Strange flat with ZWO ASI O III narrowband filter

It’s valuable to remember what flats do:

1) Correct for cos^4 radiometric irradiance fall off in the field
2) Correct for mechanical vignetting due to any part of the main light bundle being blocked by edges, dust, or dings in the optics.
3) Correct for PRNU (Photon Response Non-Uniform).

Notice that all of these factors are multiplicative to the signal.  Flats won’t correct for additive signals and in fact, additive signal (like stray light) will screw up flat correction. The last item, PRNU stands for Photo Response Non Uniform and it’s an effect that is most evident in the bright regions of the image.  It’s caused by the fact that the responsivity of the individual pixels across the sensor varies a little bit.  PRNU can mimic a random white noise-like variation across an image but it can also manifest as larger features that result from larger spatial variations across the sensor.  Features like graininess, curtains, bands, and patchy regions that show up in properly produced flat data  are evidence of larger, slowly varying spatial variations in PRNU across the sensor.  None of this stuff is necessarily evidence of “crappy manufacturing”; rather it’s evidence that the sensor manufacturing technology is up against a limit.   Each new fab process moves the technology forward and sensors continue to improve, but no sensor will EVER be perfect.  That’s a key reason WHY you do flats in the first place!  You are calibrating the sensor to produce images as if it had perfectly uniform response.  

The fact that PRNU may produce more obvious features when you use an O3 filter is simply the result of the fact that PRNU is actually wavelength dependent.  That means that the spectral response curves that manufacturers show for a sensor is merely an average response from all of the pixels in the sensor.  Each pixel may have a slightly different spectral response that varies with minor impurities distributed through the silicon in the boule that was used to produce the chip.  Again, this is an effect that flats correct.

The bottom line is that if you’ve taken flats correctly (meaning no stray light, proper exposure, and vignetting <~30%) and the image calibrates properly, the variations that you see in your flats are a good thing!  That’s the stuff that you want flats to remove and it shows that the process is doing the right thing.

John