2008.09.08 10:07 "Re: [Tiff] SAMPLEFORMAT_IEEEFP", by John

2008/9/8 Bob Friesenhahn <bfriesen@simple.dallas.tx.us>:

But I have no luck with floating point values. I use SAMPLEFORMAT_IEEEFP sample format along with 32 bits per sample.

All the pixel values are issued from an 8 bit/sample image, with no conversion or multiplication applied to them.

When I open the file with some well known image editor, the image is completely saturated, all white with values 32767...

Where should I look?

Most applications don't support floating point images like this. My application (GraphicsMagick) does, but in a somewhat crude way. Some applications like GDAL are designed for floating point and should do a lot better.

My scientific image processing application, vips, also supports floating point tiff.

  http://www.vips.ecs.soton.ac.uk

It might be useful to testing. There's a windows binary and packages for most unix variants.

John