Image integration with non-Random noise

[georg.viehoever]

Hi,

so far I have been using DeepSkyStacker for stacking my images, but now I want to give PI ImageIntegration a try. I assume I need to process my raw images with darks, flats, bias frames and normalized channels as usual. This probably is still best done with DSS, using the "Create calibrated file for each light frame" option . However, the resulting images still contain non-random artifacts, such as

- my beloved  Canon Banding
- the sky gradient

How will these influece image integration? Should I remove those artefacts before integrating (DBE+PJSR script)? 

Georg

[Nocturnal]

Hi Georg,

I fix hot pixels and bad columns in my raw images before stacking with DSS. These operations make modifications to individual pixels without changing the rest of the data so it's safe to do before calibrating/stacking.

If your script works the same way then I recommend doing it as early as possible.

[georg.viehoever]

Sander,

I thought drastic artefacts such as hot pixels or bad columns (or satellite/airliner/cosmic ray trails) would be healed by stacking/integration as long as they do not always appear at exactly the same pixel coordinates in the registered/star aligned images (even your column defects will move slightly between exposures if your mount is as imperfect as mine  ).

In my opinion, the situation is different for the banding (and maybe also the sky gradient): they are only slight variations to the "true" value of the pixel (and thus are not filtered out my median combines of stacked images). Apparently, they also don't follow the usual noise statistics (Poisson or normal distribution).

I may very well be wrong about my this...

Georg

[Nocturnal]

Hi Georg,

yes this is true but I don't always use dithering so sometimes hot pixels line up for at least several frames. The drawback of dithering is that you can't monitor differential flexure anymore. Once I'm confident my new hyperstar based setup is sold I might start dithering again.