2001.10.05 22:22 "16-bit ICC L*a*b* ?", by Dwight Kelly

2001.10.09 00:12 "Re: 16-bit ICC L*a*b* ?", by Martí Maria

Hi Bill,

TIFF Lab of 16 bits are rare.. I do have only a couple, and if remember ok were from LaserSof'ts twain source and other form ScanOpenICC, I'm not shure of this latter.

8 bit ones are more common, and here Adobe ones exhibit an important difference... they are using D50 instead of D65. This makes the transport process easy, since, leaving aside the different encoding of a, b, the values does correspond to ICC PCS, and thus a 1:1 identity profile can do the trick.

About the profile... I did found a TiffLab8Spac.icm in an old torture-test sample from Microsoft, for Win95 ICM. I figure this profile is in public domain, since Microsoft put it available on their site. Anyway, the profile is easy to duplicate, it only does the a, b encoding translation plus a D65->D50 white point remapping.

16 bit TIFF in general are not very used, but there are some professionals using cameras like the Nikon D1, D1H, D1X that does use very low gamma to store the image preserving dark areas. They used to save in 16 bits to avoid the posterization effects by different gamma of display/printer. Humm.. gamma 1 should *never* be used, but this is like a "hard dump" of the camera memory.

Here 16 bits are needed. Also CMYK separations for high-end prepress are rendered to 16 bits on final stage on RIP, just because the dithering/blue noise stuff.

On the 16 bit Lab, well, this is another question. Lab has a enormous gamut, so converting from any 8 bit RGB to 8 bit Lab is for shure lossy. Just because Lab can contain a lot more values than any 8 bit RGB space. The image will be perceptually similar, but numbers will not match, and a images with bad gamma (1.0, for example) will lost a lot of information. So, Lab16 is required by numerical apps or precission color ones, not for "normal" imaging.

Of course, this is only my opinion and I could be wrong :-)

Best Regards,
Marti.