[ptx] 16 bit and float image support in nona/hugin

[Kai-Uwe Behrmann]

Am 02.05.04, 01:11 +0200 schrieb Pablo d'Angelo:

> I was wondering how an alpha channel look like in these format, and if

alpha:
unsigned values - MIN -> Max  (0 .... 65535)
signed values   - 0   -> MAX  (0 .... 32767)
float/double    - 0.0 -> 1.0

> somebody uses signed 16/32 bit images etc.

modern cameras, scanners, artists, scientists, hobbyists ;)

> Are there other popular formats for HDR images that should be
> supported?

Greg Wards old HDR format from radiance. And some more on
<http://positron.cs.berkeley.edu/gwlarson/pixformat/index.html>

Sample implementations for:
RGBE (hdr)
OpenEXR
TIFF (int8,int16,float,double,SGILogLuv)
cineon
in cinepaint <cinepaint.sf.net>, sources available

Basically RGBE/hdr and
tiff (because You have allredy some code) are good starting points.

Here comes the tiffinfo output on an newer libtiff test suite image:
TIFF Directory at offset 0x483b4
  Image Width: 279 Image Length: 220
  Bits/Sample: 64
  Sample Format: IEEE floating point
  Compression Scheme: Deflate
  Photometric Interpretation: RGB color
  Samples/Pixel: 3
  Rows/Strip: 3
  Planar Configuration: separate image planes

All tags are definately needed, even so Compression Scheme and alpha can
be adapted to ones needs.

The tiff stuff You should have allready, except of double reading
and SGILogLuv, which is available on the patches site
<http://sourceforge.net/tracker/index.php?func=detail&aid=834911&group_id=75029&atid=542708>.

Hope this helps
Kai-Uwe