TIFF and LibTiff Mail List Archive

1994.11.10 22:24 "CIE Lab images", by Gordon Clare

I have a question regarding the use of Lab in TIFF images.

The "a" and "b" components are in the range -127 to +127. My question is, how are these values stored in the TIFF file?

The impression I get from the spec is that they are stored "as is", ie, as signed values in an 8-bit byte, although it is not explicitly stated anywhere. It would seem preferable though to store them re-ranged as 0 - 255, ie, effectively add 127 to these components. This is pretty much what happens to the Cb and Cr components for YCbCr (using the reference white and reference black field).

Simply storing -127 to +127 in a byte and then treating it as unsigned will place the extremes of the range (-127 and +127) right in the middle of the "new" range (0 through 255). If lossy JPEG compression is performed that's expecting unsigned byte values as input, you'll get very bad things happening as -127 could easily change to +127 (127 goes to 128 in the unsigned value) etc.

Is it as simple as defining reference black / reference white settings to perform the appropriate transform, like what happens with the Cb and Cr components of YCbCr?

Lab is not mentioned at all in the reference white/black section of the colorimetry section of the spec.

Does anyone have any examples of Lab images that either they have generated or that they have picked up from somewhere? If so, I'd appreciate a couple of them!

Thanks in advance,

- gordon

    Never keep a dog and bark for yourself.                     -- harrison
Gordon Clare                                 Net:
Canon Information Systems Research Australia  Ph: +61 2 805 2738
P.O. Box 313 North Ryde, NSW, Australia 2113 Fax: +61 2 805 2929