Re: [CinePaint] masks; detail 8 vs. 16+ bits

On Thu, 13 Mar 2008 01:27:30 +0200
> 
> I wonder why?   Now that the RAW image is up at
> http://www.colorist.org/images/DSC_1180.NEF  do you want to see if you  
> get different results from The Gimp?   Perhaps I've somehow got a  
> funky The Gimp.  thanks for testing with a conversion of the  
> lightroom.tiff to 8-bit.  But the question is still to be resolved.
> 

After downloading the NEF file, I converted to PNM format using
dcraw:

dcraw -v -w -c -4 -q 3 -o 0 DSC_1180.NEF > DSC_1180.pnm

This command does nothing except demosaic the RAW file to a
16-bit image.  There is no other processing or modification.

Then I reduced the depth and copied the 16-bit file to 8-bits
using netpbm:

pamdepth 255 DSC_1180.pnm > DSC_1180-8.pnm

Loading the 16-bit version into cinepaint and the 8-bit
version into gimp, followed by adjusting the gamma on
each image to 2.2, produced *identical* images.

The difference you see in your gimp version is hard
to explain without knowing the *exact* manner of processing.
Unfortunately I cannot load the NEF file directly into
gimp because I don't have the necessary plug-ins (I choose
to do all conversions manually as above.)

Possibly the stored NEF tags were used.  Here are some tags
excerpted from the NEF file using exiftool:

Gain Control                    : Low gain up
Contrast                        : Low
Saturation                      : Normal
Sharpness                       : Normal

The gain and contrast differ from their "null" values
and could have been somehow exploited.  But that's just
a quick guess. I don't know anything about how gimp loads
and processes a RAW file.

Fortunately, Linux allows complete control of processing
through these simple command line tools such as dcraw and
netpbm.  There are no "black box" modules or plug-ins that
perform some obscure operation -- unless that's what you
desire.

But the original question is resolved.  In this case
there has been no effect of bit depth on the rendering
of fine details and this is what is expected.

FP


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/