Author Topic: Background Extraction difficulties  (Read 1359 times)

Offline rtemen

  • Member
  • *
  • Posts: 56
    • View Profile
Background Extraction difficulties
« on: 2017 March 27 20:03:15 »
I have finished stacking all of my files, and when I try to use either DBE or ABE, I cannot seem to clean up the background at all.

Here is the stacked file.
https://www.dropbox.com/s/ik592a86bgwo24c/M101_Integrated.xisf?dl=0

Can anyone look at this file and tell me what might be wrong with the file, or what the proper procedure would be to have more success?

Thanks,
Rich

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4525
    • View Profile
Re: Background Extraction difficulties
« Reply #1 on: 2017 March 27 20:52:22 »
i don't think there's anything wrong with the file at all... however it does show very weak SNR in the R and B channels, which is to be expected if you're using a DSLR. also there seem to be residual hot pixels in the subs as there is the diagonal streak "rain" noise that's typical of hot pixels + differential flexure while guiding. dithering should help you reject these hot pixels more easily during integration. alternately, cleaning up the calibrated subs with CosmeticCorrection may be helpful.

i've increased the tolerance to 2, generated a grid of DBE samples, and moved any samples that fell on stars away, then ran DBE with subtraction mode.

rob

Offline msmythers

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1176
    • View Profile
    • astrobin
Re: Background Extraction difficulties
« Reply #2 on: 2017 March 27 21:48:41 »
Here I used ABE. Using the Draw sample boxes and Just try samples you can fine tune ABE quickly. The lower left image is the default settings sample boxes. You can see how many sample points would not be used or used poorly. By tweaking the Deviation you can see the difference in the samples and the corrected image. Not perfect and ABE can't replicate DBE but it can do a decent job many times.


Mike

Offline rtemen

  • Member
  • *
  • Posts: 56
    • View Profile
Re: Background Extraction difficulties
« Reply #3 on: 2017 March 28 06:59:57 »
Hi, Rob. Thanks for the response.
I am new to PI, so I was wondering if you could educate me a little.
I thought that a hot pixel is actually a pixel on my camera sensor that is 'stuck' in a white mode, so we need to have the program not think it is a star.
How would that account for that diagonal streaky 'rain' pattern that is evenly across the whole picture?
You can point me to a tutorial that describes this as well if that is easier.

Thanks,
Rich

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4525
    • View Profile
Re: Background Extraction difficulties
« Reply #4 on: 2017 March 28 13:20:05 »
sure. first off there is a zip file there with the DBE process, it looks like you didn't download it.

what happens is that hot pixels are indeed stuck, usually at 1. but when you calibrate, if you scale the darks, the hot pixels in the darks get scaled down with the rest of the pixels, and then when the scaled dark is subtracted, the hot pixel is not fully subtracted. maybe it's value is 0.2 or 0.3 in the calibrated image, something like that. and then if these pixels are not rejected during integration, you end up with those pixels in the final integration.

the reason you see the diagonal streaks is that when you guide with an external guide camera, any flexure between the main camera's optical train and the guide camera's optical train shows up as a steady drift from one subexposure to another. if you look at  StarAlignment's output you'll see the dx and dy values steadily increasing (or decreasing) with each successive subexposure. the guider is doing it's job keeping the guide star centered on the guide camera, but the main camera is drooping.

so first you should try to eliminate this flexure. it will limit your subexposure time - eventually it shows up as oval stars. this usually means making sure the guide camera is well-affixed to the guide scope, and making sure the guide scope is well attached to the OTA. if the OTA is in rings and the guidescope is on the rings, that can be the problem right there. the whole OTA can move independently of the guider. it's best if the guider is attached to the OTA itself. next is trying to shore up the main camera attachment; perhaps the focuser droops a little bit.

second you can try using CosmeticCorrection on the calibrated lights to try to get rid of those 'warm' pixels before registration and integration. or you can try to be more aggressive with the pixel rejection in ImageIntegration - if the pixels values are still outlying enough,  II can detect that and reject them.


rob

Offline rtemen

  • Member
  • *
  • Posts: 56
    • View Profile
Re: Background Extraction difficulties
« Reply #5 on: 2017 April 16 10:00:07 »
Hi, Rob.  Sorry for the delay since our last communication.
Some questions if I may.
You indicated that there is a zip file 'there' that I didn't download.
Where is this zip file and what is it for?

You described how the pictures can drift a bit as each picture is taken.
I am pretty sure that with the same equipment and procedures, I sometimes can get a collection of shots with no drift, and sometimes they will drift from pic to pic.
What do you think would be the cause of this?  Perhaps I need to better adjust something.

Thanks,
Rich

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4525
    • View Profile
Re: Background Extraction difficulties
« Reply #6 on: 2017 April 16 16:27:20 »
hi rich, it is easy to miss but above the image i attached in my response, there is a .zip file that contains a DBE process icon that you can load into PI so that you can see how i did the background extraction.

look up "differential flexure" on the web - the problem is likely that your guide camera and main camera are not rigidly 'attached' to one another, and so because one or both cameras are shifting independently from one another as the telescope tracks, although the guiding might be solid, you still see trailing in your images.

rob