Seestar S50 Pro vs Dwarflabs Draco

In our last AP-SIG meeting I was pondering how they do a 2×2 binning with a one-shot color (OSC) sensor. The answer is very simple, and has to do with the other observation I made about how binning in CMOS sensors is not the same (perhaps shouldn’t be called binning?) as in CCD sensors: Because the binning is done in software it doesn’t matter how the Bayer array is structured, as long as the software knows the structure. Specifically, I think the 50MP sensor has exactly the Bayer array you would expect, with a 2×2 area having 1 red, 1 blue, and 2 green photosites. This is almost the “obvious” way to deal with a OSC sensor: Each 2×2 area produces 1 RGB pixel. In reality, the processing is probably a bit more complicated than that, but this is a good first-order approximation.

The non-binned case is what is typically used in a OSC camera, although it is less obvious: Each 2×2 area produces a single RGB value as described above, which can then be used to define a luminance value for each of the 4 photosites. For example, when looking at the “red” photosite we can determine what the luminance value is by factoring in the transmission of the red filter for the overall (2×2 area) luminance. It will generally be VERY close to the overall luminance, but where there is a sharp edge (a transition in luminance) there can be a significant difference in the effective luminance for each photosite. Again, this is a simplification of what really happens in processing, but is a good approximation.