Hi Mert,
The first thought that I had was that 2x2 Binning of a CFA image effectively removes all of the colour information that would otherwise be contained in an equivalent 1x1 image. However, that thought depends a lot on how, or where, the binning takes place.
If the binning takes place 'in the camera software', then it is conceivable that the manufacturer can put in place an algorithm that creates a 2x2 RAW image from the original full-resolution array. Conceivable, yes - but far from simple, and verging on the impractical.
In fact, it could even be argued that there is absolutely nothing to be gained by performing 2x2 (or any other kind of) binning in an OSC camera (except when dowload speed is an isuue during the pre-acquisition 'image framing' stage - which is alsmost always performed on a grey-scale image anyway). Al you really gain is noise - the very thing that imagers do everything they can to eliminate in the first place! And the one crucial thing that you lose, straight away, is the pixel resolution that you presumably paid big bucks for when you selected that particular camera.
The question that you really have to ask (yourself) is, "Why do I feel that binning my OSC data is needed?".
Sure, you hear of other imagers working with binned data - but they are (almost) exclusively 'mono' imagers. They have no issues with acquiring their colour data at a lower resolution - once they have built their colour image, they will be quite aggresive in their noise-reduction techniques, at least as far as their chrominance data is concerned. And, as a result, they will lose quite a bit of 'detail' because of that. Where they score is in the fact that they acquire full-resolution Luminance (and perhaps even NB) data. When they then re-integrate the Luiminance data with their Chrominance data, the 'missing' detail is replaced from the information in the Lum channel.
Finally, those 'wierd' stars that you are seeing - I would imagine that they are caused by the very fact that you are trying to re-align 2x2 binned OSC data. Just remember that, typically, an OSC camera has an array of pixels that are effectively grouped into 'super-pixels' - each of which is a 2x2 array of something like a Red pixel, two Green pixels and a Blue pixel. And, therein lies the problem (when it comes to 'aligning' the binned colour data, as you are trying to do) - which of those pixels is the 'centre' of the 2x2 array?
You also have issues such as the fact that the Red and Blue pixels do not have 'direct neighbours' of the same colour, and that there are two Green pixels for every 2x2 super-pixel. How should software deal with those issues.
No - in my humble and personal opinion, if you are using an OSC camera, then download the RAW data in its native, 1x1 binned, format. You avoid all the issues that I have mentioneed and that you have encountered.
However, as always, if there are folks out there who reqularly work with binned OSC data, and have a processing sequence that works for them, hopefully they might see this thread and post their feelings and share their methods.
