›

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Sign In

Categories

In this Discussion

HELP! Weirdness with Interactive LN, Multi-Night, Large Checkboarding on Mono

Zachary Patten
edited April 2024 in PixInsight
First let me say thank you to Adam; I am learning a lot as I make my way through your courses!  I recently decided to start using the interactive local normalization with WBPP after watching your video on it and have come across a strange issue.

Some background on my process; use an ASI 2600MM Pro and I shoot in LRGB with Ha for DSOs using an EdgeHD 8 with 0.7x reducer. I use NINA and take 1x R, 1x G, 1x B, 2x L, 1x Ha and then trigger a dither looping on that order for image acquisition.  This was a suggestion I picked up from a YouTuber a while back; that is instead of dithering after each frame, take a complete set of frames and then dither (which effectively gives a dither after each frame).  I then do a 2x drizzle in WBPP.  All in all this has worked very well for me over the past several years.  Outside of that, I'm using max quality pre-select in WBPP and PSF Scale SNR for the Lights subframe weighting and have left everything else at defaults settings-wise.

After watching your videos about WBPP keywords and multi-night/month/year stacking I decided to try to stack all the images I have of the Pinwheel galaxy over the past two years and have encountered something strange.

All together I have ~600 frames, I have manually reviewed all the frames and removed all frames which had poor tracking, high HFR, etc.  Guiding RMS on the used frames is 0.5 to 1.0 and no more than 1.5 on a few outliers (say 10 to 20 at most) and visually those even look OK; no elongated stars, etc.  The data is spread over 4 nights, 3 from last year and one from this year.  None of the images exhibit any visual evidence that would cause this; that is they all look perfectly fine and normal.

Now it appears for two of the nights last year the images are rotated.  This was before I got my rotator and could easily ensure subsequent imaging sessions would match previous sessions rotation and that set of images sort of 'fell through the cracks' in that regard.  That being said, I am fine with having to crop down to just have the galaxy itself.

Back to the issue at hand, when loading into the interactive LN I was presented with a large checkboard effect on many of the images.  I had previously processes two of the four nights individually and had not seen this in the final images so I'm fairly confident this has to do with using all of these nights together.  My initial thought was that this was an issue with my darks/flats thou there are no obvious visual indications of issues with them, I started to remove those from WBPP in an effort to exclude that however the checkboard is still showing up however I still have some flats/darks being used.  Currently I am doing a WBPP run with only darks to see what happens.

At this point I am starting to wonder if this might be due to the rotation thou I'm not sure this is correct either.  This appears across all LRGB channels (though the attached screenshots are only for the Blue channel) in ILN and I have noticed that none of the rotated images seem to have this checkboard.  Also the images with the checkerboarding show a considerable degredation to their image quality compared to their raw state which I assume is a result of whatever is introducing the checkboarding.

Given how big the data set is this troubleshooting is becoming time consuming as even with my 36 CPU machine it still takes 2-3 hours to get to the interactive LN stage and excludes me from being able to do anything else on my computer during this time.  

This feels like it's something to do with the bias/flats/darks but I just don't know.

So with this I wanted to see what the more experienced of you thought about this as I'm definitely lost about what is introducing this weirdness and the troubleshooting process is starting to become frustrating.  I suspect my lack of experience is causing me to miss something which is more than likely obvious to those with more knowledge here.

Any and all feedback, comments, suggestions are welcome!

Comments

  • BlockHeadBlockHead
    Yes, this is indicating an oversubtraction of dark frames (or biases). You get the same effect typically when using narrowband images and a pedestal is added to the calibration process to avoid negative values. 
    My video here demonstrates this *exact* effect right down to catching it when looking at LN:
    (see the image on the page)

    In your case, your issue is no a pedestal issue but instead that you are using a dark that is not correct for the data that is being calibrated. The hotpixel population or overall dark level is simply too high. You should be able to numerically see this by looking at the values of the dark frame compared to the bias or your l ight frame data.

    The weird moiré pattern will only show up on images that have small interpolation correlations after registration. The rotated images will destroy the delicate moiré pattern. So this makes sense as well.

    Bottom line... bad darks.

    -the Blockhead  
Sign In or Register to comment.