2008.04.16 22:45 "[Tiff] Floating Point Tiff Format?", by Fred Weinhaus

2008.04.17 03:13 "Re: [Tiff] Floating Point Tiff Format?", by Bob Friesenhahn

I was wondering if libtiff supports floating point format tiff? I am interested in both floating point grayscale images and floating point RGB images?

Libtiff supports it by default but does nothing special for it.

Is there some documentation for the floating point format and if so where might I find that. I see nothing about floating point format in the TIFF 6.0 Specs.

A problem with floating point is that the ranges of values is not standard. Some applications normalize to 0.0 to 1.0 while others use ranges more suitable for their internal purposes. There are a few tags which can help but not enough that you can rely on them.

My questions relate to the use of a floating point or complex format for generating FFT of images using the FFTW package and storing the results in a relevant TIFF format.

As long as the range of the values is known, libtiff is very good at storing the data for you. Libtiff itself does not care about the range of values but your application will.

Another related TIFF format (also supported by libtiff) which can be used is Greg Larson's LogLuv format which stores floating point values more efficiently with the expectation that the values may span a broad range (i.e. HDR).

Bob Friesenhahn
bfriesen@simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer, http://www.GraphicsMagick.org/