Going from LAB to CMYK

imann08

In Pro PS 5, Dan mentions going straight from LAB to CMYK. In his LAB book, he talks about going through RGB to get to CMYK from LAB. Does anyone have an opinion about this?

What have you heard Rutt?

rutt

Please can you give a page reference for the PP5E statement. Dan has good reasons to go through RGB to CMYK. But If I just want the K channel as a mask or something, I don't care and will often go directly. If the move to CMYK is on the main branch to the final image, perhaps a shadow adjustment, then I am more careful.

imann08 wrote:

In Pro PS 5, Dan mentions going straight from LAB to CMYK. In his LAB book, he talks about going through RGB to get to CMYK from LAB. Does anyone have an opinion about this?

What have you heard Rutt?

edgework

imann08 wrote:

In Pro PS 5, Dan mentions going straight from LAB to CMYK. In his LAB book, he talks about going through RGB to get to CMYK from LAB. Does anyone have an opinion about this?

What have you heard Rutt?

I've seen shadows go bonkers in a straight LAB to CMYK conversion. The problem comes from LAB's ability to use colors that don't exist in any meaningful context. A lot of the color moves that produce such noticable improvement in the rest of the image yield shadows that look dark, but really have color values that RGB and CMYK would find nonsensical. But since the screen preview on which we base our opininons is already an RGB translation, those shadows don't suffer at all going to RGB from LAB. Something really gets lost in translation running straight to CMYK, probably because Photoshop is trying to put color where CMYK would have little or none, and, as a result, black has to come down.

imann08

Thanks for the reply guys. I will look for an actual page reference for you Rutt. I could have sworn I had seen it not just once but many times where he doesn't talk about going through RGB. That's why I brought it up. I'll get some references though cause now I'm interested if I misread something. Thanks again and I'll get back to you both.

imann08

OK, I have found a couple of examples for you to look at. I know there are more. Basically every time he has an image that needs to be output in CMYK and uses LAB. Anyways, here are the ones I found for now.

pg. 191
pg. 206 paragraph 2 and page 210 column 2 paragraph 2
pg. 392 last paragraph of column 1

I have read most of the book and have yet to see him mention going through RGB yet although he does do it the way he does it in the examples given many times. Look forward to hearing what you have to say. Thanks in advance.

rutt

In the classes of his I have attended he is always careful to add the step of going through RGB. I think it is an omission. Why not raise it with him yourself on his forum? http://tech.groups.yahoo.com/group/colortheory/ You'll have to register, but there is no risk at all in doing so. (I'm a moderator and will make sure your post is approved.)

imann08 wrote:

OK, I have found a couple of examples for you to look at. I know there are more. Basically every time he has an image that needs to be output in CMYK and uses LAB. Anyways, here are the ones I found for now.

pg. 191
pg. 206 paragraph 2 and page 210 column 2 paragraph 2
pg. 392 last paragraph of column 1

I have read most of the book and have yet to see him mention going through RGB yet although he does do it the way he does it in the examples given many times. Look forward to hearing what you have to say. Thanks in advance.

imann08

Does he read that site himself? I have yet to get on it as I just read your post but was interested in the chances of him seeing it.

Do you think it wasn't a simple omission and he has changed his opinion on the issue in some way? Doesn't sound like that's the case as he mentions in another place in the book something similar to what Edgework mentioned regarding shadows. He mentioned it a number of times in the LAB book so I find it hard to believe that I cannot find it anywhere in this book yet I see multiple examples of him ignoring that step.

It's good to get your feedback since you have read both books and have had experience in his classes. I appreciate your replies for this reason. BTW, my name is Isaac. I know if I ever forget yours I just need to look it up in the book

rutt

Dan not only read his newsgroup, but is very active in moderating it. Since it's set up to require each post to be approved, that means he reads them all.

Post this with a title like "LAB->CMYK conversions, omission in PP5E?" and he'll be all over it (though it sometimes takes him a week or so to get to things.)

imann08

rutt wrote:

Dan not only read his newsgroup, but is very active in moderating it. Since it's set up to require each post to be approved, that means he reads them all.

Post this with a title like "LAB->CMYK conversions, omission in PP5E?" and he'll be all over it (though it sometimes takes him a week or so to get to things.)

I just posted my comment. I hadn't read your post yet but I wish I had as I did not title it like that. I've got the same name on that board as I do here. I can't wait to see a response.

BTW, I found two more examples. Page 224, last paragraph and the first paragraph of page 246. These may actually be better ones than I had sent before. Thanks again for your replies. I value your opinion.

imann08

Just FYI, my question has been answered by Dan in his color theory group. I know this sounds corny but it was actually kind of thrilling to get my question answered directly by him. Some people want to meet a movie star, I want to meet him. I have some serious problems. lol

jfriend

What Dan said

imann08 wrote:

Just FYI, my question has been answered by Dan in his color theory group. I know this sounds corny but it was actually kind of thrilling to get my question answered directly by him. Some people want to meet a movie star, I want to meet him. I have some serious problems. lol

Well, since I imagine all of you want to know the answer as much as I did, here's what Dan said in the colortheory group:

>>The question/issue I have is that in Dan's LAB book, he specifically said that we should move from LAB to RGB and then to CMYK.>>

No, I did not say this as a general rule, although there are rare occasions when it is necessary.

>>In ProPS5E, I don't notice it mentioned once and see him making the move from LAB directly to CMYK a number of times.>>

In all but the rarest cases, going to RGB first doesn't hurt but it's a waste of time--the results are close to identical. The only exception is where the LAB file contains imaginary colors, or colors that are far out of the RGB, let alone CMYK, gamut. In that case, converting LAB to RGB to CMYK gives a different result from LAB to CMYK, and you are likely to prefer the first because it will be more similar to how the LAB file is displayed on the monitor.

I wrote (p. 179 of Canyon Conundrum) "Under ordinary circumstances, there's little difference beetween a file that goes from LAB to CMYK and the same file converted from LAB to RGB first and *then* to CMYK. If the LAB colors are even close to bing within the RGB gamut, the original LAB and the LAB-to-RGB file are indistinguishable for purposes of a subsequent conversion to CMYK. If imaginary colors are lurking about, however, all bets are off. An imaginary color in LAB seems to the conversion algorithm to be a very different animal from the same color that has been arbitrarily crammed into the RGB gamut."

Deliberately introducing imaginary colors before converting to CMYK is a highly unusual procedure. I include the warning in Canyon Conundrum as a matter of completeness in a book about LAB, but consider it too esoteric to include in a general color correction text.

Dan Margulis

imann08

I probably should have done that myself. Thanks John.