1994.09.20 03:03 "TIFF CIEL*a*b* Question (with Answer)", by John M Davison
A while ago, I posted the following question to comp.graphics:
> In plain English, exactly what compression schemes are allowed for
>CIEL*a*b* images?
Tom Lane (tgl@netcom.com) replied:
>Any compression method defined by the spec. Where's the problem?
I replied:
> One problem is on page 30 of the TIFF 6.0 Specification, in which it is
> said that "CCITT Group 3 1-Dimensional Modified Huffman run-length
> encoding...is defined only for bilevel images."
Tom replied:
>You're right of course, CCITT fax compression is only capable of working
>on black-and-white images. This is a fundamental characteristic of the
>CCITT method, not some arbitrary limitation imposed by the TIFF spec.
>CCITT is *defined* in terms of alternating runs of black and white pixels.
>
>As far as I can recall without having the spec in front of me, the other
>compression methods are all applicable to color data and thus to CIELab
>data. Whether they offer useful compression is another question...
>but the TIFF standard does not impose any limitations of its own on
>which compression method you apply to what data.
It would seem, then, that Compression values of 2 (CCITT Group 3 1-Dimensional Modified Huffman run-length encoding), 3 (T4-encoding), and 4 (T6-encoding) are acceptable only for images in which all of the following conditions are satisfied:
- BitsPerSample[0] is equal to 1.
- PhotometricInterpretation is 0 or 1.
- m = (number of ExtraSamples) is equal to SamplesPerPixel minus 1.
Is this true?
--
John Davison
Electronic Mail: davisonj@ecn.purdue.edu
WWW Home Page: <http://en.ecn.purdue.edu:20002/~davisonj/HomePage.html>