Back to forums

RCC - Wizard in Narrowband

Blackmagic Design DaVinci Resolve Request for feedback Sh2-142 Sharpless Catalog NGC Objects
7 replies•402 views
dk101 avatar
dk101 avatar
I would really appreciate some views/comments on my take on Sh2-142 / NGC-7380

https://www.astrobin.com/v858rx/

I took this from London so narrowband was a must and, yes, there was quite a bit of moonlight about but with so few cludless nights that we have been having there was very little choice.

I processed images in PI and I didn't do anything fancy. Colours are close to pure SHO and then i tweaked them based on vectorscope output (DaVinci Resolve) since i am colourblind in red-green. The workflow was: integrate (manual process, not batch), DynamicBackgroundExtraction, linear fit, tiny bit of deconvolution and denoise, SHO mix, histogram & saturation based on vectorscope. Histogram & saturation were done in PI and the workflow of tweak-export-Resolve-load-view was a pain - I could do with a vectorscope in PI

This is my first nebula so please let me know what I could I do to make this image (and future ones) better.

Thank you in advance!

Daniel
Helpful Engaging
andrea tasselli avatar
andrea tasselli avatar
Well, since you asked…

I think you have an issue with the flat fielding of the image, darker areas around the main body where there should be none. Also, lighter corners where it should be rather flat/dim. Too much green in the main body if you ask me so very little contrast between OIII and Ha/SII. I appreciate this is going to be rather difficult for a colour blind person but there you are. Possibly as bi-colour version (OIII = G&B, SII+Ha=R) would be easier?

Also, Even with SHO palette is better to recreate the stars using the bi-colour palette mentioned above and replace in the star-less image. Just use the standard photometric calibration in PI and you shan't be worried with the import-export kerfuffle.
Helpful Insightful
dk101 avatar
dk101 avatar
Thanks Andrea! That is super useful!

I did see too much green (well, didn't see it but..) and i wondered if i needed more OIII/SII signal as when i look at the amount of photons in each band for equialised backgrounds, the OIII signal is way down. SII too but that is to be expected. Looking at other people's images of the Wizard i see a lot of blue and I just couldn't get much blue through without overwhelming the background. So, could the signal level be a problem or do you think it is just the mixing/histogram?

Interesting that you mention flat fielding. The images, pre-processed/corrected look fine. Only when they are integrated i get these, erm, "blooms" in places. I also haven't seen these on other images i've been doing and the flats collection etc is always the same - i use NINA flat wizard and an LED panel with ND filters in front of it. For the Wizard I took 30 flats for each filter per session and in each session max 12 images per filter were taken. For darks, i have a library i created and they do not have any grandients on them. I sometimes use bias and sometimes not and tbh i didn't see much difference. i guess 2600MM-Pro is a decent camera and bias is not really important there. Manual integration or with the batch process gets me the same "bloom" in the corners but, like i said, looking at other target images i've been playing with, i don't see these.  I wonder where they are coming from?

If you (or anyone else) has any ideas why this may be happening I would really appreaciate it. Esp if my flat-fielding is wrong as, unless that is right, the outputs will be, well, terrible.

Great idea with the photometric calibration! I will try that and i will have a play with bi-colour.

Thanks again Andrea! Much appreciated!
andrea tasselli avatar
andrea tasselli avatar
Hi,

Normally this issue doesn't come from the flats themselves but crops up during processing if you apply ABE or DBE wrongly or too heavily. Or that you have a light leak somehwere and you overcorrecting at the center and undercorrecting at the edges. Ths said I think it could be fixed in psot-processing.  I can't judge the various channel levels by the posted image but if you are willing to share the orginal master raw channels we (well, me at least)  will endeavour to see what the issues might be.
Helpful Engaging Supportive
dk101 avatar
dk101 avatar
Andrea, let me go back and check every step in the process. I had a look at a sampling of the raw images and they don't have the bloom so perhaps i did something wrong when doing the flat-field correction or the DBE. I'll post the results here.

Thank you again for your feedback. I really appreciate it. As i said, i am new to all this smile
Respectful
dk101 avatar
dk101 avatar
Bit of an update... I went back and had a look at flat-fielding and then at the integration and I think I may have spotted something that (embarrassingly) I should have realised before.

The images i took were not properly rotation aligned. In fact, to say that they were aligned in any way would be pushing it. There is quite a bit of variance with a large proportion being at 90 degrees to others. This means that integration introduced effects that i didn't understand at the time.

Here is analysis i just did for my Ha frames. I integrated them again for all possible pairs of Combination and Normalisation in PI ImageIntegration keeping the rest of params default. Then I collated the results in a "chart" below, where Normalisation goes along x-axis and Combination along the y-axis.



The images above simply had PI STF applied individually so levels between them are not comparable. I was interested in the overall effect, not in direct level comparisons although it would be interesting to look at the signal & noise values for all and compare those too.

In my original image I used Average + Local Normalisation and I (wrongly) assumed that moonlight & light pollution are to blame for the blooms in the corners which I cropped. The result was then DBEd to remove any leftover background which then introduced further problems (I guess - this is TBC).

@andrea tasselli , do you think the above makes sense? It does to me but i have (clearly...) been wrong before

I also hope that others may find the above "chart" informative.
Helpful Insightful Respectful Engaging
andrea tasselli avatar
andrea tasselli avatar
I would use Average + Local Normalization (relative to the best, cleanest frame of the entire set) or Average + Additve with scaling (but never ever moltiplicative which is only for flats) since they yield the best SNR all other things being equal. The question that puzzle me is how come you have rotation between single frames on a single session?
Concise
dk101 avatar
dk101 avatar
I had five sessions over September and October, and I rotated my camera for another target and then forgot to move it back for one of the Wizard sessions. I really should have looked at the images as they were being taken...

I agree that Average + LN (rel to the best frame, of course) works best in almost all cases. Mathematically that makes sense. I did all the combinations in the above just to show the effect. Looking at SNRs for the images above, Avg LN > Avg NN > Avg Add with Scaling > Avg Add > Med LN > ... as expected. SNR for Average + LN is twice that of the Average + Multiplicative.

I am embarrassed that while i know how integration works I didn't think about the effect the rotated frames would have. Even worse, i assumed other causes were to blame. A lesson there for sure.

Thank you @andrea tasselli  for your help. I have to admit that i wouldn't have looked had you not asked. I'll have another go and see what comes out.
Helpful Insightful Respectful