Why Not Edit in Output Device Mode?

Z
Posted By
zloi
Jun 8, 2004
Views
1761
Replies
42
Status
Closed
I’ve been having a hard time grasping this issue. Working from a scanned image, which initially has my scanner’s profile, I move to Adobe RGB to perform all my editing. When I’m done and ready to print, I can check how the image will look in my selected profile (in my case, I output everything to an Epson 2200) by going to View/Proof colors. The final step before printing is to go to Print Preview and switch from Adobe RGB to whatever profile I choose (generally for the type paper I am using) and selecting "No Color Adjustment" in the final print screen. I have worked out this technique based on what I have heard and read from "the experts," ie, anyone more knowledgeable than I.

Question: why am I bothering to work in Adobe RGB at all? For a time, the first thing I did with my scanned file was go to Image/Mode and Convert it to whatever profile I anticipated printing from. But since the "Convert to profile" command, as I understand it, changes the actual color values in the file while keeping their on-screen appearance about the same, it is not intended to give an accurate idea of how the printed file will look. The "Assign profile" command actually changes the colors on screen (without doing so in the actual file) so you can see the difference in using various different profiles. But just as a painter can only paint from whatever colors he can create on his palette, why would I not want to work with just those colors that my printer (in this case, the 2200) will actually print for the given profile? What good does it do me to edit in a mode which may give me a lot of out-of-gamut colors? I mean, so long as I can access all the filters, commands, and controls available in RGB mode, if I know my output profile in advance, I don’t see why I should not edit in that mode from the get-go. Anyone have any comments or suggestions in this regard?

How to Improve Photoshop Performance

Learn how to optimize Photoshop for maximum speed, troubleshoot common issues, and keep your projects organized so that you can work faster than ever before!

LH
Lawrence_Hudetz
Jun 8, 2004
I have had some of the same questions about these matters. The answer that works for me is that I am not simply editing for one device, paper and ink. I am editing such that I can send the file to a commercial printer, or a web designer, or my Canon printer, or my Epson printer etc, and know that, in the end, I will have a file that will output as I intended, with smaller deviations to be accounted for by the end use.

BTW, I also scratch my head over "Assign" and "Convert". The only practical use I have seen for it is if I am going to the Web using "Save for Web" I convert to sRGB first, because if I don’t, apparently the process will take my Adobe RGB and Assign the sRGB. Then, watch out for color shifts!

Finally, you have access to "Soft Proofing" in PS. That will get you into the color space you will output, and you can do your editing there.
J
Jim
Jun 8, 2004
I suppose it would be OK to edit in the output color space if you are certain that this space will never change.

Alas, I don’t have that degree of certainty.
wrote in message
Question: why am I bothering to work in Adobe RGB at all? For a time, the
first thing I did with my scanned file was go to Image/Mode and Convert it to whatever profile I anticipated printing from. But since the "Convert to profile" command, as I understand it, changes the actual color values in the file while keeping their on-screen appearance about the same, it is not intended to give an accurate idea of how the printed file will look. No, you are wrong. The convert to profile changes the in memory data to whatever profile you select. It may or may not change the appearance of the image on the monitor. Most likely some changes occur. The only time that the file on disk is changed is when you direct such change.
The "Assign profile" command actually changes the colors on screen
(without doing so in the actual file) so you can see the difference in using various different profiles
No, this command merely tells PS what profile it should use to interpret the in memory data. If you load an image which does not have a profile, the PS is left with no information about what the numbers mean. It must therefore make an assumption which is almost certain to be incorrect. Jim
L
LenHewitt
Jun 8, 2004
Zloi,

An editing space really needs to be as linear as possible. Output device profiles can be anything BUT linear, and so are certainly not ideal as editing spaces.

Additionally, an editing space should encompass all the possible output device colour spaces that you are likely to come across. Edit in the 2200 colour space (for which paper surface?) and then you find you want to print on a different paper, or your 2200 goes belly-up and you replace it with something different, or something much better comes on the market and you upgrade.

Now your saved images may well be in a smaller colour space than your new equipment…
BG
barry_gray
Jun 8, 2004
I might add that usually all you need to do edit wise is a hue/sat and maybe a curve adjustment. Keep your Adobe1998 space, add one or to adjustment layers in soft proof mode, lable those layers as that profile.
You’ll always have the untouched original for other OP’s.
MO
Mike_Ornellas
Jun 8, 2004
This is all BS and can be addressed with a prebound workfow.
PF
Peter_Figen
Jun 8, 2004
Refining Len’s comments regarding output spaces. Epson RGB spaces are perhaps more non gray balanced than non linear, which is a more difficult situation to deal with. The big attraction of all of the intermediate working spaces in PS is that in each one of them a neutral gray is achieved when red, blue and green are equal. That almost never happens in output spaces, at least not across the entire tonal range. That makes it extremely difficult to find and set a neutral when you need to and impossible to desaturate. Try a complete desaturation while your in your Epson profile and you’ll see. In addition, most good Epson profile also compensate for the natural state of plugging of the shadows below somewhere around 20 or so on the RGB scale, and map a 0,0,0 RGB up to the point of separation of detail. It would be all too tempting to push the black point down much closer to 0,0,0 and end up with lost detail. Much better to set the endpoints in a working space and let the profile map them to where they need to be. Now add to the fact that the 2200 Epsons seem to need reprofiling every six or eight months due to machine drift, which is remarkably infrequent considering the level of user control available. The previous comments about repurposing all apply.
Z
zloi
Jun 9, 2004
With all due respect, I was not asking about "repurposing"–obviously, if I think I will have different output devices, then I need a "generic" color mode. But I am thinking specifically of just ONE–the 2200, in this case. I will do some testing and return to this topic, but as of right now, I don’t really follow some of the posts. You’ll have to specifically define "linearity" and "non-gray-balanced" and indicate why they adversely affect working in the output mode rather than in Adobe RGB & then switching over. Use a concrete example, if possible. As for "prebound workflow," I imagine that could be the answer–if I only knew what it was.

Maybe someone else has a clearer explanation. I am (obviously, I guess) not a print professional. Try to keep jargon out of your response, please!
MO
Mike_Ornellas
Jun 9, 2004
drift wood…
DM
dave_milbut
Jun 9, 2004
mike, your smarts are showing again. better tuck em in and zip up!
LH
Lawrence_Hudetz
Jun 9, 2004
A scientist goes out one morning with a ruler, measures the height above ground to the bottom of a camels belly, and finds out it is 5′ off the ground. Next, the camel driver gets up on the camel, and the scientist re measures the height. It is now 4′ off the ground. Now, a passenger gets up on the camel and again, the scientist measures the distance and it is now 3′ off the ground.

That’s linearity.

Finally, the camel drivers wife walks up to the camel and tosses a canteen full of water to the driver and the camel crashes to the ground.

That’s non-linearity.
L
LenHewitt
Jun 9, 2004
Zloi,

I’ll try and keep it simple, although you have opened an extremely complex topic with your question. In keeping it simple there will be vast generaliszations and some things which are not strictly accurate, but will demonstrate the theory.

In a linear environment, a change in say contrast, will affect all colour channels equally and across the full range of tones equally. In a non-linear space the same change may affect each channel differently and have different effects at different levels within each channel.

So, if you started with a neutral grey gradient running from black to white and in RGB mode, in a linear space any linear alterations would still leave you with a neutral grey across the whole range. In a non-linear space you could end up with say a blue cast in the lighter areas, a green cast in the mid-tones and a red cast in the shadows.
AP
Andrew_Pietrzyk
Jun 9, 2004
All theories aside,

Do major edits in your (linear perceptually uniform working space), convert to printer profile and fine tune your adjustments in printer color space. See how you like it. Experimenting will answer your question better than any theoretical discussions. Think about the forest while you’re wasting all that paper.

drift wood…
MO
Mike_Ornellas
Jun 9, 2004
Curves, Levels, Selective color, are not affected by the color space.

Masks are.

The working space is only really valid if you convert to it from a good known capture space.

Assigning a working space to a file that hasn’t been converted is the 2nd choice. Not a better one, but if you don’t know what created the image, it’s basically guess work, OR assign a color space that you like perceptually and then convert to the working space.

The whole point of the current color mgmt. workflow is to preserve the images (look) as best you can after a conversion is done.

Working in an output space may or may not give you the desired end results.

If you are only going to be using only one output device and that’s it, than working in the output space is just fine. Especially if it’s a CMYK device. (long story)

If you need to repurpose the file for multiple devices, you should work in a working space, but have to realize the limitations of such workflow and usually these issues won’t arise until your face is pushed up against the glass.
BL
Bill_Lamp
Jun 9, 2004
Let me add to Mike’s reply.

"only one output device"

It is FAR easier to match output from one printer to another if you are working from a device independent color space. I ran into this when I had to match Epson 1270 prints with an Epson 2200. I’m not saying I had an easy time of it because it wasn’t easy for me. But it was a lot easier with the files with a device independent space.

Bill
L
LenHewitt
Jun 9, 2004

M.O,

Did you deliberately ignore the:

"and some things which are not strictly accurate, but will demonstrate the theory."

or was it an oversight on your part? <g>
MO
Mike_Ornellas
Jun 9, 2004
Color is subjective.

;o)

We can do blanket statements when necessary.
LH
Lawrence_Hudetz
Jun 9, 2004
Pendelton blankets are nice! 😉
Z
zloi
Jun 9, 2004
Well, now I know how linearity relates to camels. Too bad I wasn’t asking about camels. Okay, still no one has explained why, if I am using a 2200 profile for the 2200, that profile is not a perfectly valid one for editing in. Sure, it’s nice that your "linear" profile keeps all the colors, well, "linear", but what good is that if the printer can’t reproduce all those nice "linear" colors, but only the ones which, supposedly, the profile was designed to encompass???
MO
Mike_Ornellas
Jun 9, 2004
we have answered your question.
T
textmonkey
Jun 9, 2004
Yeah but the message isn’t getting across.

You edit device specific, your color corrections will only be valid for that device and may not be valid 6 months from now when toner, paper, etc has been changed a few times. If you decide to print that device specific image to another process, it will may look drastically different in color and you may not be able to repair the "damage" done by working in a device specific space.

If you’re color correcting in CMYK this is even more true, but color correcting in CMYK gives you more control over color correction in a device specific work flow.

Yes, the working space isn’t a perfect color correction option, but for those of us that understand the pitfalls of reprinting a legacy file to a new process or the same process months (in some cases even weeks later), it’s the best compromise.

As has been said and implied, it’s what works best FOR YOU. Don’t worry about what everyone else does is that doesn’t fit how you prefer to work.
RW
Rene_Walling
Jun 9, 2004
zloi,

You can always edit in your printer space.

But you’ll be kicking yourself the day you need to repurpose the image.
Z
zloi
Jun 10, 2004
The only reason it seems people can come up with has nothing to do with "linearity" or other technical issues. It has to do with outputting the file to a different device. I tried to clarify that that was NOT what I was asking. But people keep coming back with that answer.

One guy evaded the question by saying "it’s a big complicated issue..blah blah blah" and talked a lot of condescending blarney rather than address the question. Another guy evidently considers himself a Sage above the fray and replies with cryptic, inscrutable phrases (even claiming he helped answer the question!). Zowie, this is quite a geeky group!

Speaking of geeky, I guess if you want an answer, sometimes you just have to figure it out for yourself. I’ll go edit a number of files now in several Epson profiles, edit the same files in a couple non-Epson (linear) profiles, do a lot of careful printing and record-keeping, and see what happens. Heck, I should’ve done that to begin with, altho it’s something of a bore. I just thought maybe somebody actually knew already what would happen and why.
SS
Susan_S.
Jun 10, 2004
Well actually this was something that was puzzling me, and I think I learnt somethig form this thread..so to everyone that bothered to answer thank you.

My summary of what I learnt (just in case it helps) –
(a) it’s harder to repurpose (yeah i know it’s not what zloi wants to hear but it’s relevant to most users)
(b) if your output space is non-linear and non-neutral it can make the edits harder to make – extrapolating from the camel example – small discrete changes (of pick variable of choice) in a non-linear editing space can lead to unpredictable results – Len gave some really clear examples in post #10. And the complexities of trying to colour correct in RGB where equal amounts of R,g,b, don’t give a neutral grey make my head spin. Of course if you get to know your colour space’s foibles then you can work around this I imagine and get used to what you need to do – as long as you don’t need to reprofile the output device or change it as then you wil not only need to re-edit existing images, but also get used to a different set of (unpredictable) behaviours.
Susan S
GA
Gordon_Anderson
Jun 10, 2004
zloi

I think most people think they’re helping even if they’re answers break away a little and go to a more generalized (but still problematic) consideration. It’s the same if a teacher was asked a specific question and chose to answer the question to the class rather than just the student. With these program issues there are some Q & A’s that are very specific and can be answered directly. But there are many questions that dig into the complexities of the programs and thus illicit a range of comments. Try not to feel slighted at this departure. Overall (I think) it helps the most people, not to mention it’s unavoidable.

Gordon
MO
Mike_Ornellas
Jun 10, 2004
How about this.

Edit in output space and when you want to print the file to another device, you’ll find the two outputs don’t match.

When you come to that conclusion, you’ll understand why we edit in a working space.

it’s pretty simple math.
LH
Lawrence_Hudetz
Jun 10, 2004
No, you learned nothing specific to camels. You asked about linearity. The example I gave was a classic, if somewhat humorous illustration of linearity from Engineering 101. If you concluded camels instead of linearity, I can see why you are having problems. 😉
MV
Mathias_Vejerslev
Jun 10, 2004
The camel hasnt dropped yet.
LH
Lawrence_Hudetz
Jun 10, 2004
LOL!
RW
Rene_Walling
Jun 10, 2004
The camel hasnt dropped yet.

We’re missing the straw.
CC
Chris_Cox
Jun 10, 2004
Many of the filters and adjustments in Photoshop make assumptions about the colorspace – that it is perceptually uniform, equal color values make gray, etc.

Most device color spaces don’t have those properties, and can lead to some serious surprises when trying to use adjustments, gradients, blending, etc.
LH
Lawrence_Hudetz
Jun 11, 2004
We should collect all this and put it in a folder for reference somewhere. Then we can add to it as time goes on.

You can keep out the part about measuring camels, if you like. I won’t mind!!;-)
RW
Rene_Walling
Jun 11, 2004
But I like the part about the camels
Z
zloi
Jun 11, 2004
My apologies to Mike. Somehow I missed your message #12 when it first posted and only just now, reviewing this thread, came across it. Thanks to all who shared here (camel included–someone could at least unweight him while we’re figuring this out). Since the 2200 is the only device I have output photos to in the last 2 years that I’ve owned it (or whenever it came out), that was why I really was looking for a specific answer. People have made a lot of assertions here. I don’t know which and how many of them are based on solid fact, since I haven’t seen this question discussed by any "authoritative" authors. I keep all my files in Adobe RGB (my working space), but to reiterate yet again, if I know I’m going to be printing to the 2200, why not continue with my manipulations–contrast & color changes, montages, filters, etc etc–in the mode of that device? However, Chris’s comments, if valid, are exactly the specific "contras" I was looking for. Some filters specifically alter color and blending, so it’s hard to generalize about which color space to work in with them. The results may simply be different from one to another, but not necessarily "better" or "worse." Other filters I can’t see it would make a difference. Unsharp mask–don’t know. Anyways, the only real way to test these multiple theories, onerous as it is, is to cut down some trees, stock up on ink (I still don’t print enough to justify using anything but the cartridges) and get some monitor burn on my eyeballs. I agree with Lawrence, some of the topics on this forum should be turned into folders where this info could be accessed by future Lost Generations. Permission granted to edit out my ranting.
‘Night all.
GS
Gustavo_Sanchez
Jun 11, 2004
"Chris’s comments, if valid"

Pshe… I don’t know… He’s just one of the blokes that makes Photoshop… (take a peek to the splash screen) 😉
MO
Mike_Ornellas
Jun 11, 2004
Though I’m not an author of a book, I know a few.

Authors that is.

;o)
PF
Peter_Figen
Jun 11, 2004
zloi,

The biggest reason is that your 2200 isn’t going to last forever, and you might, as some point want to try a different paper in that printer, which will, of course, change the entire equation, and, as I said earlier, these printer DO change over time. I have to reprofile mine every six months or so because of subtle but noticeable changes in the way it prints. A photographer friend of mine did exactly what you are asking about, and then had to trash all of his files when he got a new printer, but hey, no one has a gun to your head. It’s a free country (sort of) and every thing that’s been written in this thread is based on peoples real world experience from people who are trying to help you avoid big problems down the road.
Z
zloi
Jun 11, 2004
to Peter: click.

to Gustavo: sorry, I’m not a PS groupie. To be honest, I thought some elves way up north did all the work.
CC
Chris_Cox
Jun 11, 2004
elves? Hey, I’m not that short, and I do NOT have pointy ears!
MO
Mike_Ornellas
Jun 11, 2004
we CAN fix that.
LH
Lawrence_Hudetz
Jun 11, 2004
Hub boy, can we do that! <grin>
BL
Bill_Lamp
Jun 13, 2004
/\ /\

© ©
^
_

Fixed that.

Bill
I
info
Jun 17, 2004
E-Mail:

good thanx
schrieb im Newsbeitrag
I’ve been having a hard time grasping this issue. Working from a scanned
image, which initially has my scanner’s profile, I move to Adobe RGB to perform all my editing. When I’m done and ready to print, I can check how the image will look in my selected profile (in my case, I output everything to an Epson 2200) by going to View/Proof colors. The final step before printing is to go to Print Preview and switch from Adobe RGB to whatever profile I choose (generally for the type paper I am using) and selecting "No Color Adjustment" in the final print screen. I have worked out this technique based on what I have heard and read from "the experts," ie, anyone more knowledgeable than I.
Question: why am I bothering to work in Adobe RGB at all? For a time, the
first thing I did with my scanned file was go to Image/Mode and Convert it to whatever profile I anticipated printing from. But since the "Convert to profile" command, as I understand it, changes the actual color values in the file while keeping their on-screen appearance about the same, it is not intended to give an accurate idea of how the printed file will look. The "Assign profile" command actually changes the colors on screen (without doing so in the actual file) so you can see the difference in using various different profiles. But just as a painter can only paint from whatever colors he can create on his palette, why would I not want to work with just those colors that my printer (in this case, the 2200) will actually print for the given profile? What good does it do me to edit in a mode which may give me a lot of out-of-gamut colors? I mean, so long as I can access all the filters, commands, and controls available in RGB mode, if I know my output profile in advance, I don’t see why I should not edit in that mode from the get-go. Anyone have any comments or suggestions in this regard?

How to Improve Photoshop Performance

Learn how to optimize Photoshop for maximum speed, troubleshoot common issues, and keep your projects organized so that you can work faster than ever before!

Related Discussion Topics

Nice and short text about related topics in discussion sections