Convert from RGB > CMYK

Tech

Well-known member
Hi All,
Another long discuss came up regarding making correction in RGB vs CMYK. The current practice by our specialist is perform all changes/corrections in RGB with printer profile turn on in Proof Colors. He makes corrections based on the soft CMYK proof mode with his calibrated monitor. So unless instructed on saving as CMYK with profile, he usually leaves final files in RGB without printer profiles.

1) What is the point in making corrections in soft proof color mode and not go ahead making final CMYK conversion and embedding the file? He kept talking about dropping colors if we make the conversion—no kidding, even if color is converted by vendor we'll still lose colors due to gamut and smaller color space in CMYK. IMHO, this is a CYA practice—if printed color proofs returns wrong, he/someone can blame the printer.

2) He also uses Convert to Profile when he is requested to assign and embed printer profile. Now, correct me if I'm wrong, doesn't this method actually changes image color info just like converting from RGB to CMYK mode. The difference here is that Convert function maintains color visually from RGB to CMYK whereas a straight change from RGB to CMYK does not. Hence people thinks this is the accurate method of converting RGB > CMYK?

This brings up my point of working in CMYK mode instead of RGB if you already have printer profile. I tested all methods, with 3 files (RGB with proof mode, converted CMYK, RGB > CMYK) setup on screen and zoomed in 3200% to compare color pixel info, the same pixels in all three files clearly displays different color info. I see no clear advantages from working in RGB-color proof mode, unless one wants to keep an RGB layer version and a final CMYK flattened version.

Am I crazy or is this RGB workflow a damn pain created by photographers wishfully thinking this is the best method on translating their work into 4-color printing?
 
>1) What is the point in making corrections in soft proof color mode and not go ahead making final CMYK conversion and embedding the file?

I'll refer to Dr. Abhay Sharma's analogy. If you have a box of 64 crayons, and drop down to 32 crayons, you'll never be able to get back to 48 crayons (snapping in half doesn't count. In short, when converting to CMYK you commit the file to a specific process that may or may not be appropriate to the final output, and you can't go back to where you were.

>2) He also uses Convert to Profile when he is requested to assign and embed printer profile.

If we're talking RGB, you can't assign a CMYK profile to an RGB image, so conversion would be appropriate. Converting will maintain the look of the orginal in the color space of the output profile.

>I tested all methods, with 3 files (RGB with proof mode, converted CMYK, RGB > CMYK) setup on screen and zoomed in 3200% to compare color pixel info, the same pixels in all three files clearly displays different color info.

Are you sure you're using the same profile/rendering intent for all, because this isn't my experience?

An RGB workflow has advantageous and disadvantages (from a workflow perspective if you're unaccustomed), but on the whole it offers the ability to repurpose the image for a much broader output intent. So if you don't know the intended output, keeping the images in RGB seems perfectly viable, if not desireable to me.
 
1) Yes, we are aware of CMYK conversion is final. The point I'm questioning is passing corrected RGB files to printer/prep house and having them make the final conversion instead.

2) In RGB file, setup Proof Setup > turn on Proof Colors > your instant preview of the profile on RGB and that's what he is basing his corrections on, screen soft proof.

3) I'm certain I have 3 test files setup correctly side-by-side with RGB with Proof Color (using printer profile) vs RGB Convert to Profile CMYK vs RGB straight to CMYK—color info indeed changed in info palette—we are not talking about major color shifting but noticeable depending on your monitor.

As noted, aside from re-purposing files, what other RGB advatange is there? I'm in book publishing, all final products will be CMYK. It's not like we reuse the files for web content.

The frustration is doing all the extra work, driving everyone crazy with color management and realizing a lot of it is just overkill.
 
Last edited:
Definitely Wrong

Definitely Wrong

Sending RGB files out for production is not a good policy. There will always be color out of gamut and you will undoubtedly suffer from some color shift when converting. This shift will happen at some point in the files journey from screen to print - why not see and correct these issues before you dump it into someone else's lap. This seems to be merely a way of passing the conversion off on someone else when it could and should be done by the person charged with the color correction. The best policy would be to correct the images (in RGB if preferred) and then convert them to CMYK before saving them for use as production files. This way the operator can see if any undesired color shift has occurred and make the necessary corrections in CMYK before the files go to print. Some people may open and re-save RGB files in Photoshop, some may let their RIP convert the art, or Acrobat, or Pitstop, etc. Considering the different ways RGB images can be converted to CMYK, and considering that they all may render different results; shouldn't the art go out as CMYK to begin with so as to reduce unwanted alteration.

In addition, how would this RGB workflow handle print ready PDFs? If you were to create a print ready PDF with your RGB art embedded the printer could not open the art to convert it in Photoshop. This further limits the printer's ability to successfully convert the RGB to CMYK. All around sending out RGB artwork is a bad idea.
 
The method the client used is fine provided that the RGB image has the correct ICC

The method the client used is fine provided that the RGB image has the correct ICC

There are however different RGB spaces. Where the ECI is the one that has most close match to coated. (Adobe RGB has a gamut missmatch in intense yellows wich makes out 100% CMYK yellow unacheivable in Adobe RGB) and RGB is 75% of file information.
The advantages of working in RGB are many, for one many filters work much better in RGB. If using adjustment layers you can leave the photographers original data in original values, so that all conversions are done in one step.
Less maths means less risk of bading.
RGB has a one colour has one value, where as CMYK has any colour that is not primary or secondary has numerous possible definitions. And then comes the flexibility, CMYK is device/substrate dependant, RGB is not.
The conversion is dependant on the CMM so you will need to do the conversion from RGB to CMYK in an Adobe Colour Engine.
To leave the colours in RGB letts you make late decisions in paper or process, also it gives the freedom of the images being ready for cross media. The converting to CMYK would then be either in an Adobe PDF print engine or in saving the layout for PDFx1a in InDesign. The person preparing the layout has the responsibility.
In of gamut colours are converted to device cmyk if the settings are correct. Gamut warning identifies what colours to take note of as not printable but the proofer preview will show you what the file will look like if those colours are not brought in gamut (by selective colour editing)
In extreeme images a file can be converted to a smart object and then converted to CMYK giving advantages of both worlds at the price of disk real estate and processing time.
It is a perfectly safe workflow.

It is much harder to fix a badly made CMYK image than an RGB image.
 
@Mike,
I wholeheartedly agree on not sending out RGB files to prep house/vendor. I'm against it from start and still am. Unfortunately, the practice setup here is not for me to change and we are treating case by case as we go. Sometimes we send out CMYK files and other times we send our RGB files or even a mixed batch.

There is a misunderstanding here that somehow CMYK conversion is better when perform downstream when we don't have vendor's profile. How is it better when we have no idea how vendor's profiles were setup and whether is created based on the ink/press/paper our job would run on?

To which I suggested a color-corrected untagged CMYK is fine and can be assigned/tagged downstream. Specific vendor profiles ultimately doesn't vary a great deal from one to another. With expectation that color shift/lose is inevitable, out weighs unknown downstream conversion from RGB in the varies methods you mentioned.


@Lukas
You are absolutely correct that making correction is RGB is more flexible than in CMYK. I agree with this under three conditions: 1) absence of target vendor profile 2) your final product will not be printed 3) you expect further changes at later date or re-purpose the same content--this leads to saving 2 or more versions of the same file (one in RGB, one in CMYK), then storage becomes another issue.

This CMS-RGB-CMYK topic comes up every four to six months, I had already voice my opinions, but learned to avoid talking about it because change is not what people here want to see. They are comfortable with RGB workflow, knowing they can place the responsibility on someone else downstream. As you can see, I do not share the same views because I see minimal benefits for a book publisher that will always be limited by CMYK.

Lastly, I like to apologize for ranting--it's just one of those days. I rather they stop coming to me for advice regarding color profile and CMS every time they find something they done and don't like the results. They listen but don't make any changes from my suggestions, then why waste my time and theirs?
 
Proof setup - proof colors and rgb-CMYK from the drop down menu will both use the profile/rendring intent and cmm from your color settings perfrences. If you covert using these same settings, the results will be numerically indentucal. If you are seeing differences, than something isn't set up correctly.

I also do not reccomend sending out rgb to the printer for some of the raspna mentioned. But Rgb workflows can be a number of different things, and keeping images at least archived in a repurposable state simply makes sense to me.
 
Proof setup - proof colors and rgb-CMYK from the drop down menu will both use the profile/rendring intent and cmm from your color settings perfrences. If you covert using these same settings, the results will be numerically indentucal. If you are seeing differences, than something isn't set up correctly.

RGB file
Proof Setup: US Sheetfed Coated
Proof Colors turn on

RGB Convert to Profile CMYK file
Same settings with file converted to the actual profile, visually looks identical but @3200 pixel info changed and is not matching

If this is wrong, please point out what I need to correct. I want to make sure I'm not going insane and jumping to wrong conclusions.
 
Love what artificial intelegence can come up with, so your client is abot to beasure 3200 pixels to get their values? Do you get a delta-e on how much the pixels vary too? What number of pixels are you comparing? If you can find in your production run less than 3200 pixels in your final print with delta-E = 0 you are living on another planet in another universe ;P

There are different approaches and key is just that all those that will be involved in the production are in agreement. That the issue comes up is not a problem, technologies change and we need to keep testing to find if yesterdays bestpractices are still valid today. On the CMYK argument if you do produce the same material with a FM or AM (linescreen or stochastic) raster you will need different CMYK values on the same substrate.

Your stating:
There is a misunderstanding here that somehow CMYK conversion is better when perform downstream when we don't have vendor's profile. How is it better when we have no idea how vendor's profiles were setup and whether is created based on the ink/press/paper our job would run on?
Is doubble edged.
What is it you want in an image? A physical colour on that is reflected by the paper or a numerical representation on an intermediate by product (the print plate)? If you do decide to stick with the CMYK (old school- prooven trackrecord) then I would recommend that you send a PDFx1a with the output intent to what you proofed the image to (as far as I can read US Sheetfed Coated).
If not you are giving the printer the bossibility to vary the appearance of mid tones by 10% (due to the differences in allowed tollerances in ISO standard between different production techniques).

I will not argue with a PDF/x1a that has a output intent. If that is what you are giving then you will be safe.
 
Are you using dithering with your CMYK conversion? Proof set up doesn't preview with dithering and pixels would be different between a dithered and non-dithered conversion. In fact, two dithered conversions could have non-identical pixels all else being equal. You normally wouldn't notice this except at high magnification, such as 3200%.
 
2) He also uses Convert to Profile when he is requested to assign and embed printer profile. Now, correct me if I'm wrong, doesn't this method actually changes image color info just like converting from RGB to CMYK mode. The difference here is that Convert function maintains color visually from RGB to CMYK whereas a straight change from RGB to CMYK does not. Hence people thinks this is the accurate method of converting RGB > CMYK?

I think you may be confused on this point. Convert to Profile... performs EXACTLY the same function as Image/Mode/CMYK (or as you put it "straight change from RGB to CMYK"). Convert to Profile allows you to choose a destination profile, CMM (engine), rendering intent, Black point compensation, & dithering when you convert from RGB to CMYK. Image/Mode/CMYK does the same conversion, except that it picks up all of the above options from your Color Settings (under the Edit menu). So if you choose exactly the same options in both places, you will get the exact same CMYK numbers. Both methods can be accurate if you use the correct settings.

On a similar note, your Proof Colors option should show you the exact same CMYK numbers as long as ALL options are the same as in Color Settings and/or Convert to Profile. Mike Eddington mentions how dithering could create some discrepancies at 3200%, but I have such a set-up on-screen before me and the largest difference I can find on any given pixel is one percentage point in one channel. Most pixels read identical. If you are getting different values from these 3 methods, then your settings are not identical.

There is a misunderstanding here that somehow CMYK conversion is better when perform downstream when we don't have vendor's profile. How is it better when we have no idea how vendor's profiles were setup and whether is created based on the ink/press/paper our job would run on?

If you do not know what ink/press/paper a vendor used when creating a profile, then it is indeed worthless. It's possible for CMYK conversion downstream to be better (as in more color accurate), but only if the vendor uses a profile created specifically for the printing condition they will be using to print your job. I agree with others on this list that it is bad practice to release RGB files to a printer, and especially bad practice to mix RGB and CMYK files in the same release. The fact that you release some CMYK, some RGB and sometimes mixed means your company doesn't really have a fixed workflow. You say the people you work with "...are comfortable with an RGB workflow", but in fact it sounds like a "mixed" workflow.

-Todd Shirley
 
In addition, how would this RGB workflow handle print ready PDFs? If you were to create a print ready PDF with your RGB art embedded the printer could not open the art to convert it in Photoshop.


That's what the PDF/X3 standard is for and it works just fine in my experience.

Additionally, I've been releasing files in RGB (and occasionally mixed-mode for legacy stuff) for over a decade for both a major stock photography vendor and for a large consumer entertainment business and have great success. As an example, files released by my team today may go to any of six print vendors, some domestic, some in Asia. Each of those six vendors has their own profile, in some cases several profiles. Some of those vendors are likely to subcontract some of the work to vendors I don't know, much less have profiles for. In three months, we may do another run of a given product but at a different vendor.

Given the constantly changing swarm of destination profiles and the fact that the multinational approach to sourcing means that there isn't a single standard (SWOP, FOGRA , Gracol, etc.) that's appropriate, I'm gonna fire off X3s or live files with the output profile conspicuously left blank. And I swear it, it just works.
 
As long as you are talking coated paper the SWOP, FOGRA, Gracol is really not so big a deal, not compared with differences between coated and uncoated.

Pork why X3 and not X4, RGB and Transparency can be issue in X3, since you may get part of an image converted to CMYK and anoher part not (that was my experience, but it was a couple of years back when I decided to go X1a until X4 was ripe)?
 
Looks like third attempt = luck, I'm finally able to get all 3 test files to match in color info, but on my home system. Can't tell what I'm did wrong previously, same exact steps and but this time I used a digital photo instead of a scanned image.

Werby, thanks for correcting me on Image/Mode/CMYK = Convert to Profile. The conclusion came from wrong test result.
 
To which I suggested a color-corrected untagged CMYK is fine and can be assigned/tagged downstream. Specific vendor profiles ultimately doesn't vary a great deal from one to another. With expectation that color shift/lose is inevitable, out weighs unknown downstream conversion from RGB in the varies methods you mentioned.

Tech, please don't send untagged files. RGB, CMYK, Lab, whatever - I don't care - if you don't tag it then you hang the printer out to dry.

The advantage to having the printer do the final conversion is that they are the only folks who know the final output condition.
 
As a printer we recommend that images be sent RGB, and as Rich said, with the embedded profile. From a color correction standpoint I would much rather correct RGB images for many reasons. Equal numbers equal neutral in RGB, and you do not have to worry as much about the black build when making corrections. Plus, you always have the information that you started with.
Regards,
Todd
 
Tech, please don't send untagged files. RGB, CMYK, Lab, whatever - I don't care - if you don't tag it then you hang the printer out to dry.

The advantage to having the printer do the final conversion is that they are the only folks who know the final output condition.

Correct me if I'm wrong here, profiles for output only interprets image info, if so, vendor downstream should be able to assign their profile to an untagged CMYK without problem. How does this leave them hanging?

The advantage of having printers oversea converting our images not knowing what their profile is calibrated to really doesn't mean much. If I want to get to the bottom of it, I'll need to butt heads with production and their secretive CSRs—AKA too much headaches with minimal return. I would rather provide them with generic CMYKs that will output within an acceptable ballpark—color-wise. In the past, they have received RGB > CMYK converted proofs back and were not happy, their only option is to kick them back to printer and request more corrections. This is hardly an efficient process when you are dealing with oversea vendors.
 
Ok letts say I order a folder from you, a 21x21. (nw sending you an untagged number)
How would you interpret that 21 what? pixels? mm? inches? cm? could be anything.
Now if you did the research and found I was in europe you may assume mm or cm, then make a conclusion that I mean cm since mm is a rediculos size for a folder. But you would have to do research and guess work to get there. A computer or rip doesn't think like that.
If you you are aimng for a specific colour you will have to find out enough about your vendor to make an intelegent guess, sending information according to (what you guess) are his/her assumptions.
If you would at this point tag it to say, this is what I assume you meant. It would be like confirming the order by saying "oh you want a folder 21x21 CM?" That would then prompt a Yes or a No. All communication clear and in the open.
If at the point you say OK I will send you CMYK data and assume you are working to eg Euroscale Coated, I have clearly stated what I am expecting back. They can say we have no Idea what you are talking about… and that would be a heads up (or look they so cheap they using unquallified staff, I get what I get, but at least it's cheap) for the QC.

A printer can tagg untagged CMYK. You (not the printer) are thereby accepting all responsibility of the colour. You would be wisw to know what dotgain that press has on that stock is, or what is a standard they come close to.
A printer can untagg tagg files if necessary. If they untagg or ignore they have accepted your information. I would recommend using the output intent tag, because it is the easiest to override and it is a clear signal of your assumed intrnded output. It also has the advantage of triggering Acrobat to softproof correctly. And measuring in separations preview would give the CMYK values you expect. Also in separations you can toggle the black.

In this world I find there are designers frustrated at printers who have dugg their head in the sand and printers disgusted with the lack of knowledge of desigers who are flicking every swithch they find with a new feature. Get to know your printer and work as a team.

What really getts me is those aggitating colour consultants and printing pimps that are telling the printers to just go industrial. Demand print ready PDF's n mash them throuh production. The same people also try to sell in a proofing device to the designer saying that they no longer need to communicate with their printer but can surf the lowest price. All the knowledge that is wasted…sigh. You get wha you pay for.
 
A printer can tagg untagged CMYK. You (not the printer) are thereby accepting all responsibility of the colour. You would be wisw to know what dotgain that press has on that stock is, or what is a standard they come close to.
A printer can untagg tagg files if necessary. If they untagg or ignore they have accepted your information. I would recommend using the output intent tag, because it is the easiest to override and it is a clear signal of your assumed intrnded output. It also has the advantage of triggering Acrobat to softproof correctly. And measuring in separations preview would give the CMYK values you expect. Also in separations you can toggle the black.

In this world I find there are designers frustrated at printers who have dugg their head in the sand and printers disgusted with the lack of knowledge of desigers who are flicking every swithch they find with a new feature. Get to know your printer and work as a team.

What really getts me is those aggitating colour consultants and printing pimps that are telling the printers to just go industrial. Demand print ready PDF's n mash them throuh production. The same people also try to sell in a proofing device to the designer saying that they no longer need to communicate with their printer but can surf the lowest price. All the knowledge that is wasted…sigh. You get wha you pay for.


I agree, when a vendor "untagg or ignore they have accepted your information", this is great if I just want to cover my butt, in case something goes wrong. I don't work that way though. I have no problem accepting responsibility for submitting untagged but color preserved CMYK files when we have no printer profiles to work with. I still prefer this over sending RGB then pray conversion to CMYK and proofs would comeback acceptable by designers.

Let's not forget color is a relative thing and very subjective to different type of designers. Ideally, we would have direct communication with vendor's prepress/pressman, but we don't. So imagine when one designer is happy with RGB inkjet results and expect your oversea printer to match it. Ideally, he/she would sign off on first proof, great right? What if you have a designer whom is more picky? That translates to wasted time in proofing images on top of actual content of product. I have seen this happened a few times this past year, in most instances, the difference is negligible, but the time lost puts unnecessary stress on everyone that are trying to keep a schedule.

BTW, I appreciate all the responses. This helps me to sort things out.

Added: I'm not surprised person in charge of photo dept doesn't agree with working with smaller generic CMYK color space. The difference between getting colors that can be reproduce on paper vs what he/she thinks should be is the real issue.
 
Last edited:

PressWise

A 30-day Fix for Managed Chaos

As any print professional knows, printing can be managed chaos. Software that solves multiple problems and provides measurable and monetizable value has a direct impact on the bottom-line.

“We reduced order entry costs by about 40%.” Significant savings in a shop that turns about 500 jobs a month.


Learn how…….

   
Back
Top