Hi John!
It think it could be possible to get the extra planes without Standard Debayering.
However, you will never get better data than with the debayer und channel extract methode, because
the algorythm for the debayering is sophisticated.
What would i do if i could see a win.
I would take an image of plain green paper, one seperately from a plain red paper and one from a plain
blue paper.
You load these 3 images into PI as B/W-raws and you will see in the red B/W at the red position a bright pixel
and on the other positions a more or less black pixel.
This is valid also for the BLUE B/W picture it will have one bright pixel per 4-matrix and 3 dark pixels.
On the green B/W image there will be 2 bright and 2 dark pixel in each matrix of 4.
There are different tools to put the bright pixels in each image to the value 1 and the darker ones to the value 0.
(RangeSelection, PixelMath etc.) - iif($T >0.5,1,0)
This gets you for each color a separate mask where for example on the blue mask there are only the blue pixel
POSITIONS on 1 and the others on Zero in each matrix of four.
So you will end up with 3 images with the same size than the original. one for red, one for blue and one for green.
I call them Masks.
Finally with Pixelmath you take the RAW astro image Image $T with an iif expression
you ask simplified iif(RedMask = 1, put the sáme pixel of thé raw image to the 3 neighbouring pixels in the Raw Image $T)
(xpos+1,ypos,
xpos,ypos,
xpos,ypos+1) .
This will fill the 3 neighbouring pixel of the red with the same red value.
Create with the result a new image as RED
Load the unmodified RAW again and do the same using the BLUE mask.
Here the neighbouring pixels in matrix of four are in the vice versa position.
For green you have to fíll, after you have saved before the blue raw image,
only 2 pixel (averaged) in the matrix. Using the the Green Mask as logic.
You should receive gray images the same size as the original RAW for each color channel.
Now you can do your linear processing of the channels seperately, than combine the channels with RGB channelcombination
after doing a Linear Fit to each channel. Would not recommend any stretching of these RAWs.
However, your image will be always less sharp because you more or less trashed one important
green pixel, and you have Super Pixels of 4 for each channel seperately. Means less resolution generally.
Hope this idea helps. Would be interested what you really want to achieve?
Aloha
Gerald
