Profiling, education & death . . .

[Journeyman]

How do we feel about LAB color space (device independent)? If I'm scanning/creating an image, it's my choice because no clipping occurs before an output device (which has its own gamut) tries to maximize/profile said output.
As far as client interaction goes, the analogy I use most often is:
"Ever go to Circuit City, Best Buy, Sears, wherever and see the TV displays? They are all outputting the 'same file', but the screens all look different. Right . . .?"
Sometimes there's a glimmer of "the print shop's environment is different than mine"?! Designers, marketing reps, even photographers love to learn . . .
I try to, in return!

[time2go]

For me I scan in LAB since thats the scanner native color space. Sometimes I'll
do some color cxs in Lab but most of the time I'll convert to rgb and finish all
the rest of the cxs in Photoshop. Since Photoshop is very weak in LAB and it's
stongest in RGB cxs I feel it's best to do my work in RGB. Just my two cents.

[clod]

For me I scan in LAB since thats the scanner native color space.

Not sure what you mean by that...the scanner (any scanner) is starting with RGB, not Lab or CMYK, even if you can't see when the conversion occurs. We at one time scanned Lab, and you can get decent results. You can do some interesting things with editing/retouching, but I found we got more consistent results with properly profiled RGB scans.

[time2go]

I should of said that the scanner software is set up to scan in LAB not RGB.
When I profiled the scanner it was done in LAB not RGB thats why I use LAB.
Just so you know it's a Linotype-Hell ChromaGraph S3400 using Linocolor 6.0

[clod]

Hey, small world...Hell Chromagraph 3800/LinoColor 5.somthin'. That's exactly the workflow we used many years back when we veered off from "direct" CMYK scans. I did like the LCH color setting that Lino provided. I sometimes miss Linocolor, but it really was a big step forward for us implmenting profiled RGB scanning.

[pcmodem]

but what clod is saying is that all scanners scan in RGB and then convert to LAB. So why not just scan keep it RGB?

[mattbeals]

You can get pretty wild with correcting in LAB and RGB. For a majority of work, I think that RGB is quite sufficient. There are times when a image could be "better"/"faster" manipulated in LAB vs. RGB. But for most of us RGB fits most of our work. I don't know of a scanner that scans in LAB, they all scan in RGB as others have stated. So, rather than worrying about capturing LAB vs. RGB I would rather worry about capturing in 16bit vs. 8bit color. Because with either LAB or RGB you still have the same basic gamut. After all, RGB fits inside of LAB (almost completely that is). But 16bit vs. 8bit is a order of magnitude more information. Maybe that's a bit too esoteric? All the digital captures I did were done with a digital camera back (Sinar Braun, Leaf, etc) in 16bit Colormatch RGB. The basic manipulations were done in 16bit RGB and final adjustments were in 8bit CMYK. Worked well for us.

[time2go]

The way my scanner works I kind of look at it the same way as digital camera raw.
Even though it captures in rgb it's then converted to its on color space before you
can touch it. As in the Leaf using HDR and Sinar are STI files. If I could get to the RGB
before its converted to LCH I would but this old scanner was not made that way.
At least its better then even older scanners that you could only get cmyk data.
When you bought this scanner we got a great deal on it because they sold one
to another shop that wanted it to scan straight to RGB but once the had this scanner
for a week the found out that it didn't work that way so the ended up getting a 3900
which does scan straight to RGB. By the way we don't do much scans anymore
maybe 5 a week if that. Almost all of the images we work on are from our Canon D10
camera or our new Sinar Braun 22mp back for our Fuji GX680. And of course files
from our customers who get them from who knows where.

[digitaldog]

-->For me I scan in LAB since thats the scanner native color space.

No like every other scanner on planet earth, it scans in RGB (as has been correctly pointed out above).

So some conversion is taking place. If only done on 8-bit, you?re losing a lot of data. You?re also converting into a HUGE color gamut (the gamut of human vision) which is a far, far cry from the native gamut (actually color mixing function) of your scanner. What does LAB provide? Well it?s self defining. Since it?s a device independent color space, you don?t have to tag the files. There are some controls, like adjusting tone without altering color which is somewhat useful. However, it?s really not worth the potential data damage and the use of a pretty non-intutive color model.

This really isn?t anything like a RAW file which is essentially a Grayscale, linear gamma encoded file.

[digitaldog]

Often, starting with a really big gamut color space isn?t all that it?s cut out to be. This of course depends on the subject and it?s color gamut and the output gamut (and profiles, rendering intent etc).

You can test this yourself. Make a spectral gradient in Photoshop (or better, the Granger Rainbow which is a tad more work). The RGB color space is not important because you?re making a synthetic image. Duplicate this so you now have two identical files with identical numbers representing a spectral gradient.

Now assign sRGB to one, Adobe RGB (1998) to another (or even ProPhoto or LAB). You have two synthetic files defined by different color gamuts (that?s because they are in two different RGB working spaces which define the scale of the RGB primaries).

Now convert each to your output profile and print them out. You should see some interesting results! The source profile description (sRGB, Adobe RGB (1998) etc) play as critical a role as the output profile (which in this test is the same). Play with both the RelCol and Perceptual intent for both tests. Of course, there is no standard in how a Perceptual intent is constructed. It?s up to the profile manufacturer to decide what they feel produces ?pleasing color?.

Point is, a big honking color space, even in high bit can sometimes produce issues with conversions to the same output color space as a smaller gamut. Bigger isn?t always better.

LAB is huge and the only device that can use it in reality is the standard observer (we humans).

LAB was somewhat useful as a ?capture? (from conversion) and editing space prior to Photoshop 5 (some would say 6.0). Now with well behaved RGB working spaces based on math (like LAB), it serves little use.