Subframe Selector - Trying to understand FWHM Better

[james423896]

Since switching to a mono ASI1600 the increased sensitivity has forced me to pay more attention to my star shapes and sizes. Coupled with the new subframe selector process I have looked back through my data and noticed my measured FWHM is always around 6" which seems incredibly high for a 6" newt unless I am seeing/guiding limited? I went back over some of my dslr data and found one dataset on M81 where my FWHM was 2.18". Visually the ASI1600 subs look better so I am wondering if I have something wrong in the subframe selector script?

Some example files:
https://1drv.ms/f/s!AtooZYq7lKxzgTWdvW_aGEfCbJhn

The M81 data is shot with a 100D @ 1.33"
The IC405 data is shot with the ASI1600 @ 1.15"pp

Guiding performance for IC405 was 0.7" Tot RMS with both axes < 0.5". Not sure what the guiding was on my M81 data but it was with my AVX mount so I would expect it in the range of 1.5" Tot RMS

Any thoughts on what is causing the difference, or is this confirming I typically have 5" seeing? with that M81 data a rare (for me) night of 2" seeing?

[james423896]

The weather was poor last night (windy and rain showers) but I recollimated and shot some short exposures of M81 between the clouds to see what my FWHM would be to try and get comparable data (from a target perspective).
5x10s = 4.19"
5x30s = 4.34"

[james423896]

The screenshot attached shows the stars from a 10s, 30s (ASI1600) and a 300s DSLR sub. The DSLR sub allegedly has a FWHM of 2.2".

The DSLR sub is clearly more bloated and SAO15018 (separated by 8.9") is more defined in the ASI1600 subs despite them having a FWHM > 4"

I also noticed the DSLR subs had a star count 4719 compared to 100-300 for the ASI1600 subs. It therefore looks like I have something off in the SubframeSelector script when looking at the DSLR subs?

Interestingly if I use DynamicPSF and select a few stars in each image (the same ones) the results are as follows.
10s  - 3.46x3.22px @ 1.15" pp
30s  - 4.17x3.18px @ 1.15" pp
300s - 5.59x5.16px @ 1.33" pp

This confirms what I am seeing, but I cannot work out why SubframeSelector and FWHMEccentricity are thinking the FWHM is way lower in the DSLR data.