From a recent discussion on Cloudy Nights:
There was a very interesting thread recently on guidescope to scope ratio, that got a bit off topic, and veered into a discussion of FWHM. I'd like to continue that discussion, and follow up on something Frank said:
"I think it's fine to compare images as a way to assess guiding quality - but it should be in addition to stating the fwhm, in arc-seconds, of stars in the raw images. I think anyone involved in autoguiding and striving for improved results should measure these numbers on each imaging session and try to make them smaller - assuming they do want optimal results with their equipment"
I agree completely with this statement. When I started imaging, I was completely in the dark so to speak, and had no objective way to evaluate the quality of my data acquisition. Guide RMS is fine for some purposes, but doesn't say how round and small your stars are. That has the be the ultimate measure of the raw data quality. I agree that we care about how the final image *looks* but we have to start with high quality data.
Stars are point sources, and a good FWHM analysis tells us a lot about guiding, focus, polar alignment, sky conditions, etc. As I said, I didn't realize this at the beginning, and just kind of flailed around, wondering why some frames were better than others, and why some images were a bear to process.
After reading and listening to other more experienced imagers, I finally started tracking FWHM during acquisition. I use CCD Inspector to watch my data folder, and I look at the FWHM for each frame as it's downloaded (if I'm awake!) I can see right away if there's a focus or tracking problem. The FWHM will start to increase, or the roundness will decrease. I now have a good feel for my equipment and skies ( around 1.8 to 2.2 arc seconds depending on skies), and have a lot more confidence that I'm getting good data.
One of the things that initially confused me was that I couldn't see how big stars and little stars could have the same approximate FWHM. Well, they do! This assumes that the image is well sampled on the CCD - an undersampled image can be very misleading, as we need the star image to span a reasonable number of pixels. I've taken a raw frame and analyzed it star by star in PixInsight, looking at individual FWHM values. The average is very close to the CCD Inspector value for the whole frame. This tells me that using CCD Inspector (or similar software) during data capture is a valid and very important step.
I encourage all of us to listen to what Frank has been saying, and to keep paying attention to FWHM. It's not the only determinant of how your final image will look, but it's a critical first step in a good image.