LAB Mode with 8-Bit Images ???

[slejhamer]

I've heard that I shouldn't use Photoshop's LAB mode to process the 8-bit images I get from my G1. (JPEG or RAW converted to TIFF.)

I'm still a relative newcomer to Photoshop, and have not tried LAB mode until very recently. (This past weekend, to be specific.)

However, it seems to me that the results I got by adjusting levels and sharpness on the lightness channel in LAB mode were much better than what I got making the same adjustments in RGB mode. In particular, adjusting levels on the L channel seemed to work better than contrast masking in RGB as there was no halo effect. Colors seemed more natural, too.

So what's wrong with processing 8-bit images in LAB mode? Will my prints (generally 5x7, but sometimes 8x10) look bad? What "damage" am I doing to my images when I convert back to RGB?

And, if LAB mode is really a no-no, then is there a technique to reduce or eliminate the halo effect from contrast masking?

Thanks in advance,

[Roger_Cavanagh]

Mitch,

I don't know why 8-bit images should have a special problem with Lab mode. PS uses Lab as the intermediate space when converting from one colour space to another, so I don't see why 8 should be different from 16.

Sharpening the lightness channel in Lab mode is quite a well-used technique. Pekka uses it in LS342, Fred Miranda uses it in EdgeSharpen, the DigiEdge chap (wojcek?) uses it.

I don't believe there should be any degradation converting from RGB to lab and back. However, I did find one banana skin wrt colour spaces. PS has a working colour space; most people use Adobe RGB. If you are working with an image that has a different colour space, you can convert to Lab - no problem - but when you convert back to RGB, PS will convert to the working space not the one your image was previously tagged with. This circumstance can _seriously_ screw your images.

I would suggest you take a look at Fred Miranda's EdgeSharpen action. For a few bucks, you get a good tool.

Regards,

[slejhamer]

Thank you for your comments, Roger. Sounds like I've nothing to fear with respect to Lab mode.

I had read (on other sites) about quantization errors specifically when converting 8-bit images to Lab, and was worried that this could lead to artifacting.

I've also read that all conversions to Lab, whether 8- or 16-bit, are "destructive." If it's on the internet, it must be true... ( Ha! :) )

I did pick up the "light" (freeware) version of Fred Miranda's EdgeSharpen - a nice tool, indeed. I didn't realize that it used Lab mode - thanks for pointing that out. I also use unsharp mask with a high radius/small amount, which I saw described as a "haze remover." On my images it seems to be a nice complement to EdgeSharpen.

Thanks again,

[Roger_Cavanagh]

Mitch,

ALL edits in Photoshop are destructive. That is the name of the game. Conversion in and out of Lab mode will not leave your image entirely untouched, so I'm not recommending going back and forth willy-nilly. :)

However, as you have discovered sharpening the lightness channel can be much less destructive than USM in RGB mode.

You can examine how much the image changes by looking at the histogram before and after Lab conversion. It does change, but not much - a 1/100 here and there on the mean and standard deviation. You can assess the impact of the changes with this procedure:

1. Open an image, convert to 8-bit, if necessary.

2. Duplicate the image.

3. Using the dupe, convert to Lab then back to RGB.

4. Using the dupe, select all pixels and copy.

5. Go back to the original and paste a new layer.

6. Select the new layer and change the blend mode to "difference". This will send everything black as it show the RGB differences between the two layers.

7. Run histogram and you will find that the vast majority of pixels are 0,0,0 or 1,1,1 with a handful showing differences greater than this.

Regards,