Hi there,
When you say "RGBG is not common enough" what, exactly, do you mean by this?
In a standard four-colour Bayer Array (also commonly referred to as a CFA) there are four possible meathids of aligning the CFA to the underlying monochrome pixel array. It is very common to not know how the CFA array has been aligned by the CCD manufacturer, and even if a datasheet is available, there is also then the fact that the actual manufacturer of the imager (of the camera) may have used some undefined offset, image-flip, image-flop or rotation (or a combination of all four transformations). Further, the raw image can also be changed by the image capture software. And, finally, some software packages (like PixInsight) actually offer the user a means of trying to compensate for these transformations at the time the raw image is loaded into PixInsight for processing.
So, all of that sounds scary, doesn't it? After all, if a user cannot know how the CFA is supposed to align to the pixel array, then how can they ever expect to be able to deBayer an image to recover the colour content?
Fortunately, when the CFA sub-matrix (the R, G, B and G elements that you referred to) is in the form of a 2 x 2 grid, you only have four possible choices to consider - and all of these are on offer within the DeBayer Process (and associated BatchDeBayer Script). Only one of these will work 'correctly', and all you have to do is to try all four and determine, empirically, which one is the 'correct' one. Once you have done that then, unless you change imaher or acquisition software, you should not have to change things again (and PixInsight can even memorise a 'Default' selection based on your particular setup).
And the process is not actually that difficult at all - all you need to do is to open a single RAW (un-calibrated) sub-frame. Then, using the DeBayer Process, apply each of the four possible CFA settings to your source image - one at a time. You will, of course, generate four new, colour, images. You can apply a ScreenTransferFunction (STF) to each of these images, and you can zoom in to examine what the images look like.
Typically, I find that two out of the four unages are noticeably 'mosier' than the others - these can be discarded straight away.
Then I would try 'un-linking' the channels in the STF dialogue, and re-applying the STF to the remaining two images. One is usually 'better' than the other in terms of what you might expect your un=processed colour image to look like. Set that method as your CFA selection (and default) and try your new selection on your image set.
Do noy overly concern yourself if your fully calibrated, aligned and stacked 'MasterFrame' doesn't seem to have much colour in it as you start post-processing - this is normal, and subsequent steps will slowly reveal the colours that you have captured.
Another very useful method for empirically determining CFA is to simply put a piece of coloured paper (or see-through film, such as a wrapper from a certain well-known brand of boxed chocolates !!) over you imager (no need to be attached to the OTA for this) and expose to a reasonably 'white' source of light. Now, when applying each of the CFA selections, only one should result in an image approximating the colour of your filter.
Hope this helps but, if I have missed the point of your query entirely, please come back to us with a more detailed question.
