Assign Profile vs Converting to Profile

[Printer_Rick]

This relates to a very lengthy thread in the InDesign forum, "RGB vs CMYK images and resolution"

I have a lot of questions (perhaps confusing) relating to RGB color gamuts. To simplify let's start with 2 gamuts, ProPhoto and Adobe RGB

I have a profile editor that can view both of these within potato-shaped Lab gamut. They are of course both triangles, I believe all RGB gamuts are. I can see ProPhoto is considerably larger than Adobe RGB, containing more fringe colors

I also see that the gamma of Adobe RGB is 2.2. The white point is 6500K

The gamma of ProPhoto is 1.8. The white point is 5000K

I understand gamma to be "black point". Or better yet "black density". On a press sheet, ink density can be measured with a densitometer. In my experience a density reading of 2.2 on a press sheet would be very dark. Is my understanding correct - that gamma (RGB) is comparable to ink density (CMYK)? Perhaps better to state as an analogy: Gamma: RGB as Density: CMYK

My monitor RGB profile has a gamma of 1.8 (mac standard). This tells me that the Adobe RGB gamma of 2.2 has to be re-interpreted on my display. Is that correct?

As for white point, that would be the RGB equivalent of CMYK paper white.

The InDesign forum has a lot of discussion about assigning profiles, vs converting to profiles. My understanding is that assigning a different RGB is actually a "pure" conversion. The pixels are left completely intact. There is no move to Lab, and back to RGB. It's taking the image and effectively dropping it into a brand new gamut, The price for this, of course, is that the appearances of the colors are completely redefined, and this appearance shift can at times be radical.

For example, if I have an ProPhoto image open, then assign Adobe RGB, I can see very clearly that the image becomes darker on-screen, and the color "shrinks"

As a prepress person, I have often used re-assigning in RGB mode as a very effective color correction tool. Usually it's turd polishing, to be quite honest, when critical color match is not an issue. The scenario is usually a crappy sRGB image. I assign Adobe RGB, which as the Adobe description states is ideal for conversion to CMYK. I must add that I always use proof preview, I am well aware that Adobe RGB has colors far beyond a standard CMYK gamut. But when I convert to CMYK, using Adobe RGB as the source, the image color is expanded, and the result on press is often vastly improved.

I will also add that as a prepress person, I don't go re-assigning in this fashion without the customer's consent.

In the InDesign forum, this "re-assigning" has been referred to as "random color". There is a lot of emphasis on color appearance, and maintaining color appearance. The consensus therefore is that if you had an sRGB image, you should convert to Adobe RGB. But then it is my understanding that you miss out on the often huge benefit of gamut expansion. If you wanted to expand color after converting, you have to do color corrections, which alters the pixel data and in the strictest sense is destructive (unless you use adjustment layers).

All this leaves me wondering - if assigning is such a no-no, why is it available? Probably the main reason for the assign capability is to assign profiles to images that don't have an embedded profile. Sometimes users unknowingly discard profiles, if the color settings policy is set to off. When another user open the image, he quickly sees the image does not have a profile.

Normally he would assign his working space, since that is affecting his visual on-screen appearance. But he can't know for sure if that's true to the original capture.

Which brings up another point. Any device doing the capture (camera or scanner) has a gamut. This gamut is an input profile.  When the image is translated from device capture into digital file, should this input profile be embedded in the image?

At this point I'm not sure about this. I have a 7.1 MP camera, and the downloads always have sRGB embedded. Not a profile specific to the Kodak model. My guess is that sRGB is a universal standard, representing the gamuts of monitors and desktop scanners. It is the working space of the world wide web. So it's more or less the default RGB, and is also the default working space in all Adobe applications (North America general purpose).

But the description of sRGB is very clear. It is not ideal for prepress, this is stated in Adobe's description. It is small. This may make it comparable to CMYK, but it is still not ideal for conversion to CMYK. And in fact there are CMYK colors that fall outside of sRGB. Especially if you are dealing with the larger CMYK gamuts corresponding to new offset screening technologies (FM screening and concentric screening)

So why in the world would someone convert from sRGB, to Adobe RGB? There's no benefit at all. May as well leave it sRGB, instead of converting. And the even bigger question - how do you know that sRGB is "true" color? To me, the true color is the original subject. In the case of a photo, that might be just a memory. In the case of a scan. it's the original, but the user might not even have that, if someone else did the scan and all he has is the digital file. So who's to say that the embedded profile - sRGB - is a fair representation of the original?

Re-assigning RGB profiles may be an odd way of adjusting color. But it can be effective. Why would the assign option be readily available, if not to translate colors to a different gamut, without altering pixel data? Seems to me it is the primary reason Adobe developed the assign option in the first place.

I know this is a lot of questions. Any input on any of these matters would be greatly appreciated.

[Printer_Rick]

Thanks for that. Maybe you can help with this question, I posted earlier in color management (at Ramón's suggestion):

"Should input RGB device profiles (camera, scanner) be embedded and used as a source space, for conversion to CMYK? Or converted to another RGB space (such as Adobe RGB), then converted to CMYK? I don't see where 2 conversions are beneficial, but I could be wrong..."

The main reason I ask this is, I very rarely see RGB images with scanner and camera profiles embedded. To be honest most I receive are sRGB, or don't have a profile at all.

Not being a photographer, I don't download camera images for correction in Photoshop. My scanner produces a LAB image (I work in Linocolor when scanning), so no RGB profile there either.

[Was DYP]

"Should input RGB device profiles (camera, scanner) be embedded and used as a source space, for conversion to CMYK? Or converted to another RGB space (such as Adobe RGB), then converted to CMYK? I don't see where 2 conversions are beneficial, but I could be wrong..."

If you are going to do anything (adjustments) to a photo you will want to convert to a working profile space. If no adjustments are needed you could certainly convert to CMYK from the device profile. But again like re-assigning, what ever works best for your desired finish result. But certainly do not save an image with any chance to passed along with device profiles embedded.

[Printer_Rick]

Was DYP wrote:

But certainly do not save an image with any chance to passed along with device profiles embedded.

Why is that a bad idea? I think I might know but I'd rather get your answer first..

[Was DYP]

It takes some intervention and if someone really does not know what they are doing the output could be pretty screwed up. Although now days PS is pretty smart about this situation as long as you have the right color management policies selected. But if the images is used in something else, who knows.

[Printer_Rick]

Was DYP wrote:

It takes some intervention and if someone really does not know what they are doing the output could be pretty screwed up. Although now days PS is pretty smart about this situation as long as you have the right color management policies selected. But if the images is used in something else, who knows.

If I understand correctly, by pretty screwed up you mean the device profile is discarded, and is therefore no longer the source space.

Also let's say the image is placed in another Adobe app (such as ID), and the source (device) profile is honored. No problems then, right?

If that is the case, then why are so many design images sRGB? Are designers purposefully converting to sRGB? The only use I see for sRGB is web output (my personal opinion)

Again I need to state, in case someone else reads: I am not suggesting that a prepress workflow should forcefully remove sRGG profiles from images, and re-assign. But if you're a designer, my feeling is you are free to do that if you want to.

[Miguel_Curto]

Printer_Rick wrote:

Thanks for that Miguel, I appreciate it. I did not post my original questions to anger anyone

I should clarify. I receive an sRGB image. I assign Adobe RGB. On-screen the color is more saturated. This saturation carries over into the subsequent CMYK conversion.

Try it with any sRGB image. Convert to CMYK, save a copy. Then assign Adobe RGB, convert to CMYK. Compare the two CMYK images, I think you'll see what I mean.

It is not a specific color correction by any stretch, but it's not random rearranging of colors either.

Youre partially right,I does look more satured,I made a mistake assuming it would be the other way around but I can see whats happening.I know why but it would be really hard for me to explain...its the values in between that disapear (or not represented by Photoshop) and not the extremes...not more color but more gradient if you get my meaning)

on the other hand can I suggest you something?:

Save those both images (original sRGB and the other you assign to AdobeRGB) and save them both (as JPGs for instance).

Now open both in a non ICC aware aplication (Internet Explorer will do just fine).Can you spot the diference?

[Printer_Rick]

Miguel Curto wrote:

Youre partially right,I does look more satured,I made a mistake assuming it would be the other way around but I can see whats happening.I know why but it would be really hard for me to explain...on the other hand can I suggest you something?:

Save those both images (original sRGB and the other you assign to AdobeRGB) and save them both (as JPGs for instance).

Now open both in a non ICC aware aplication (Internet Explorer will do just fine).Can you spot the diference?

I don't have Explorer on my mac. Safari showed a difference. Preview showed a difference. Quick Time did not.

Apparently Quick Time just defaults to monitor as source. So it must be what you refer to as non ICC aware app.

I think the reason (correct me if I'm wrong) that assigning Adobe RGB expands the color is because the number values take on a new definition in the larger gamut.

By the way, how can 1.8 gamma be considered stone age? Would 2.2 give a better CMYK preview?

[Miguel_Curto]

The diference between those two images is only the tag or the labbel so to speak.ICC aware apps as Photoshop,Mac Preview or Safari (this is new to me) are "smart" enough to try to translate those RGB values to your monitor values.The values between those images are exactly the same...youre only changing its name to trick those apps to render the values in a diferent manner.

It may work,and toure right..its not random but it would be considered as "exploit" and in my view two wrongs dont make one right.

Having said that,theres a lot of things that make no sense that in the end might suit some needs.

Ramon might slam me for this but Ill give you an example:

Theoretically if you start with a 8bit/channel image theres no point in converting to 16bits for editing.But it so happens that Photoshop "mechanics" work better in 16bits so in the end (for extreme corrections) there is a real and mesurable gain (Wich I can post here to proove)

In that process youre throwing away some "finess" in color witch should be adress as editing image but nothing is writen in stone and as long you REALLY understand the implications of what youre doing theres no WRIGHT or WRONG.

[Printer_Rick]

Theoretically if you start with a 8bit/channel image theres no point in converting to 16bits for editing.But it so happens that Photoshop "mechanics" work better in 16bits so in the end (for extreme corrections) there is a real and mesurable gain (Wich I can post here to proove)

In that process youre throwing away some "finess" in color witch should be adress as editing image but nothing is writen in stone and as long you REALLY understand the implications of what youre doing theres no WRIGHT or WRONG.

I've heard some discussion about 16 bit vs 8 bit before.  Let me ask this. Assume you have a raw RGB capture with device (camera) profile, and it is 8 bit, would you recommend changing to 16 bit before converting to RGB working space? To smooth loss of gray levels?

[Miguel_Curto]

Only pays off on (very) extreme corrections..otherwise might as well leave it at 8bits because the results will look the same and you save time.

If you start (capture) at 16bits (most scanners and cameras do 12bits but Photoshop assumes as 16 anything more than 8bit/channel) its a diferent story...

[Printer_Rick]

Only pays off on (very) extreme corrections..otherwise might as well leave it at 8bits because the results will look the same and you save time.

If you start (capture) at 16bits (most scanners and cameras do 12bits but Photoshop assumes as 16 anything more than 8bit/channel) its a diferent story...

Sounds reasonable. Now let me ask this. You are provided a really bad sRGB image. It's my understanding you would:

- Convert to your working RGB space, and do massive color correction to correct appearance, instead of

- Assigning your working space, avoiding massive color correction (still perhaps doing considerable adjustments)

Either way, you are greatly changing the original color (which you know is terrible). But with assigning, there is no trip to Lab and back, which is (technically speaking) destructive. (Adjustments don't have to be destructive when using adjustment layers)

[Miguel_Curto]

Printer_Rick wrote:

Sounds reasonable. Now let me ask this. You are provided a really bad sRGB image. It's my understanding you would:

- Convert to your working RGB space, and do massive color correction to correct appearance, instead of

- Assigning your working space, avoiding massive color correction (still perhaps doing considerable adjustments)

Either way, you are greatly changing the original color (which you know is terrible). But with assigning, there is no trip to Lab and back, which is (technically speaking) destructive. (Adjustments don't have to be destructive when using adjustment layers)

I get your point but I would go for convertion mainly because of Rendering Intents and to preserve color relationships.I would have to go and see a 3D model of sRGB and AdobeRGB to give you a more detailed answer (im off to bed now,really late here) but you can see it here among other places:

http://www.drycreekphoto.com/tools/printer_gamuts/

Without knowing exactly what color (and amount) are being cliped cant give you a straight answer.

Since both profiles are grey balanced (add equal amounts of color you get a neutral grey) the clipping might not be that obvious to unbalance tonal relationships and im not that "geek" to know it by heart.

[Was DYP]

Save those both images (original sRGB and the other you assign to AdobeRGB) and save them both (as JPGs for instance).

Now open both in a non ICC aware aplication (Internet Explorer will do just fine).Can you spot the diference?

I do not know why you would want to save a re-assigned image or view it some where else, it is only a means to an end, CMYK.

[PeterK__]

Printer_Rick wrote:

No, the opposite. I receive the sRGB. Assign Adobe RGB. Color is expanded.

You're screwing up your client's intended colour. They viewed it as sRGB, saved it with sRGB because that's the final image they settled on, and then you just arbitrarily decide to assign AdobeRGB and convert to CMYK? Let me guess... then the client complains that the colour is off and you charge them for "edits."

It's just a bad workflow. If the client's file is not coming out on press as it looks on screen, you need to profile your press better so that you can convert the client's RGB to your CMYK and have it match. Assigning a different profile because it suits your subjective taste is ridiculous.

(besides, skin tones often take on an ugly reddish tone by incorrectly assigning AdobeRGB to a sRGB image.)

If I send you a sRGB file, that is how I viewed it in Photoshop, how I saved it, and how I expect it to look. Even for a client that request "pleasing" colour, if they're sending you a sRGB the assumption is the same. They approved that image and sent it out the door like that, so for what reason are you screwing with their intended colour from the outset?

[Printer_Rick]

PeterK.. wrote:

You're screwing up your client's intended colour. They viewed it as sRGB, saved it with sRGB because that's the final image they settled on, and then you just arbitrarily decide to assign AdobeRGB and convert to CMYK? Let me guess... then the client complains that the colour is off and you charge them for "edits."

It's just a bad workflow. If the client's file is not coming out on press as it looks on screen, you need to profile your press better so that you can convert the client's RGB to your CMYK and have it match. Assigning a different profile because it suits your subjective taste is ridiculous.

(besides, skin tones often take on an ugly reddish tone by incorrectly assigning AdobeRGB to a sRGB image.)

If I send you a sRGB file, that is how I viewed it in Photoshop, how I saved it, and how I expect it to look. Even for a client that request "pleasing" colour, if they're sending you a sRGB the assumption is the same. They approved that image and sent it out the door like that, so for what reason are you screwing with their intended colour from the outset?

Peter I think you've misunderstood me. It's not my workflow, not by any stretch. I only do it when a client requests color correction. Otherwise I always honor the sRGB as source.

I was just suggesting this for a really bad colorless sRGB image, as a quick fix, that's all.

[Was DYP]

Usually when I open these sRGB images, reassign Adobe RGB, then convert to CMYK, the result is much better than sRGB - CMYK. If my RGB policy was set to Off, I would get the same effect as re-assigning - my working space is Adobe RGB.

I'll add I only throw out the embedded sRGB if the the customer requests pleasing color, and the proof preview is crud. Otherwise I have to honor the sRGB.

I know exactly what you are talking about, and there is no reason not to use this approach if you achieve the desired result by doing so. I have many timed re-assigned (I like this name much better than false profile) an image in Adobe RGB to some other RGB space like Ekta Space or others to achieve the desired look when converting to CMYK. It is no more of a trick than to make image adjustments for the same reasons.

[Printer_Rick]

Was DYP wrote:

I know exactly what you are talking about, and there is no reason not to use this approach if you achieve the desired result by doing so. I have many timed re-assigned (I like this name much better than false profile) an image in Adobe RGB to some other RGB space like Ekta Space or others to achieve the desired look when converting to CMYK. It is no more of a trick than to make image adjustments for the same reasons.

Finally! Someone knows where I'm coming from. It's a simple concept. Why is this process hated so much? As long as you proof and like the results, re-assigning is effective.

[Printer_Rick]

My original post had valid questions. I was searching for valid answers. I was not trying to suggest how someone should set up a color management workflow.

Why have I angered you so much? That was not my intent. If you have insight, you can share it without including insults or personal attacks.

[Printer_Rick]

WRONG.  Profiles are complex three-dimensional bodies in space.  You're looking at a simplified 2-dimensional projection on a single plane that tells you very little.

I am aware color is 3 dimensional, wasn't trying to imply it wasn't. RGB is displayed as a triangle in simple models, that's what I was getting at

PS— sRGB is the lowest common denominator, where the "s" stands for sh¡t.

sRGB is not my working RGB space

Finally, you should know that there is a dedicated Color Management forum in these Adobe forums, and that's where this type of discussions belong—not here, really.

Then I will post my questions there. Again my apologies.

[Miguel_Curto]

Adobe RGB  and Kodak Pro Photo RGB are large gamut (PRoPhoto being one of the largest) device Indepedent profiles (sRGB is too).

sRGB is a narrow gamut profile intended to try to reproduce an uncalibrated monitor and atenuate diferences from the same image shown in diferent monitors.

All CMYK are a device dependent profiles that determine a print intention.As a usefull exercice go to Edit - > Color Settings -> CMYK and choose "Custom CMYK" so you can see what in fact a CMYK profile is.

Being new to this forums dont know if I can do this,but I would recomend you read late Bruce Fraser book "Real World Photoshop" focused on Color Management or at least read some of his articles at Creativepro:

http://www.creativepro.com/articles/author/127446

[Ramón_G_Castañeda]

Miguel Curto wrote:

Being new to this forums dont know if I can do this,

Absolutely, Miguel.  We link to books and web sites all the time.

Welcome to the forums!  

[Printer_Rick]

Miguel,

Thanks for the response I will look into it.

[Ramón_G_Castañeda]

Printer_Rick wrote:

Why would the assign option be readily available, if not to translate colors to a different gamut, without altering pixel data? Seems to me it is the primary reason Adobe developed the assign option in the first place.

Gawd, that's so wrong that I hardly know where to begin.

Let's see…

ASSIGN is there so you can make educated guesses as to what color space an untagged file was created handed to you by a moron.  When you run into such a moron who does not embed the profile in his images, first cycle through possible color profiles in the ASSIGN dialog box until you make and educated guess by finding the one profile that makes the image look most plausible, then go beat up the moron who handed you that file with a baseball bat.

No, Adobe, didn't develop ASSIGN so you could wreck images willy-nilly.

Now, I'll stop.  My blood pressure is ggoing through the roof already.

[Printer_Rick]

No, Adobe, didn't develop ASSIGN so you could wreck images willy-nilly.

That's fine if you see assigning as wrecking images. It works for me, I've done it with good results.

[Ramón_G_Castañeda]

Plonk!