared by either of you, is #209.

Dan


Bruce Jamieson
 

My thoughts!

Lab color easily can describe a color that is simultaneously paper-white and red as a stop-sign. But out here in the CMYK world we have no such luxury. So we must pick what is the most saturated CMYK build of a color can be, and sacrifice in one of two ways: either keep the color as saturated as possible, harming the shadow and highlights, and flattening the shape of the object, or: Keep the shadows and highlights, or even emphasize them, knowing that the overall saturation will drop. On a real job, I might query the client about this matter, since it is apparent that this robe’s representation with ink and paper must reflect the client’s emotional idea of what it should be. Communication is key. Presenting several versions from which they could pick would be a great tool to bridge what is aspirational to what is physical.

Some notes not directly about the red robe:

Skin colors or neutrals shifted in almost all of them, in my opinion for the negative. Some hands begin to blow out, some get old and craggly-looking. Since these colors were never out of gamut, except maybe where the robe reflects the strongest on them, they were the most accurate using a straight Relative Colorimetric (or the Default) conversion. The robe was always the point of the assignment, so by doing the acrobatics required to achieve this has turned the rest of the subject into the problem.
Those immune to this problem: 203, 206, 208, 217

Drop shadow banding occurs in many of the images. This area needs help to begin with: on a real job I would have completely cleaned up the shadows and assured their neutrality, but in this case I took them as if they were desired. However, many of the solutions also made this area far worse.

FOGRA27 has a total ink limit of 350. We can make the decision to go over that mark, but I try to respect the profile’s rules unless I have a specific reason. Going over this limit without regarding it is a classic error and may lead to some very soggy pages. The black areas which are usually the problem are thankfully not too big, so maybe it won’t be a problem. Maybe check with your printer before trying it. Specifically looking at the darkest areas of our image, these images go over the total ink limit: 202, 205, 214

My individual notes on this matter are below. I am not posting my favorites because I strongly suspect it's mine, and a vote for myself feels disingenuous. 

200 Default: Reds have great saturation but obviously sacrifices too much detail. Besides that, I’m using this a reference for yellows, hand, foot and drop shadow.
201: Nice robe shapes but it looks kinda dark and I’m not seeing any max saturation areas.
202: Red has a nice subtle contrast and shows the shape well. Yellows are exciting but moves away from RGB. Neutrals are yellow
203: Good saturation, subtle highlights and shadows. Neutrals stay that way.
204: Too dark, blurry for some reason, banding in shadow, yellowish skintone.
205: Great saturation but much is lost in the highlights, and may be unacceptably so. Some saturation bands occur between highlights & saturation. Saturation, sure, but at what cost? Banding in drop shadows. Hands look pleasant, maybe a touch pink in the highlights.
206: Strong representation of the robe shape, not very saturated. Neutrals stay that way.
207: Great robe shapes, maybe a bit cooler than they should be. Hands look cold and dead.
208: Great saturation, robe highlights get lost a little. Neutrals stay that way. Nice!
209: Yellow cast, nice robe shapes. Hands look grey and dead.
210: Pow! Has both saturation and contrast but maybe a bit too strong for some clients. Cool cast, most noticeably in the highlights. Hands & shoes also get that pop, which I don’t love. 
211: Nice soft look to the robe and looks pretty much on par. Hands look grey and dead, mask is showing on pant leg.
212: Strong yellow cast, blue banding in drop shadow. Robe shapes are understated but never seems to get 100% saturation.
213: Too light, yellow cast. Smooth robe shapes, but some saturation banding seems to occur.
214: Hits the saturation, nice robe shapes. Very nearly par!
215: Robe shapes are too soft, hands are getting some strange colors.
216: Has that exciting contrast pop. Yellows (and to a lesser extent the reds) aren’t as saturated as they could be, hands are grey and dead.
217: Neutrals and hands are good, yellow has a nice pop. More saturated than par (sacrificing robe shapes,) and looks nice! Black gradient wasn’t part of the assignment.
218: Good saturation and nice robe shapes. Hands greatly lose and look like putty flippers.
219: Overall lighter than par. Good saturation, good shape. Hands lighten and add saturation, which is nice to match the rest, but shoes and drop shadow gain a yellow cast.
220 Par: Great robe shapes and saturation, and I expect we may come to the consensus that for the red areas the Par version works the best. However, I prefer the rendering of all the other areas from the 200 Default conversion!


Jim Sanderson
 

Thanks very much for the exercise!!

, with purpose and great thought. 

On May 17, 2020, at 8:05 AM, angelhairpasta177 <flapjack@...> wrote:


My thoughts!

Lab color easily can describe a color that is simultaneously paper-white and red as a stop-sign. But out here in the CMYK world we have no such luxury. So we must pick what is the most saturated CMYK build of a color can be, and sacrifice in one of two ways: either keep the color as saturated as possible, harming the shadow and highlights, and flattening the shape of the object, or: Keep the shadows and highlights, or even emphasize them, knowing that the overall saturation will drop. On a real job, I might query the client about this matter, since it is apparent that this robe’s representation with ink and paper must reflect the client’s emotional idea of what it should be. Communication is key. Presenting several versions from which they could pick would be a great tool to bridge what is aspirational to what is physical.

Some notes not directly about the red robe:

Skin colors or neutrals shifted in almost all of them, in my opinion for the negative. Some hands begin to blow out, some get old and craggly-looking. Since these colors were never out of gamut, except maybe where the robe reflects the strongest on them, they were the most accurate using a straight Relative Colorimetric (or the Default) conversion. The robe was always the point of the assignment, so by doing the acrobatics required to achieve this has turned the rest of the subject into the problem.
Those immune to this problem: 203, 206, 208, 217

Drop shadow banding occurs in many of the images. This area needs help to begin with: on a real job I would have completely cleaned up the shadows and assured their neutrality, but in this case I took them as if they were desired. However, many of the solutions also made this area far worse.

FOGRA27 has a total ink limit of 350. We can make the decision to go over that mark, but I try to respect the profile’s rules unless I have a specific reason. Going over this limit without regarding it is a classic error and may lead to some very soggy pages. The black areas which are usually the problem are thankfully not too big, so maybe it won’t be a problem. Maybe check with your printer before trying it. Specifically looking at the darkest areas of our image, these images go over the total ink limit: 202, 205, 214

My individual notes on this matter are below. I am not posting my favorites because I strongly suspect it's mine, and a vote for myself feels disingenuous. 

200 Default: Reds have great saturation but obviously sacrifices too much detail. Besides that, I’m using this a reference for yellows, hand, foot and drop shadow.
201: Nice robe shapes but it looks kinda dark and I’m not seeing any max saturation areas.
202: Red has a nice subtle contrast and shows the shape well. Yellows are exciting but moves away from RGB. Neutrals are yellow
203: Good saturation, subtle highlights and shadows. Neutrals stay that way.
204: Too dark, blurry for some reason, banding in shadow, yellowish skintone.
205: Great saturation but much is lost in the highlights, and may be unacceptably so. Some saturation bands occur between highlights & saturation. Saturation, sure, but at what cost? Banding in drop shadows. Hands look pleasant, maybe a touch pink in the highlights.
206: Strong representation of the robe shape, not very saturated. Neutrals stay that way.
207: Great robe shapes, maybe a bit cooler than they should be. Hands look cold and dead.
208: Great saturation, robe highlights get lost a little. Neutrals stay that way. Nice!
209: Yellow cast, nice robe shapes. Hands look grey and dead.
210: Pow! Has both saturation and contrast but maybe a bit too strong for some clients. Cool cast, most noticeably in the highlights. Hands & shoes also get that pop, which I don’t love. 
211: Nice soft look to the robe and looks pretty much on par. Hands look grey and dead, mask is showing on pant leg.
212: Strong yellow cast, blue banding in drop shadow. Robe shapes are understated but never seems to get 100% saturation.
213: Too light, yellow cast. Smooth robe shapes, but some saturation banding seems to occur.
214: Hits the saturation, nice robe shapes. Very nearly par!
215: Robe shapes are too soft, hands are getting some strange colors.
216: Has that exciting contrast pop. Yellows (and to a lesser extent the reds) aren’t as saturated as they could be, hands are grey and dead.
217: Neutrals and hands are good, yellow has a nice pop. More saturated than par (sacrificing robe shapes,) and looks nice! Black gradient wasn’t part of the assignment.
218: Good saturation and nice robe shapes. Hands greatly lose and look like putty flippers.
219: Overall lighter than par. Good saturation, good shape. Hands lighten and add saturation, which is nice to match the rest, but shoes and drop shadow gain a yellow cast.
220 Par: Great robe shapes and saturation, and I expect we may come to the consensus that for the red areas the Par version works the best. However, I prefer the rendering of all the other areas from the 200 Default conversion!


Paco
 

" I found problems with a group of pictures regarding the costume of "Domino", a deep red (more yellow than magenta) that converting in CMYK as you often emphasise, loses almost all its shape.

I’m not interested in the exact color reproduction, my book (if I Will succeed in my project) will never be a fabric catalog, 
what I want is the feeling of the costume, his shape indeed."

Take notice to what the client says he is looking for: "
not interested in the exact color" and "I want is the feeling of the costume, his shape indeed." and "red (more yellow than magenta) which, since no #'s are given, it is "orangish."

When finally going to paper... do not tonalities and contrast darken in an exact amount of "a couple of skoush's.?"

I'll lift a toast to all of us who did do it! 


Paco
 

Exactly how I felt once I thought I was done with it. I masked and tweeked M. Make me feel OK that you caught it too!


Roberto Tartaglione
 

Thank you angelhairpasta177
for your in-depth analysis of all the files; I almost agree with all of your choices except for #208 and #214.
I’m not sure as you say that the default conversion (200) is preferable for the hands and all the other areas but 
is true that maybe I was too much concerned by the robe and neglected the rest.

About  the ink limit, I try to never exceed the value of 300; but  in this case in my attempt,
the shadow of the black mask is 321: is a very small area it should not be hard to print.

I think that we all agree that affording images like this, is not only a matter of personal choice but also a compromise between what we want and what we can.
That’s more true if a client is involved when working on assignment. 

As Gerald Bakker says:
>>>>"The question then is, do we want a version that has so much detail that the structure of the fabric leaves no doubt at all”? 
"Or should we strive for a version that has stronger color, and just enough detail to discern what we are looking at”?

Or in other words, according to Dan Margulis:
>>>>>"The problem is that we have to make it look like a real garment and not a red blob. To do that we have to make some folds darker, by adding cyan and/or black, and/or make other folds lighter, more reflective, by reducing magenta and yellow. Unfortunately both of these things diminish redness, so they have to be confined to small areas. Otherwise, even if the maximum red is there someplace, overall the image seems grayer."

Regards
Roberto 
angelhairpasta177 wrote

>>>>220 Par: Great robe shapes and saturation, and I expect we may come to the consensus that for the red areas the Par version works the best. However, I prefer the rendering of all the other areas from the 200 Default conversion!


Beat C
 

I have found the results on the website - quite impressive.
Are the results not downloadable? I want to compare my sorry attempt with them :-)
Preferably the complete folders in one go?

Thanks, Beat


Harvey Nagai
 


My thanks to Mr. Margulis for conducting this case study for us, and to
Mr. Tartaglione for providing the example.

I put most of the emphasis on Mr. Margulis' objective: "make it as colorful
as possible in CMYK", followed by "subject to retaining enough detail for
believability".

I also paid heed to Mr. Tartaglione's description "(more yellow than magenta)". 
However I couldn't reconcile "deep red" with the original image, so I made
allowances for darker reds when reviewing the submissions (but which I did not
with mine).

With that in mind, 209 emerged has my choice as the most "colorful" in
head-to-head comparisons with each of the others. Granted the reds are
a bit too yellow, particularly the glosses, but with a bit of masking
that can be reduced by blending in the original color. It also has enough
shape and detail to look like synthetic costume material rather than
a fabric.

Reading the previous comments, I obviously had a different interpretation
of the problem than Mr. Tartaglione. C'est la vie, and kind of pointless
to rank the submissions further.

Looking at them as a whole, I'm a bit surprised that there was so little
variation, maybe that was the nature of the problem, or at least our
interpretations of it.

And I have to say I was a bit taken aback by how poorly Mr. Margulis' version
converted back to RGB (without black point compensation). Much of the reds
became two-toned, light and dark with the nuances of the cmyk version nearly
gone. Whereas the par version merely got a bit flatter, retaining transitions
from lighter reds to darker reds much better.

I do like the way the cymk version appears to have some "glowiness" to it,
and that seems to have been smushed out in the conversion back to rgb.

My understanding of cmyk is limited, but it looks like the channels are mostly
lights and darks with very little in the midtones, and the conversion
exacerbated this in the rgb channels?


Roberto Tartaglione
 

Hi Beat,
Of course you can download the files; just go to the link provided by Dan Margulis
https://groups.io/g/colortheory/album?id=246660 (CMYK versions) select one by one the Thunbnails 
and click to the download icon (here is an example). I don’t think is possible to download the whole folder.


Kind regards 
Roberto Tartaglione



 Beat C <b.cornaz@...> wrote:

I have found the results on the website - quite impressive.
Are the results not downloadable? I want to compare my sorry attempt with them :-)
Preferably the complete folders in one go?

Thanks, Beat



Dan Margulis
 



On May 17, 2020, at 5:49 PM, Harvey Nagai via groups.io <hnagai@...> wrote:

And I have to say I was a bit taken aback by how poorly Mr. Margulis' version 
converted back to RGB (without black point compensation). Much of the reds 
became two-toned, light and dark with the nuances of the cmyk version nearly 
gone. Whereas the par version merely got a bit flatter, retaining transitions 
from lighter reds to darker reds much better. 

I do like the way the cymk version appears to have some "glowiness" to it, 
and that seems to have been smushed out in the conversion back to rgb.

I’m not sure which version you’re referring to but I have to wonder about a profile mismatch. I looked at the par version #220 just now in both RGB and CMYK and don’t see a difference offhand.

My understanding of cmyk is limited, but it looks like the channels are mostly 
lights and darks with very little in the midtones,

Correct. That’s how CMYK produces very intense, saturated colors.

and the conversion 
exacerbated this in the rgb channels?

Two factors in play here:

1) the black channel disappears and must be merged into the red, darkening it; and to a limited extent lightening the green/blue as well.

2) Adobe RGB has a much larger gamut in reds than CMYK (which is why we could never match the RGB original). Therefore even without the influence of the black channel being added, the red channel has to be much darker than the cyan of CMYK. Otherwise, the resulting color would be half a kilometer outside of the CMYK gamut and we’d be back in the same mess we were to begin with. For the same reason the green channel of RGB must be slightly lighter than the magenta of CMYK, in this particular case. In more everyday images we rarely see such an effect.

Dan