Exclude outlier pixel values when creating master dark frame

[lowenthalm]

I saw this new feature in the release last week. Very cool. I had just run scripts a few weeks earlier to capture an all new -10C and -15C dark library for my 2024 observing season and found only two out of 64 were bad. Oddly enough, they both occurred in the longest bin 2 (1 and 2 minute subs) that would take an hour and 2 hours to recollect, respectively. All of my other darks for bin 1 and bin 2 for 2, 3, 4, 5, 6, 10 and 30 second subs (with some of these at a couple of different gains) were fine.

I recaptured one of the bad darks last night, but still saw some banding (which I though was excessive enough to possibly reject) that I thought would have been suppressed by this outlier rejection feature.

Question: After the first test of the new feature, it occurred to me that I'm running all my dark captures using SharpCap sequences. I noted that checking the reject outlier box in the dark capture dialog and cancelling the dark collection after a couple of frames would get the reject outlier option set to on the next time I opened the capture dark dialog. However, I don't know if the current state of this checkbox in the capture dark dialog is honored by the capture dark sequence command. Do I only have access to this new feature when I manually run dark collection via the dialog?

Is there any reason to have this function available in the capturing flat frames function as well? They can also suffer from occasional bad frames.

[admin]

Hi,

the designed use case for that new feature is that one or more of the individual dark frames contain hot/warm pixels that are absent from the other frames. Including these pixels in the calculation of the master dark leads to a 'lukewarm' pixel in the master dark that can punch black holes in the dark subtracted image since the pixel is not usually hot in the light frames. Because of this, the code is looking for pixel values that fall more than 10 standard deviations away from the mean of the (non-hot) values at the same location. The large sigma threshold of 10 SD means that this will have no effect on banding in the images, which is usually down near to the 1ADU level.

Essentially the new feature means that master dark frames can be immune to the effects of intermittent hot pixels. Coming up in the next release, you will be able to enable hot/cold pixel removal *and* dark subtraction, meaning that intermittent hot pixels in the light frames can be handled properly.

The setting for using this feature is remembered automatically from one use of the capture dark tool to the next, so if you turn it on in the UI, it will be used in sequence dark frame capture steps. There is, however no way to turn it on/off via the sequencer at this point.

cheers,

Robin

[lowenthalm]

This is a good idea to suppress cosmic ray hits and such. I had a couple of cases before this feature was added where i think a single frame just had complete data corruption for a 10th of the frame. Would this be enough to causes the data in the frame to be suppressed?

[admin]

Hi,

yes, data corruption would probably get excluded too, as the random corrupted values will be a long way from the normal mean value for those particular pixels.

cheers,

Robin