1998.07.30 18:35 "RE: 16 bit grayscale", by Ed Grissom

The library already does it. It has for years now.

Yes, exactly right! we have our own (non-LibTIFF) Raster I/O library, and have been using LibTIFF to verify the correctness of our implementation. Recently I upgraded our code to do 16-bit stuff and was pleasantly surprised to find that LibTIFF could handle 16-bit.

However, I found some problems when using anything but uncompressed - full 16-bit data. 12-bit data is common in satellite and medical imaging, and there are problems in either the library or some of the distributed tools when handling such data - especially with compressed storage (mainly dealing with "packing" of unused bits as described in the TIFF spec).

I sure wish Adobe would either update the dang spec or give it to someone who will.

AGREED!!! I have a list of features that we desperately need in TIFF, but some of them are fairly major changes that we don't want to go off on our own and do. We would much rather have an agreed upon standard to follow. These include:

• ability to store large images (greater than 4 GB),
• photometric interpretation of multi-band images (7-band landsat for instance)
• sparse images ( tiles with "NO DATA" in them)
• ability to specify a transparent pixel value (not a mask)

Why don't you just give the TIFF spec to a capable committee of people

I may not have been very vocal on this list recently, but I participated in both the 6.0 TAC and the TTN2 discussions and would relish the opportunity to be on such a committee. Please feel free to include my name if such a chance arises.

--
ed grissom
egrissom@ingr.com