Forums

We have had some lively discussions regarding late binding, one in particular comes to mind:

http://forums.adobe.com/thread/477312

The crux of the matter is the conversion of RGB to CMYK, and when it occurs – at design stage, or prepress stage. Here are some hurdles to overcome:

1. There are many different RGB color spaces, of various shapes and sizes. It makes things confusing.

2. There are many different CMYK color spaces, of various shapes and sizes. It makes things confusing.

3. CMYK is usually a much smaller color gamut than a source RGB gamut. This can be difficult to explain sometimes. If a designer loves the source RGB, the CMYK destination can oftentimes be a letdown.

4. Color management works as long as everyone in the production line does his or her part. It is easy to adhere to a good workflow, but it is also very easy not to. And it only takes one user to break the chain.

5. Good proofing is essential. This includes soft proofing and paper proofs. Time, money and effort are necessary to make good proofs. It may be tempting to cut corners, but it's much better to invest in a good proof than be disappointed in a print result.

That is a valid question.

I personally like Lab but by and large in the design community I believe it's not as user friendly, certainly not in Photoshop.

But it goes beyond that. Since there is no Lab device, no image starts out as Lab. The initial photo image capture has to be some sort of RGB, whether it's a digital camera or scanner that's used.

Not only that, the monitor is the medium for viewing files and monitors are RGB devices. Lab is monstrously huge. If you design in Lab there is great potential for creating colors way beyond the monitor's gamut (however the same can be said for large RGB spaces like ProPhoto).

Also, placing RGB images and converting to CMYK on output from InDesign is a workflow that is increasing in popularity. I don't believe you can convert these images to Lab when outputting from InDesign (I could be wrong about that). On the other hand preserving Lab color in output is possible, and technically it is the proper color space for Pantone spot colors.

Lab is extremely useful for measuring colors in the real world. For example, if you have a previously printed sample there's no way to know the CMYK values. But you can always use a spectrophotometer to measure the color, and then the color is defined. Once defined it can be mapped to RGB or CMYK.

Actually two users can share Lab files and see consistent color provided the monitors are very accurately calibrated and profiled. Monitor calibration devices are based on Lab readings so good color communication using Lab files is possible.

Likewise two users could share an Adobe RGB file and if one of the monitors is inaccurate, the color is still not communicated properly. So the RGB color space does not create an advantage over Lab in that respect.

But the problem remains, as pointed out that Lab transcends any monitor. There is a chance that the file has colors that you don't see. One way I know of to check for this in Photoshop. Go to View: Proof Setup. By Device to Simulate select the monitor profile. Rendering Absolute. Hit enter, then View: Gamut Warning. This shows the file colors that don't exist on the display.

However there is another big problem using Lab images pertaining to InDesign. InDesign has no Lab flattener color space. I see problems if a Lab image interacts with transparency. That may be enough reason to steer clear of Lab images.

I guess I'm not understanding right. I thought Lab colors were more or less fixed. For example L25 a15 b-68 (Pantone 286 C) is what it is (at least in the proper viewing and lighting conditions).

If a Lab color display is untrue, it either means the color is beyond the display gamut, or the monitor is not calibrated and profiled properly.

Your example of two monitors, one calibrated to 5000K Gamma 1.8, the other 6500K Gamma 2.2 probably would display the same Lab image differently. But wouldn't it also display the same tagged RGB differently? My thinking is that one of the monitors is more accurate than the other. Or maybe one is more accurate in one part of the spectrum, the other is more accurate in another part of the spectrum.

A tagged RGB image still relies on Lab color as an intermediary for display purposes, so there could still be disparity between the two displays. The difference with a Lab image is the source color is already committed. What remains uncommitted is the destination color, CMYK in the instance of print design. I would think two users with different accurate calibrations could still soft proof Lab color in the same destination CMYK and maintain good color communication.

Spot colors have Lab definitions too, and a user can soft proof a spot color as long as it in within the display gamut.

Not trying to be disagreeable, I know you are knowledgeable in this area. Maybe my logic is wrong?

you can start your paper off by saying that with all the advancements in software technology the print indistry is still processing garbage.  Enough said.

I don't think I understand your point, Rob. If I place an RGB image in an InDesign document, and ask for CMYK output, at print/export time InDesign converts the RGB data to my desired CMYK space - using either embedded or assumed ICC profile data for the RGB image, and the targeted CMYK profile. I think it does this essentially by converting it from RGB to Lab, and then Lab to CMYK. If I place an Lab-based image into InDesign, and ask for CMYK output, it seems to me that the application simply doesn't need to do as complex a transformation.

I've just tried this with a few images - RGB and Lab versions side-by-side in an InDesign document, and the output seems pretty much the same - at least to the casual observer.

Can you please explain your thinking a bit more?

BTW, I wouldn't particularly want to edit in Lab, as I find it a bit counter-intuitive, but I suppose that may just be a matter of what I'm used to. I also used to find RGB editing a bit odd...

I've just tried this with a few images - RGB and Lab versions side-by-side in an InDesign document, and the output seems pretty much the same - at least to the casual observer.

 

In order to display Lab it needs to get converted to RGB for display (you don't have a Lab monitor right?). So if you make edits based on your displayed soft proof, what RGB space are you working in? I'm pretty sure it's your monitor RGB (see my #8). So as long as you display the Lab image on your monitor and stay in a closed loop, there won't necessarily be a problem making a conversion from Lab to CMYK or displaying Lab.

It gets more problematic if you send out the Lab file. I would have a hard time replicating your softproof of Lab on my display, because you might be calibrated and profiled at 2.2 gamma and 5000K, and I might be at 1.8 and 6500K. Lab is going to display in our respective monitor RGB specs differently. If you convert the file to AdobeRGB and send it to me with a tag, then we at least have a chance of seeing the same softproof because now the file's RGB specs are the same—2.2/6500/Adobe Primaries (assuming our monitor's are accurately calibrated so the conversion into AdobeRGB is correct).

You could argue that there would be no harm in placing Lab and exporting a PDF/X-1a and letting the conversion to CMYK happen on the other end, but there's no advantage and there's still the problem of no source profile and a potentially inaccurate softproof at the other end.

OK. I see where you are coming from, Rob. I was assuming  a single system or at least all monitor calibrations constant.

I don't quite understand what you mean by "other end" here:

You could argue that there would be no harm in placing Lab and exporting a PDF/X-1a and letting the conversion to CMYK happen on the other end,

I would assume that the workflow would be edit images in Lab in Photoshop, place in InDesign as Lab, export to a tagged CMYK PDF (like PDF/X-1a). That way nobody else needs to know/care about my Lab work. Perhaps that's what you meant above.

This all assumes that there is some reason to edit in Lab which as I mentioned, I do not happen to see... In any case, thanks for the clarification, Rob.

rob day wrote:

 

 

 

I'm not seeing that, I just filled a Lab PS file with a color, dup'd it, put both into ID and exported to PDF/X-1a and got the same values. ID's Separation Preview is also showing the same separation.

That's right, but you cannot convert to Lab from source color using InDesign.

With an all Lab PDF there's no calibration, but unlike uncalibrated source color it can't be reassigned. Maybe that would be an advantage of sorts.

In other words, source profile is unknown, source appearance is committed, destination can be whatever you need it to be.

Let me explain

Say you have Pro Photo and you output.

The next user decalibrates your PDF and assigns Adobe RGB as source. Then converts to a given destination. The appearance relative to the first user, not preserved because of the decalibration.

Now the first user converts the Pro Photo to Lab.

There is no opportunity for the 2nd user to false profile because it's Lab.

Just a note, I don't think a Lab workflow would be useful, but you never know, maybe there's someone out there who wants ID to convert to Lab on output.

rob day wrote:

There's nothing in the Lab file that tells user 2's color management system that the source view display was ProPhoto  1.8|5000K|Custom Primaries

There doesn't need to be. The ProPhoto1.8|5000K|Custom was already honored in the previous file conversion to Lab. Bottom line the Lab result

Pro Photo > Lab

is what it is, whether a display does it, or if it's a file conversion.

rob day wrote:

 

If that were true we would have no need of RGB profiles and everyone would be sending out Lab

The reason people don't send out Lab is

1. From InDesign you can't convert to Lab, which means a bunch of extra conversions in Photoshop and duplicate files.

2. Lab is never a source, and it's not a device output (monitor, printer, press). With Lab you're in the intermediary. I'm on my way but I'm only halfway. So most people don't bother with it.

3. You can't read Lab values in Acrobat (at least I don't think so)

4. Lab does not yield as much design capability in Photoshop (filters, etc)

5. Illustrator doesn't have a Lab mode

6. InDesign has no Lab blending space

7. There are a lot of people who just don't like Lab color, because it's bigger than anything else (except ProPhoto), it's strange, and the numerical values are not user friendly.

rob day wrote:

 

 

If you display Lab 50|0|0 on a monitor profiled at gamma 1.8 and then on a 2nd monitor profiled at 2.2, the same value is going to have to display darker on the 2nd monitor—it doesn't matter what kind of conversion got you to the 50|0|0 value.

If that's true, then source Pro Photo RGB:

100|100|100

that produces the Lab output

50|0|0

would also be inconsistent on the two monitors.

To summarize, this is what I think you’re implying:

1. When a user looks at a Pro Photo, or Adobe RGB, or whatever, Lab doesn’t figure in at all. Somehow the RGB file goes straight to the display RGB.

2. If RGB values do get converted to Lab in a file conversion, the initial color frame of reference is lost forever. When different users open up the Lab files on different calibrated profiled displays, they all see different results, even though the Lab numbers are fixed and the Lab color space defines the gamut of human vision.

3.  If I take an RGB file, make a copy and swing that to Lab, the copy looks like the original on my display. But if I send both files to someone else, somehow there’s a chance the two could look different on that person's display.

4. If User 1 submits Lab files to User 2, soft proofing goes out the window.

5. Somehow user 2 could unintentionally wreck the source color, even ought there’s no profile to strip away. Bear in mind: user 2 is not trying to edit color. This person is building an InDesign document, or submitting to a RIP, or some other later stage part of the workflow. Both parties agree on the destination color space.

If you agree with all those things, OK, we can agree to disagree.

Technically with ProPhoto there are colors that can be lost in a conversion to Lab. But these colors belong to other species and aren't part of what we observe in the natural world, and I don't think they would cause disparity on different monitors.

It is my belief that the reason hardly anyone uses Lab is because it’s not a device space. Nothing is captured in Lab, there’s no Lab proofer, and there’s no Lab press. But it’s still the gateway for all conversions and the foundation for color management.

rob day wrote:

Not if it has a profile included—in that case its display gamma is 1.8 on both monitors.

The 1.8 gamma is honored in a file conversion to Lab.

ProPhoto and other RGB spaces do have synthetic colors that reside outside Lab:

Of course, a display gamma can be anything. And the Lab image has no gamma.

But it's not a requirement with Lab. The 1.8 gamma of ProPhoto has been replaced with Lab values.

Here are 2 soft proof color conversions when viewing ProPhoto on 2 different calibrated monitors:

ProPhoto, 1.8 gamma, 5000K > Lab > Monitor, 1.8 gamma, 5000K

ProPhoto, 1.8 gamma, 5000K > Lab > Monitor, 2.2 gamma, 6500K

The image should display consistently on both. And Lab is the intermediary.

Now the first part of that chain is replaced with an actual file conversion. Yes, the resulting Lab file does not have a gamma or white point. But it doesn't need it. These attributes have been replaced with Lab values.

Lab > Monitor, 1.8 gamma, 5000K

Lab > Monitor, 2.2 gamma, 6500K

will still display consistently. It's not like the ProPhoto appearance was discarded in a conversion to Lab. The colors that make up the file are still there in Lab numbers, and can be soft proofed on a calibrated monitor.

No, I'm saying that the Lab will have the same appearance as the RGB from which it originated on the same monitor.

My monitor is calibrated to 2.2, 6500K. I can open a ProPhoto, convert to Lab. The appearance does not suddenly change, even though my monitor is not 1.8 gamma 5000K like ProPhoto.

I can send both the ProPhoto and the Lab to someone else with a monitor calibration of 1.8 gamma, 5000K. This person opens both the Lab and the ProPhoto and the appearance of both is consistent on his or her display.

The appearance from one monitor to the next may not be entirely consistent, because the mediums have different calibrations and color spaces. Much like printing on different substrates. You may convert a file properly for both substrates and the intent is to preserve appearance, but the appearance cannot be entirely consistent, because the color gamuts are different.

So yes, 50|0|0 Lab can look different on different displays. But so can 100|100|100 ProPhoto.

100|100|100 ProPhoto converts to 50|0|0 Lab, no matter what monitor is being used. Lab is always the intermediary, even if the conversion is a dynamic soft proof conversion that occurs automatically without user intervention.

The promise of color management was to ensure consistent Lab values in any conversion. To do this ICC profiles were required for the device dependent color spaces, to ensure a proper transition to and from the intermediary Lab.

If user 1 converts the file to Lab, we're already in the intermediary.

rob day wrote:

If you send me a tagged ProPhoto file it displays to the profile's specs (1.8|5000) not my monitor specs (2.2|5500).

It doesn't display to the profile's specs. It translates to your monitor specs. Through Lab.

The monitor can't display ProPhoto. A conversion must take place.

When you convert a ProPhoto to Lab, do you see a big appearance shift?

If you sent both files versions to me, would they somehow look different on my monitor, side by side?

The image is 1.0, converts to Lab, then converts to the monitor which is 2.2. So yes, it is displayed at 2.2 in the final soft proof.

In theory no, but the two monitors have different color spaces.

Different white point temperatures can make significant differences in appearance. Just like paper with color casts. If I have a bluish stock, then a yellow stock, a relative colorimetric conversion to the CMYKs for each of these will try to account for the paper casts. But the appearance will always be different, because 0C 0M 0Y 0K don't look the same.

rob day wrote:

 

To keep things simple just consider gamma. If I convert RGB 100|100|100 from my custom 1.0 ProPhoto to 2.0 ProPhoto the resulting value is considerably lighter 150|150|150.

But with this very step you're going through the Lab portal again.

That's what makes the whole color managed environment work. Lab.

So if someone converts to Lab they can't possibly break CM. The only thing lost are the name and parameters of the RGB source (ICC profile) and the RGB numbers. But the appearance of color is completely and utterly intact, cemented in the file conversion to Lab.

That's why the old Linocolor workflow worked pre-color management. You didn't have to worry about someone altering untagged RGB appearance unintentionally, because the step to Lab occurred at the scanner capture stage.

Again – take any RGB file. Convert to Lab. Do you see a color shift? With the RGB and the Lab result side by side – do they look the same? I'm sure they do.

If you send both of those files to me, and I look at them both side by side on my display – will they look different? No.

So if I receive a Lab file, how can anything relative to source appearance be sacrificed? If it, and the source, side by side, would look the same anyway?

Thanks, Rob.

I'm not so sure about this ProPhoto to Lab thing now. If I fill a ProPhoto document with

0R 0G 255B

then convert to Lab, there is a significant shift on-screen.

If I fill ProPhoto with the same color, and convert to Adobe RGB, no shift on the display (and the numbers stay the same too). If I then take the Adobe RGB and convert to Lab, no shift. So going that route, strangely enough, I end up with what looks like a much better Lab match to the original ProPhoto (but may in fact not be).

I realize the original ProPhoto blue is way, way beyond my monitor. I guess it's beyond Lab too. But how is the display shift when converting from ProPhoto straight to Lab explained? My argument earlier was that a conversion to Lab has to happen as color translates from source to monitor, behind the scenes, when viewing any RGB image.

What's strange too is that I don't see the shift when converting from ProPhoto to Adobe RGB. I thought Lab was in between every conversion. But if I go ProPhoto > Adobe RGB, I get a different result than when I go ProPhoto > Lab > Adobe RGB. What sense does that make?

This may not matter a lot because the ProPhoto blue is something you would never see in a normal image. And so far with every RGB image I test, I see absolutely no shift moving to Lab. It only happens with the fringe ProPhoto blue.

Maybe you have thoughts on this.

(To the original poster I apologize if I am getting too far off subject)

OK. I still don't fully understand, but enough about ProPhoto.

Switching gears – let's assume that both artists and commercial printers subscribe to a late binding workflow.

This means that all files are set up in source RGB (or Lab), and conversion to the destination CMYK is done by the printer.

Images would normally be RGB. Swatches (except for [Black] could be Lab, or perhaps RGB.

Who benefits more from this, the designers, or the printer?

What about "flying blind" with source color?

Meaning the artist has all proper color management in place, but has no way of knowing destination color, because the print buyer hasn't finalized who's printing, or the job specs.