2008.01.08 15:00 "[Tiff] Tag handling in LibTiff - past, present and future", by

2008.02.01 09:06 "Re: [Tiff] TIFF T.30 vs TIFF Class F", by Ron

While I won't dispute Joris as a greater authority on what the TIFF specs have to say about this, I would suggest that if you are planning to send this to an actual FAX machine, the ITU spec (T.30) is going to have to take precedence (assuming you have a current, error free copy of that).

That is the spec by which any FAX manufacturer is going to have to have their product approved for use, so unless some (all?) actually support both in practice, I don't really see how you can do otherwise.

You might want to take a peek here: http://www.soft-switch.org/ and perhaps have a chat to Steve. He's probably one of a relatively small handful of people who've done much with T.30 in software that might be able to toss you a few practical clues.

Are you certain that the answer is not _both_ (the tiff is encoded lsb-msb while the transmission stream is sent msb-lsb)? I haven't looked too hard at T.30 yet, but I've seen crazier things in telephony. (just look at how MS packs GSM in wav for a fine sample)

Cheers,
Ron

On Fri, Feb 01, 2008 at 09:05:11AM +0100, Joris Van Damme (AWare Systems) wrote:

> Bob,
>
> Bob Friesenhahn wrote:
>> It seems that the T.30 FAX specification says to send the
>> most-significant bits first while TIFF Class F

>> (http://palimpsest.stanford.edu/bytopic/imaging/std/tiff-f.html) >> recommends (requires?) sending the least-significant bits first.

>>

>> I have people in both camps claiming that their specification is the >> right one to use and that it can't work any other way.

>>
>> Can someone please educate me in the ways of FAX and what the best way
>> is to address this situation?
>

> I don't remember a T.30, only a T.4 and T.6. But I assume we're likely > refering to the same thing with 'FAX specification'.

>

> The compression scheme specification specifies a lot of stuff that is > instead specified on the IFD level in TIFF. Most obvious example is color

> and dimensions: the FAX specification says some stuff about how wide images > are, and what is white and black, whilst in TIFF this is determined by the

> IFD ImageWidth and IFD PhotometricInterpretation tags and can be anything. > The same does indeed also apply to bit order, if I remember correctly.

> Thus,

> in TIFF either bit order is correct, and this applies to any compression > scheme and is always to be indicated by the FillOrder tag.

>

> The general rule that applies, is that the TIFF specification is the higher > priority. From the FAX specification we need to read only and exactly what

> is the compression scheme and does fit in TIFF without contradicting or

> overwriting TIFF. Otherwise FAX compression would end up a mess similar to > the OJPEG mess, with tag meaning depending on compressed data properties,

> at
> which point all TIFF hell breaks loose.
>

> I've elaborated on the general rule and some specific applications, with

> regard to FAX compression in TIFF, in this mailing list before, attempting > to help get the record straight. Please see

> http://www.asmail.be/msg0054894761.html. In my experience, most people

> comply with this point of view, I've seen very little of a second camp and > instead, fortunately, much good interchange.