Problems with flats

[turfpit]

I got this message from Malwarebytes when trying to access the link:

Capture.JPG (21.46 KiB) Viewed 3319 times

Dave

[MikeB838]

Hi Dave,

I scanned the zip-file with Avast and it is okay.

May be your guard blocked the file because it is a zip archive and/or because my home page can not be accessed by https, but it should be safe to allow the access.

Michael

[MikeB838]

Hi Menno,

good to hear, that you can reproduce the effect and I seem not to have made a stupid mistake

Michael

[turfpit]

Mike

The lack of https access will have been the problem.

Dave

[admin]

Hi Michael,

thanks for sharing the files and details of how you captured the flats.

I selected "None" as Black Level offset correction when capturing the flats.

This is almost certainly the cause of the issues - the 'None' option is OK for when you are trying to get rid of dust donuts on lunar/solar/planetary images (even then, it's not perfect, but it will get 90 or 95% correction or so, which is good enough in those cases). On deep sky imaging (where you stretch the final image) whatever small percentage goes uncorrected is boosted right back up to visibility when stretching the image. You need to either use the 'Subtract bias' or 'Subtract dark flat' options for deep sky imaging.

To give an explanation as to why this is important, it all starts from the fact that the camera won't produce a pixel value of zero for a pixel where no light landed - it produces a base value bigger than zero due to the offset control (on some cameras called black level or brightness). Maybe if your camera is in a 16 bit mode with maximum pixel value of 65535, the value of a totally dark pixel might be 200 or 500 or something. Let's take 500...

Now, if a spot of dust cuts out 5% of the light, in a flat frame it would reduce the pixel value by 5% - say from 30000 to 28500 - if it wasn't for that offset level of 500. Actually, the pixel value will change from 30500 to 29000. If SharpCap knows about the offset of 500 (because you chose Bias or Dark Flat) then everything is fine, but if SharpCap doesn't know about it, the correction made by the flat will be based on 30500/29000 or 5.17%, not the required 5% and the flat frame will tend to over-correct.

cheers,

Robin

[MikeB838]

Hi Robin,

thanks for your explanation and surely I will give it a try.

But should not the view be the same (in Live Stack and in Live Stack with folder monitor)?
I used the same settings in folder monitor and no matter how I change the histogram sliders in folder monitor, the donuts are always visible.

At the telescope all looked well and in postprocessing with folder monitor (more time to check the best settings) no good setting can be found.

BTW: I have shortly checked the black level/offset value again. Now I have a value of 20. But with the old value of 60 I also had the donuts.

May be you can explain the different behaviour in folder monitor, thanks a lot!

cheers,
Michael

[admin]

Hi Michael,

yes, the result of the processing (if given the same inputs) should be the same either way. However, it may not look the same on screen due to differences in either

* Live Stacking display stretch levels (how strong the stretch is)
* The display stretch in the mini histogram (if that is active).

Unless those are both set the same then comparison of what you see on screen is a bit dubious.

cheers,

Robin

[MikeB838]

Hi Robin,

I understand that, but I had the mini histogram off (live and at folder monitor) and I set the stretch levels in the folder monitor case to almost
the same values as during the live session (you can see that in both setting files in data.zip).

And the donuts in the folder monitor are seen at a very wide range of the stretch levels (even far away from the levels I used to set at the scope).

So please consider to check again the code for folder monitor. That would be great, thank you very much!

cheers,

Michael

[admin]

Hi Michael,

interesting - if I use the raw frames with the 'Test Camera 1' (you can load any individual raw frame into that and it just repeats over and over), it all works fine. With the folder monitor camera there is something odd as the flat doesn't seem to be doing its job properly. I would have expected both to be the same, so will have to dig into it further. Thanks for persisting in pointing this out - without checking I would have been 100% convinced that the working behaviour in Test Camera 1 would carry over to the folder monitor camera.

cheers,

Robin

[MikeB838]

Hi Robin,

nothing to thank for, the donuts are a showstopper for me and if you can find something it is a win - win

cheers and good luck and thanks a lot for your great support!

Michael