Coverting sRGB scans to Adobe (1998) RGB

[jason_elsworth3]

I recently used a scanning service who produced 67MB tiffs for me

from 35mm slides using a Nikon Coolscan 8000. My expectation was that

they would be in the Adobe RGB (1998) Colour profile (the scanning

company said this was their default profile, it is the profile

required by Almay and is considered the best colour space for CMYK printing).

 

However when I open them in PhotoShop I get an

embedded profile mismatch message (because it doesn't match my working

space of Adobe RGB (1998) saying that Nikon sRGB 4.0.0.3001 is the

embedded profile. I am a bit concerned, as obviously sRGB isn't

generally consider a good choice for pre press work. I have contacted

the company who did the scans and they have just said I should convert

the images to Adobe 1998. Should I be concerned and is converting from a

narrower to a wider Colour space like this a viable option? My bottom line is

will these files be suitable for CMYK printing after conversion?

 

Any advice would be greatly appreciated.

[william_john_smith]

Send it back and have them do it right. The company stated that Adobe RBG was the default

profile so that is what you paid for. Them telling you to convert is unprofessional to say the

least. You can also check with who is going to do your CMYK work and see what they think.

[AaronFalkenberg]

Unless there are some highly saturated reds and greens, your images should be fine. If these are chromes, you should take a good long look at them on a light table to see if the Nikon sRGB scans have captured the hues faithfully. You won't lose anything by converting to A98.

[mike butler]

I think they've given you a device-dependent rgb that has to be a converted to a working

space. With that size of file, I would think any data loss would be trivial in 16-bit.

[timothy_hughes]

"Should I be concerned and is converting from a narrower to a wider Colour space like this

a viable option?"

 

Going from sRGB to Adobe RGB is not going to increase the color gamut (as I'm sure you

already know). It's just going to change the color mode. Colors should remain nearly

unchanged. You could probably get away with converting the files to Adobe RGB and no

one would likely know, lol.

 

"My bottom line is will these files be suitable for CMYK printing after conversion?"

 

Yes, it should be fine. Designers and printers prefer Adobe RGB for conversion, but CMYK

is a limited color gamut anyway. It's up to you about how critical you want to be about the

color space. I'll tell you, if the photos are going to end up in newsprint, or on the web it

won't make a darn difference.

 

Commercial photo processors use the sRGB color space. Photos output from them look

perfectly good to me! That being said, I shoot for my stock agency in Adobe RGB usually. I

don't sweat it if my work is in sRGB mode though.

 

-Timothy

[mike butler]

Jason,

 

Another thought: A "good" scan to me looks flat and unsaturated, but it contains all the info

that was on the chrome. That's what I want as opposed to rigidly faithful color and gamut

matching. You're not going to escape levels and curves and color management even with

drum scans, but you can automate to that a degree.

 

Best of luck with your project.

[twmeyer]

"<i>You won't lose anything by converting to A98</i>"<p>That is true, but this just means you won't miss what you've never had. One way to see the difference would be to pay for one of the same originals to be scanned (again) in Adobe RGB 98 and make your own comparisons... t

[ellis_vener_photography]

The question is whether or not they did scan them in sRGB or Adobe RGB(1998). They

make actually have scanned them in Nikon's Adobe RGB(1998) color space but given the

way Nikon tags their images, Photoshop (which version are you using) may be looking at

the tag and interpreting it as Nikon sRGB 4.0.0.3001.

 

Try this: when you get that profile mismatch error try choosing the no color management

option and then once the image is open in Photoshop converting to Adobe RGB(1998).

 

And now is a note for the color persnickety. Adobe RGB (1998) actually clips some of the

gamut E-6 and other transparency films are capable of recording. For scanning I

recommend either "Ekta Space PS 5, J. Holmes" (which is free) working space or the more

perceptually more uniform and recently revised "Chrome Space 100" (which isn't free)

working space also from Joseph Holmes. You can read more about these and download

the free "Ekta Space PS 5, J. Holmes" working space and a really well written PDF

explaining this by going to http://www.josephholmes.com/propages/SpacesandSets.html

 

of course to scan into directly into these spaces the people doing the scan have to be

using scanning software that let's them choose alternatives to Nikonscan's limited

selection. Alternately have them scan into Wide RGB and d do the conversion yourself.

 

Depending on which CMYK profile (there are literally millions of them depending on the

press being used, along with the inks and the paper) the generic CMYK gamut is a bit

larger than sRGB i n the yellows (i.e., sRGB will clip the saturated yellows --think wild

flowers, yellow rain coats, etc. and sRGB will clip off a goodly slice of the greens and

cyans as seen in the attached screen capture from Chromic Color-Think 2.2 ( http://

tinyurl.com/g8wjg ), a piece of software I find invaluable and useful for really knowing

what is going on with color spaces and for analyzing individual images.<div></div>

[jim_doty]

Jason,

 

I seem to recall reading lately (maybe in one of Tim Grey's DDQ emails) that using a Nikon Coolscan scanner set to capture in Adobe RGB (1998) can oversaturate colors from some transparencies, so sRGB is a better initial scan setting. Then convert to Adobe RGB (1998) in Photoshop. My own very limited experience with a recently acquired Nikon Coolscan seems to confirm this. When I tried using Adobe RGB (1998) some of my scans looked garish compared to the original slides which are more "neutral" in saturation. I switched back to sRGB on the scanner.

 

Someone with more experience can give you a better answer.

 

Jim

[jason_elsworth3]

Try this: when you get that profile mismatch error try choosing the no color management option and then once the image is open in Photoshop converting to Adobe RGB(1998). Have tried this and it converts very quickly if that is anything to go on?

 

Jim - if memory serves me right Tim was refering to Nikon Scanner RGB, not sRGB

 

Thanks for evryones help so far. I am tending towards thinking it is OK, but am going to check what the actual capture space was.

[Tim_Lookingbill]

Look at your previews and check each RGB channel histogram

for clipping. If normally saturated=(correct looking) colors are

clipped, it means they squeezed a wide space=(scanner

sensors combined with scene gamut capture capability of the

media scanned-E6?) to the narrow space of sRGB.

 

If the saturated colors are close to around 230-240 or 15-25 in

one of either of the RGB channels then it was most likely written

in AdobeRGB. You'll know from assigning either space to get the

correct look as to which it is as well.

 

And about color spaces, the operative word is Working Space

when it comes to choosing which. A larger space is better to

work in. If that Nikon wrote directly to sRGB it might have thrown

out color data that is lost for good and converting and EDITING in

AdobeRGB won't bring it back, but it's still a better space to edit

in and thus convert to a smaller space with like CMYK while

having Soft Proof turned on.

 

Most scene gamuts fit well within sRGB but don't leave much

color editing options when Soft Proofing certain colors to CMYK

like yellows mentioned before. In sRGB it's hard to get a clean

cyan-less 100% yellow, but in Joseph Holmes you can sculpt

quite a bit of clean detail in sunlit yellow flower petals.

[tail]

Jason,

 

I have examined the color-management issues of the NikonScan program recently with a 9000. During scanning with that program, one has the options of choosing a color space (sRGB, aRGB, scanner_RGB etc) or none at all. What I found out was that choosing a color space is equivalent to assigning that space in PhotoShop. Your files being identified as having "Nikon sRGB" is a consequence of having chosen either sRGB or scanner_RGB during scanning (they result in identical files). In order to get what you originally expected (as if choosing aRGB during scanning), you need to "assign" to aRGB, not "convert", contrary to what most people mentioned here.

 

It is worth noting that "assign" will change the colors according to the destination space, but it is a non-destructive, reversible operation, so you can feel free to experiment.

[Tim_Lookingbill]

Tai,

 

That's interesting to know because it leads me to suspect the

same thing with CM integration on my Epson 4870 flatbed.

 

There's a misunderstanding of terminologies and of what type of

preview is being generated at the time of capture when the

scanner CM Source setting requires you choose the scanner

profile, in my case Epson Standard, and a Target working space

like sRGB or aRGB.

 

The confusion comes in when the preview changes within

EpsonScan when I pick a different Target working space which I

was under the assumption meant that a conversion was

occuring much the same that happens in PS where the preview

DOESN'T change.

 

The preview changes much the same as ASSIGNING the

working space which can make editing within the scanner

software even more problematic. I don't know if it's as much a

problem on a Nikon, but many complain about Epson's

subsequent oversaturated previews within PS after editing in

Epson Scan with these types of CM previews. Since Epson

doesn't embed the profile many are left wondering what to

assign once opened in PS because the edits don't translate very

accurately assigning any profile.

 

Nice to confirm it also happens with Nikons as well. What's up

with scanner software these days? Geez!

[howard_owen]

"Your files being identified as having "Nikon sRGB" is a consequence of having chosen either sRGB or scanner_RGB during scanning (they result in identical files)."

 

I don't think they do result in identical files. The scanner_RGB profiled scan will come into Photoshop claiming to be sRGB, but it does not, in fact, contain an embedded profile and the gamut, according to Nikon's documentation, is the widest possible that can be captured with the scanner (using color management). You will note that when the CMS is turned off, the scanned file still comes into Photoshop as sRGB. This is because:

 

"Photoshop... reads info from color profile and EXIF data. When color management on Nikon scanner is turned off NikonScan still specifies sRGB in EXIF data and Photoshop reads this data and interprets it as if image had sRGB profile embedded." I am reasonably certain that this applies when using scanner_RGB as well.

[tail]

I was aware of the behavior of the NikonScan processed files when opened in PS. My result was not based on that. Here is what I did: I made a scan with NikonScan in its NEF format, without color management. Then the NEF file was reopened in NikonScan to generate images using various color management options. The ones corresponding to sRGB and scanner_RGB are identical bit by bit. They are differnt from those that use other icc's.

 

This is a disturbing result because as Howard noted, sRGB and scanner_RGB are supposed to have the narrowest and widest gamut on the list on NikonScan. In comparison, SilverFast has much more reliable and predictable color management -- if you have a custom profile for the scanner, you can use it (once again, it is an "assign" operation) during scanning, which NikonScan cannot (the list of icc available to NS is fixed).

 

I used NikonScan version 4.0.0 and CoolScan 9000 for the tests. It would be nice if someone else using other combination can confirm that.

[jason_elsworth3]

The profile is defintely embeded rather than being read from the exif data. I verified this by downloading a script that makes Photoshop 7.1 ignore exif data.

I have used soft proofing (working CMYK) and gamut warning to look at some of the slides and there is defintely some clipping in saturated colours, particlay ***lows. This clipping seems to remain at the same level whether I leave in sRGB or convert to aRGB. Interestingly ignoring the embeded tag and then assigning a profile seems to make the out of gamut problem much worse. Having read a multiple of posts on this subject it all still seems a little confusing and confirms just why I didn't buy my own scanner :) Here's hoping my eventual move to digital wont be quite so perplexing:)This whole process has however taught me quite a bit more about colour management.

[howard_owen]

Thanks for the input, Jason and Tai. In a word, this sucks. Ok, that was two words.

 

Jason, when you say that, "Interestingly ignoring the embedded tag and then assigning a profile seems to make the out of gamut problem much worse," I would expect that to be true given that CMYK does not handle most saturated colors (blues, greens, purples, reds...) at all well--they do not exist in the CMYK gamut. So any increase in saturation of native RGB files can only result in more out of gamut colors on the CMYK end. One would hope (for accuracy's sake) that not only are the colors becoming more saturated when a profile is assigned, but also they are becoming more numerous as well.

 

Tai, what are your suggestions for a workflow if one wants to capture all the colors the scanner can see? Should I ditch NikonScan or can it be coaxed into giving me what I want?

[tail]

What Jason just described agrees with my own experience too. Although I tend to be less pessimistic about going digital by scanning films, the process is tedious in the early experimental and calibration stages, and does require substantial financial investment if you want the best out of films.

 

The answer to Howard's question depends on whether you have a custom profile made for your specific scanner or not. Without a scanner profile, I would scan with NikonScan's CMS on, choose PS' working profile there, output as 16-bit TIF, open the TIF in PS to edit the color and contrast in order to match the original (e.g. slides viewed simultaneously on a lightbox). The modifications (via curve, level and saturation etc), once fine-tuned, need to be saved and then applied to images from the same type of film. Since it relies on visual comparisons, it is not gonna be very accurate but a profiled monitor is a must. If you work mainly with 8-bit JPG, then it would be better to do the matching within NikonScan in order to optimize the use of limited latitudes (assuming NS applies the edits before reducing the bit depth).

 

The workflow is simpler if you have a custom scanner profile: you scan with NikonScan with arbitray CMS settings (because NS only "assigns", you can always reassign in PS with no loss of data), without any editing in NS. Open the 16-bit TIF in PS, assign it to the scanner profile, then convert it to your working profile (say, aRGB). If the scanner profile is properly built, the result should be a close match to your original. Any further editing should be done only after conversion to the working space, to avoid clippings introduced by converting from a wide scanner space to a narrow working space.

 

I actually use Silverfast (SF) for its convenience and speed over NS. With SF, I scan at high resolution once for an image to archive it as a HDR file. The HDR file can then be used over and over again to produce a TIF or JPG at any (equal or lower) resolution and bit-depth to your liking. Color management, done during the conversion from HDR to TIF/JPG, does both the assign and convert outlined above.