When I tried it before, the link to Astrobin yielded an Ice In Space "Something terrible happened" error, but now it is working.
The image looks very good.
Quote:
Originally Posted by strongmanmike
....
Yes dithering more may help I guess..? Or in Astroart6, that I use, there is a column remover available in preprocessing, you just identify the column and it will be removed in every sub as part of calibration.
Mike
|
We do both those things. Firstly, GoodLook 64's "Portobello" calibration module identifies bad columns in the darks, where they are most obvious and easiest to find statistically, and then replaces those bad columns with the mean of their neighbours. Works pretty well. (We don't have any actually totally dead columns, but we have about 35 of them out of 4000 that are a bit dodgy). Then, assuming we have at least three (and preferably more) dithered subs, "Prometheus", the registration module, produces a "winsorized" sample at each registered pixel, replacing all outliers with the most extreme value that would not count as an outlier. That is mainly to get rid of cosmic rays, hot pixels, satellite trails, and ghost afterimages, but it would also get rid of bad columns even if we hadn't already handled them. As strongmanmike said, standard packages will do something similar to both these steps. The two approaches make very different assumptions and are complementary.
