This is something that I’ve been observing for awhile now and wanted to share. My usual culling routine would be to reject images with both high eccentricity and high FWHM. But when taking a closer look it seems that a lot of my subs with high eccentricity still retained good detail in the non-stellar parts of the image. I think I understand the reason for this. let’s say you’re taking 60 sec. subs and you get a little wind gust or an RA guiding bump that lasts a few seconds. Because stars are so bright, this slight bump is enough to bring the stars out of round and produce a frame that would score low on eccentricity but, because the signal is building up so slowly in the fainter areas, they are relatively unaffected. Here’s a side by side comparison that shows the effect. These are 60 sec. subs of M-51 with a uniform histogram stretch and slight noise reduction to make the detail easier to see:
📷 result.png
📷 result-2.png
You’ll note that in the core, the high eccentricity frame actually has better detail then the best eccentricity frame and is close to the best FWHM frame. I’ve done this side by side test many times and always get a similar result. Of course you still want to discard frames that have extreme elongation or multiple star images but that’s about it in terms of eccentricity. You do want to cull deeply using FWHM as that really seems the best indicator of sub quality. I’ve also found the number of stars detected in the frame often correlates well with best FWHM frames so it can also be a powerful indicator of sub quality.
Just to give some numbers here, in the FWHM subs the best has an average of 3.12”, the worst comes in at 3.75” The worst eccentricity sub. has an average of 3.17”
