›

Do i have a flat-problem?

Helge Kooistra
in PixInsight
Hi, after struggling with my data recently i wanted to post my dilemma here. 
I shoot with a mono camera (ASI1600mm). Attached are a few files, all from my green channel. The first -double- one contains the result of WBPP on the left, and the result after DBE and stretching to the right. But when i apply another autostretch (to the already stretched image) to see whats going on, i get this really funky smudged image (2nd image). To me it seems my flats didnt do their work properly. But when i look at my masterflat (3rd image) that looks quite nice to me. 

I know a strech upon a stretched image ain't the way to go, but i had quite some struggles in post processing to get to a decent image. A lot of background gradients and blobs to take care of, so wanted to see if i could find the cause. Also i had the idea my final result would end up better with such amount of integration time. (5th image)

stacked image is the result from 350x 60s frames, 5h50m in total
flats: taken with he asiair in 'auto flat' mode, 2.0s. I use a home made flat field box, consisting of a tracing led panel with a few sheets of paper on top to dim the light, and then two layers of opaque acrylic at different distances. all fitted into a foam-board box. flats are taken into the dark to avoid light leakage

Comments

  • BlockHeadBlockHead
    You can see from the flat that the paper is probably not a uniform scatterer. I do think the paper is an issue. 
    Maybe it isn't the paper... but the flat doesn't look right to me. In addition, I am not a fan of flat boxed fitted tot he front of a telescope. There is to many opportunities for scattered light issues.

    An experiment is to take a flat field image using the twilight sky. Don't worry if there are a few stars in there. Use approximately the same exposure time you are using for your flat box for a twilight.

    Divide the twilight image by your box flat. If they are not uniform (or very close)- you will be able to directly see how the box flat is introducing issues into your imagery.

    -the Blockhead 
  • Helge Kooistra
    Thanks, i will try that.

    ABout the flatbox; i tried to cope with reflection issues:
    see the attached schematic. 1 is the tracing panel, 2 are thick layers of paper, 3 is an opaque white acrylic glass, then a space of 12 cm and then number 4 another opaque acrylic glass... 5 is the telescope


    Naamloos.png
    2064 x 864 - 13K
  • BlockHeadBlockHead
    oh... you mean translucent material for 4 and 5... (not opaque!)
    So... in that case maybe the situation is better!
    But... compare with twilight flat for a sanity check.
    -the Blockhead
  • Helge Kooistra
    Eh yeah, i mean translucent. Sorry, english is not my native language 

    But i ran the tests you suggested and i noted some differences! I decided to make 'proper' flats the oldskool layers of t-shirt pointed at the morning sky-way, then i could use them right away if the test resulted in a difference as you predicted. So the exposuretimes are not equal but the gradient seems different.

    The first image is the result file; dividing the t-shirt flat with the panel-flat
    The second image are the two files compared, t-shirt flat on the left and panel flat on the right
    third image is the Blue filter flat, wich showed the most dramatic difference in my opinion. Off course exxagerated due to the exposure difference.

    Now trying to do the whole WBPP again on a seperate laptop because i need my main desktop PC for work. Let's hope it won't explode because my desktop PC seemd to have a lot of trouble with the WBPP process (due to the numer of files i think).

    Thanks for your advice!
    2022-09-06 10_13_56-PixInsight.png
    1190 x 908 - 1M
    2022-09-06 10_14_17-PixInsight.png
    2384 x 905 - 2M
    2022-09-06 10_25_00-PixInsight.png
    2430 x 1125 - 3M
  • Helge Kooistra
    edited September 2022
    And an extra result; because i started WBPP again, the masterflats are created with the new 't-shirt' flats. I considered that the masterfiles should be equal, so followed your advice again and devided the tshirt masters by the flat panel masters... With more schocking results in my opinion...

    The B, G and H showed massive difference. The R and L were exactly equal (only values of 1), So those flats are OK it seems. I'll do some more research, maybe the tracing panel emits different spectra which won't affect in case of a OSC cam, but with colour filters results in different wavelenghts being detected? If that's the case, the next question will be; wich light is the most comparable to the light you'll need for a flat file. The blue result shows that in the t-shirt method less blue light is detected than in the flat-panel way. 
    2022-09-06 10_42_12-PixInsight.png
    2378 x 900 - 2M
    2022-09-06 10_43_02-PixInsight.png
    2378 x 905 - 1M
  • BlockHeadBlockHead
    In pixel math, when you divide one image by the other- you might need to check the rescale option so that you don't end up clipping values that go above "1" (it depends on which image is the numerator).
    -the Blockhead
  • Helge Kooistra
    Thanks for your help. Unfortunately i'm still experiencing a big problem after trying to fix everything. I followed the advice in one of your tutorials and made new darks (because maybe the darks are too old), skipped Bias frames and instead made Flat-darks. After processing again it seems impossible to get an even background.

    In total i have about 27hours of integration time but i thinks this first preprocessing result puts me on the back foor heavily..

    As an example the Blue and H integrated result, combined with their masterflats. It seems the masterflat is overly aggressive. 
    The flats (and darkflats) for the Blue are 5.3 s and for the H 6.4 s
    2022-09-10 12_27_17-Window.png
    2383 x 904 - 2M
    2022-09-10 12_25_09-Window.png
    2385 x 904 - 2M
  • BlockHeadBlockHead
    Usually the problem with flats, if they are calibrated correctly- is an equipment/acquisition issue. Generating good flats is hard. Usually people that have corrector lenses or focal reducers are in trouble. Also in your case, the edge of the frame may have to much of a drop-off in light that a flat field cannot correct well. This means chip is too big (usable image circle is smaller than chip).

    The next thing to try is night sky flats. This *has* to work... and it is a nice way to make certain. Some people have filter wheel issues. But I doubt this is it in your case. I still thing it is scattered light... your flats are different than your lights I think.

    -the Blockhead
Sign In or Register to comment.