The TIFF versus JPG debate

[Don]

Armistead, Jason wrote:
> My $0.02 worth on the abovementioned subject:
> 
> TIFF has lots of ways data can be encoded within the basic TIFF file
> structure, including LZW, CCITT G3 and CCITT G4 formats.  If anyone is
> interested, get a good look at the TIFF V6 specification, although the links
> on the www.libtiff.org site seem to be broken and Adobe has now put their
> copy of the TIFF specification behind a whole lot of "developer registration
> required" pages you need to get through first.
> 
> See also http://www.awaresystems.be/imaging/tiff/tifftags/compression.html
> which describes the TIFF tags relating to compression of image data.

I've had good results with TIFF in documents that I have prepared
(user manuals, etc.).  *But*, you can't stray too far from what
is "expected".  E.g., don't use weird BITS_PER_PIXEL values
and expect many decoders to be able to handle your files!  :-(

Also, I have found that many *encoders* fail to sort the tags
properly so when designing decoders, *expect* this problem  :-/

> From where I sit, we've used TIFF with CCITT G42D (fax) compression on
> bitonal (black and white only) image documents for 150,000 engineering
> drawings with excellent results. This is a totally lossless compression
> format (what you get back is exactly what you put in) specifically designed
> to do a good job with what you'd find on pages of drawings or of text.  If
> you don't need colour or greyscale, then CCITT G42D is hard to beat.
> 
> Within the US military, the CALS (Computer-aided Acquisition and Logistics
> Support) Type 1 specification for bitonal images basically wraps the
> compressed CCITT G42D data inside a slightly different wrapper to TIFF -
> CALS is a fixed length text header, and TIFF is a lot of binary stuff.  It's
> easy to convert between the two formats when you know how.
> 
> What I like about TIFF coupled with CCITT G42D compression most is (a) it's
> lossless, (b) supports multi-page documents, (c) it's an open specification
> with (d) an open source library to manipulate the files (LIBTIFF) and (e) it
> is widely supported with hundreds of viewers available (on Windows, the free
> Imaging component works fine for most people).  I can also easily transform
> my TIFF data into Postscript (btw Level 2 Postscript more or less supports
> CCITT G42D compression too), PCL or even into PDF with not too much drama.
> 
> In comparison, PDF is locked in more or less to Adobe's Acrobat Reader (yes,
> I know there's always Ghostscript / GSView and friends ! ), and not as easy
> to manipulate - I call it more of a nearly "final form" document format than
> TIFF.

I thought PDF's could just be used to *encapsulate* regular TIFF's?
I.e., in much the same way that they can encapsulate JPEGs, etc.

> JPEG is not suitable because it was a lossy format targeted at colour images
> more than black and white, and it wasn't multi-page.  I believe the JPEG
> 2000 specification has improved on some of these such as providing a
> lossless compression option, and can handle multi pages.  I don't know if
> software to support all these features is (a) cheap or (b) available.
> 
> For my money, I'm sticking with TIFF

Agreed.