2007.03.27 15:03 "[Tiff] support for 24 bit floats?", by William B Thompson

2007.03.28 12:15 "Re: [Tiff] support for 24 bit floats?", by Andrey Kiselev

On Wed, Mar 28, 2007 at 11:06:25AM +0200, Kai-Uwe Behrmann wrote:

I should clarify myself here. I messed up libtiff with GDAL. Libtiff, of course, has nothing to do with actual data format, it will be returned as is, without any conversions (i.e., in form of 24 bit float array).

Can a similiar conversion occure in libtiff?

If I remember correctly the LogLUV pixels are converted to 32-bit likewise. The same could apply to IEEE Half.

This would increase compatibility. Blender, CinePaint and possibly more applications will soon use at least the Half float variant much more.

Well, it can be implemented through the pseudo-tags technique. But does it really needed? I think that it is up to application to do data decoding, just like GDAL does.

Best regards,


Andrey V. Kiselev
ICQ# 26871517