›

WBPP master dark and flats over subtraction.

Bill Tinsley
in PixInsight

Hi Adam,

So far your tutorials have allowed me
to resolve any problems that I have run using PI, until now. I enjoy watching
your videos of troubleshooting people’s PI projects problems. That’s why I am
reaching out with a problem that I found a solution but I can not understand
why. I have switched back to OSC because my QHY268M needs a repair, that’s
separate history.

I have never binned an osc before and tried to give a go. I
don’t think I will continue binning 2x2 since the noise went up as well, and
leaving in a bortle 6 is not helping with light pollution that I have to deal
with. To the problem, I collected a few hrs of M1 and tried to integrate them
using WBPP. The integration failed and looking at the individual calibrated
frames it looks like the WBPP is over subtracting the dark frame producing an
almost black light frame. The dark frames are recently acquired and are the
same exp time and binning as the light frames. After many tries with different
settings I got it to work, but I don’t understand why it is working.

The WBPP settings that work are all settings as per default except:

-Removing the dark-flats or bias from the WBPP window tub.

-Selecting optimize master dark.

I don’t understand this processing setting's success since
there are no dark-flats or bias masters or frames to optimize the master dark. The
final master looks like the flat is over subtracted as well. Maybe it is a common
problem on both flats and darks subtraction that is more accentuated in the
dark. Either way, I am at a loss with this and I am hoping that you will be able
to shade some light on this. I have attached some pics of the process for your
viewing and if you would like some of the frames to experiment, I will be happy
to provide you with a Google drive link for you to download as necessary.


Regards,


Bill

Comments

  • Darren Jehan
    Difficult to see what's going on Bill - especially with only 4 light frames as well....
    How have you created the master calibration frames? Were they done separately in WBPP?

    If you wanted to share the raw, uncalibrated data, I'd be happy to have a look?


    Rgds
    Darren
  • BlockHeadBlockHead
    Binning some CMOS sensors simply does not work. It depends on how on-chip binning works- and it isn't a given it works like the old CCD sensors. 

    For example Jon Talbot's description of how the ZWO6200 works.



    -the Blockhead
  • Bill Tinsley
    edited January 2022
    Hi Darren,

    Thank you for your replay, After reading my post, I can understand your observation on the difficultty understading, please forgive my english as second language as well. I'll try to answer your questions.

    -The master were created using WBPP and the individual processes tools and the result were identical no matter what method was used.
    - I only used 4 frames to speed up the testing since the 4 or 20 frames did not have any impact on the end result.
    -The first attached pic-- the black frame with 0.0 K values is 1 calibrated frame produced by WBPP full failed integration processes. However I tried image calibration alone and WBPP calibration only with the same results, both requiring the same strange settings. 

    -The second pic was calibrated using WBPP using a master flat and no master dark-flat loaded and selecting optimize master dark. 

    -The third pic is the 4 frames stacked using WBPP using the same settings as per the successful calibration..

    I have included a Google drive link with sets of light, flat-darks and dark frames if you would like to try yourself.

    Google drive link


    Thanks,

    Bill


  • Bill Tinsley
    Hi Adam,

    I watched the Jon Talbot's description of how the ZWO6200 works. Perhaps, the cmos chips binning method, of software pixel average could be responsible for what is happening. I also learned from the video that the reason that I tried binnig 2x2 was correct. My setup combination Vixen VC200L at 1800nmm focal length with the QHY168C 4.8 um pixels is oversample and binning can help with that. It looks like I have to give up on that idea.

    Bill
Sign In or Register to comment.