With perpetual winter cloud cover I thought of checking my raw subs for unusual phenomena. Plenty to go through however being in Bortle 3 sky all of mine are 2-3mins; not good for finding short duration events. What works well is short exposures 1-15s and lots of. Color/bw, image resolution, format etc don't matter. The program I am using essentially takes a folder full of images and automatically sorts these into clusters ranging from most frequently occurring frames to least occurring ones; later is where unusual stuff potentially resides. Anyone willing to share such subs or even videos, will be fabulous.
request for data
Interesting. At that shorter exposure time, lucky imaging frames come to mind (although those are obviously in milliseconds!). A program like that might be a neat way to find the individual lucky frames that include things like a bird or a plane (or balloon) across the sun, or across the moon (maybe even for ISS passes). Those would probably be big SER videos though. Would that work for your program? Cheers
Engaging
The program is for surveillance videos so it be worth trying for rare coincidence such as described as well; big SER videos be great.
Interesting. I have a 6GB SER file of the solar disk from this morning - I'm certain there are frames in there of swallows flying across the face (some nice silhouettes). We could test to see if the program finds them. How do I get the SER file to you? Cheers
Well Written Engaging Supportive
Great! I've created some space on google drive. Here is the link, thanks:
https://drive.google.com/drive/folders/1a-gx4J8hmJOLxhJXWmaRBCehfD7a3JKy?usp=sharing
https://drive.google.com/drive/folders/1a-gx4J8hmJOLxhJXWmaRBCehfD7a3JKy?usp=sharing
I couldn't get that to work for some reason (it was asking me to log in but I wasn't sure to what) but then I realised I could put it on dropbox so have PM'd you a link. Looking fwd to seeing what your program rustles up. Cheers
Hi Vineyard, thanks for the SER. In total 1000 frames were found. First attempt, I tried to analyze all 1000 in one go, didn't come up with much except for one image with a thick black line. Next ran in smaller batches of 200 frames (in separate tabs of the browser same time almost). Second batch i.e. images 201-400 had multiple anomalies. Program outputs copy and pasted below.
That's v cool - how long did the software take to run? Is it Mac compatible? B/c that would make it easy to scan for interesting overlays (like a bird silhouetted against a stacked Ha sun). Thanks!
Engaging
Didn't time it, was playing with it whilst web browsing during lunch break. Think it took about a minute to load data and do the first trial. The only thing that was a bit tedious was finding the right number of clusters for the dataset, after that it was adding folder paths, and pressing the 'run all' button. I'm using it on mbp 2018; one with the Intel chip.
CS
CS
Re-ran the program with all 1000 frames together but clusters increased from 11 to 50. Clustering time almost unchanged - 15s. Far simpler, and one more swallow.
