2000.03.30 19:12 "Re: Complex Floating Point", by Cris Luengo
I think what he means is rather than one image with one 128-bit complex sample per pixel, produce a single image with two 64-bit double-precision samples per pixel, treating one of these samples as the real part and one as the imaginary part. This should be possible to do in the TIFF format right now, without any sort of extensions.
Sorry, I forgot that. But I did because I also consider the case of images with more that 1 SamplesPerPixel, like RGB images.
If I save a complex RGB image using 6 SamplesPerPixel, I guess that the 3 first samples must be the real part, and the 3 last the imaginary. So if some one that understands floating point images, but do not for complex, can at least read the real part. Ok?
How about storing two images in the same TIFF file? One is the RGB real part, the other RGB imaginary.
Also I would recommend some other file format for storing stuff no-one is going to be able to read anyway. You are only fooling people by making a file that looks readable, but turns out not to be. (this is just an opinion)