One of the first objects I took with my trusty Canon 600D when I got into this hobby was NGC 7000, North America Nebula in Cygnus. Here is my progression from left to right, first shot taken back in July 2021, with the Canon unmodified for IR sitting on a iOpton Star Guider Pro, manual everything (focus / framing); moving to November 2022, using modified Canon (IR Cut removed), with Nina software at the helm and sitting on my Skywatcher HEQ5, finally the topic of this blog the image I took at the end of August 2023 using the ZWO 533 mc pro.
It's nice to see the progress being made. Obviously the dedicated astro camera is a significant factor and over the last 2 years the software available for the capture and the post processing certainly has given me more options in terms of framing, automation of the capture and Siril / Starnet have provided me with serious steps up. I would also modestly say that my capture and my processing techniques have significantly improved.
Anyway, on to the details of my most recent capture.
Capture
2023-08-11
41 Images taken (300s) through l-Enhance Filter
30 flats
30 dark flats used as bias
Seq # 001-041
2023-08-15
67 Images taken (300s) through l-Enhance Filter
30 flats
30 dark flats used as bias
Seq # 100-166
2023-08-20
64 Images taken (300s) through l-Enhance Filter
30 flats
30 dark flats used as bias
Seq # 200-275
2023-08-26
79 Images taken (300s) through l-Enhance Filter
30 flats
30 dark flats used as bias
Seq # 300-378
2023-08-27
80 Images taken (300s) through l-Enhance Filter
30 flats
30 dark flats used as bias
Seq # 400-379
When procesing mulitple sessions the workflow is to pre-process your Flats with the Dark Flats for that nights, then use the stacked pp_flat together with a master Dark to pre-process the lights taken that night. At this point for NGC 7000 I had 5 nights (sessions) wth 41-80 images per night, during the Conversion into 32 bit Fits files in Siril I start each session with a sequence number, this makes the registration step easier as I can simple copy them all to a single folder and run Sirils Global 2-pass Registration process across them in order to plots of FWHM/#stars/ background noise etc...
Registration and Quality Control
Total of 343 Lights.
17 failures
326 registered files
Failures:-
7 Clouds (030-041) - patchy night
10 star trails - after analysis against the PHD2 logs, three sets; (002-008;420-424; 124) I could see errors related to loss of stars and changes to star mass ... mostly clouds! So no problems to correct.
Other the 2 passes Siril selected pp_lights_229 / if#138 as the reference image.
Filtering
The following plots where obtained and used to set some quality thresholds.
FWHM
Roundness
Weighted FWHM
Stars
Background
Stacking
Attempt 1
I have posthumously called this the "Quantity above Quality" stack. In essence I took a mathematical appproch, simply using the numbers from the plots to remove what from the all the bad frames. This turned out to be a lesson learnt (hence the title of this blog!). In short I ended up with a very noisy stacked image, which when I attempted to post process was very disappointing given the integration time. It turns out that the numbers can lie, things like high cloud can be masked in the numbers so I was including really very bad images in the stack.
Anyway for completeness this the filtering I did and would subsequently redo (see attempt 2).
After the plots re-ran the registration process filtering out:-
weighted FWHM -top 90% (5.77)
Roundness [k-o] 4.4 - (0.75)
Siril Log - “Processing only selected images in the sequence (326), processing images of the sequence with a weighted FWHM lower or equal than 5.77304 (293), processing images of the sequence with a roundness higher or equal than 0.75003 (314), for a total of images processed of 287)”
Total of 287 with this filter were registered = 287 * 5m = 23h55m integration time
Removed further post register based on the following criteria:-
Removing remaining frames that are outside the 95% best (<0.00745)
272 Images left to be stacked
Attempt 2
After the above failed attempt with just using the Number I decided to actually view each of the lights and discard any to my human eye that looked "bad", this included, removal of those images with haze / high cloud, uneven illumination (probably from passing high cloud). I also made the filtering of the images harsher.
Removed roundness < 0.75
Removed Weighted FWHM > 5
Removed FWHM >3.5
This took me down to 220 images down from 272.
220*5m = 1100m = 18h20m
I used Average Stack with addition and Winsorised with no weighting.
Post Processing - Siril and Photoshop used
A quick summary:-
Cropped to remove stacking artefacts at the borders
Photometric Colour Calibrated
Green Noise Removal
Deconvolution
Starless Processing
Generalised Hyperbolic Stretch - lots of work on green and blue
Saturation - Global - fairly gentle
NEW - Installed Trial Version of Noise X-Terminator
Converted to 16 bit Tiff - Opened in PS and used Filter->RC_Astro->Noise X-Terminator
Saved back opened in Siril and converted back to 32 bit Fits format.
Recombined with a ASinh stretch on stars
Saved as FIT and JPEG
Result
There is a lot of detail in this image, I am particularly pleased with the "Cygnus Wall" part of this, the lower New Mexico with the dark strands of dust stretched across the glowing clouds of nebulosity.
One thing that I may do in the future is try processing this as a HOO image by separating out the Hydrogen and the Oxygen emissions and mapping them to some grade of red and cyan. This is a fairly well known approach in astronomy circles.
Komentarze