Color separation notes
======================

As an example I took the raw Nikon image file DSC_0100.NEF and processed it with
Cinepaint to read all 16 bits in each color.  I stored the resultant image as
DSC_0100.TIF which is a 16-bit format that ImageJ can read.

Note that ImageJ can also read the NEF format using the LOCI plugin but it
converts it to a scaled grayscale and loses bit depth.

With ImageJ I read the DSC_0100.TIF and saved it as DSC_0100.tif in ImageJ tif
formatting for future use.  When read in with ImageJ, the color tif file will
display as a stack.

I used the Stack->Images tool to separate the color planes, and converted each
one to a 32-bit floating point image.  These are stored individually as
dsc_0100-red.tif, dsc_0100-green.tif, and dsc_0100-blue.tif.

I used the Image math tool to divide the blue image (it has larger values) by
the red image.  Since the result is of the order of 1.0 over the lunar disk,
I multiplied it by 10,000 and stored that as dsc_0100_blue_div_red.tif, 
and in fits format as dsc_100_blue_div_red.fits.

The divided image is difficult to display in ImageJ because the windowing tools do
not allow easy adjustment of the windowing when the differences between the upper
and lower bounds are small.  

However, by loading the image either in ds9 or aladin you can adjust the
grayscale and view features.

 


Telescope notes
===============

Night of July 23, 2010
CDK125N
Prime focus
Nikon D200
ISO 200
Gibbous moon

moon_montage_20100723.jpg
  Montage from jpg's without change in brightness or contrast at 1:1 pixel map

moon_montage_20100723_2048_hc.jpg
  Rotation and scaling from moon_montage_20100723.jpg
  Adjustment of contrast to enhance disk 
  
moon_montage_20100723_1024_hc.jpg
  Scaling
  
The hc images started with the 4 original jpgs, montaged onto a 4094x4094 black
background:

gaussian smooth 1 pixel
unsharp masking with gimp defaults of gaussian 5 pixel and 0.5 masking parameter
repeat usm

scale to 2048 and 1024

The next day I started with the montage master and added 2 more pixels to the
perimeter to make 4096x4096, so that scaling preserves the pixel scale ratios of
1:1, 2:1, and 4:1 from the CDK125 focal plane.  Since the color shifted slightly
with high contrast, I reworked the USM with a parameter of 1.0 instead of 0.5,
and without the pre-blur 1 pixel gaussian.  Consequently the usm images are not
color shifted from the original, have a less contrast than the hc images, but
noticably enhance rays and crater edge detail.

The 2048 and 1024 versions are scaled from the 4096 usm.

  
  
Images taken in quick succession, and last one at 

2010:07:24 04:57:53

according to the NEF header.  Camera clock is within 10 seconds of correct UT.