That reminds me, one thing we also need to consider is what happens to the hfr profile graph that is displayed there now

Rob, thanks for your explanation as to what is already there. I'm not quite sure though whether the "reading of the HFR between every image" is taken by analysing a separate "focusing image" or by examining the most recent light frame that has been taken. (I never used the "autofocus if HFR >" option since I want to run autofocus as well as set the exact focus point manually. I just don't trust automatic re-focusing. Since I sit in front of the sceen all night anway, I can afford to take full manual control of it.) Deriving the HFR/FWHM reading from actual light frames would be most helpful IMO. A graph displaying the last x values would help determine whether there is a real "trend of deterioration" or just random change in light frame quality.
The HFR/FWHM reading could also be stored in the FITS header of each light frame (provided this does not violate the FITS standard). Clearly Hy's approach of logging several quality measurements all at once would take this to the next level but for now I'd be extremely happy to have a simple HFR reading - even without any graph - after each light frame.

A big THANK YOU to Rob, Hy, lhoujin and all developers for driving this forward. Currently this is the feature that I believe would improve the quality of my pictures most.

I think there is room for it to move to the focusing module. Or share the same location with tabs to switch between them.

Right, I forgot to make that part clear. It takes a separate focus image or subframe between each light image now. I agree that it would be nice to just use the light image for the hfr history and monitoring. But one issue I can see immediately is that a longer exposure will make the star appear bigger so using that for monitoring hfr and comparing it to a specific value that was derived from an autofocus image could prove difficult. Not to mention that the light images could be taken using different filters and the star could appear bigger or smaller in different filters due to the color of the star or chromatic aberration of the telescope or filter

I don't think it has to match the HFR of a focus frame, since long exposure light frames that are comparable with each other.

Thank you very much for the clarification. That's right, light frame HFR values cannot be compared to focus module HFR values. I would prefer comparing light frame HFR values only and leaving the focusing module out of the equation (for the purpose of tracking light frame quality) completely. Of course seeing will increase stars for long exposures but sub-optimal focus will ADD to it so I would expect derived HFR values to be fully meaningful still. With regard to filters HFR values should be compared among their respective groups only.

Let me clarify: I meant the user should take track of what filters have been used. I didn't ask for Ekos to keep track of it. I'd like to keep it simple and add more functionality later as needed/time allows.

So the reason I mentioned comparing hfr values with focus frames vs light frames is because the current functionality uses a focus frame between each light frame to keep track of the hfr to determine if an autofocus is necessary. The simplest implementation of the idea that was proposed would be to just take what is already there and just log the results on a graph. But if we want to change that to make it use the light frames instead to keep track of the hfr we would need to think about this so that we don’t break the autofocus monitoring that works right now.

I was not aware the autofocus monitoring worked that way. I suppose I can understand why, but it must take a fair amount of idle time to capture and download additional frames.
We must also consider the graph will be used by those who do not have motorized focusers to tell them when to manually focus. How does the current HFR monitoring system work if no focus driver is running?
However, I do see an advantage of taking separate focus frames that will be less likely to have any star trail issues of guiding is not used or was poor polluting the measurements.

I am not sure because I only found how the current hfr monitoring function works when I started looking into adding the hfr monitoring graph that started this thread. I knew that the checkbox was there, but I don’t tend to use that feature because I typically just start the autofocus myself when I feel it is needed. So I didn’t realize that it mostly did what we were looking for already, except that what I would like personally is a graph with a report as to how the hfr has progressed over time and I do not want it to autofocus itself automatically when the hfr gets to a certain value like the checkbox says currently. I still want to decide myself when to autofocus.

So we agree on almost everything. A HFR value derived from a light frame would do more than just check the focus. It would provide a comprehensive measurement of image quality and make users aware of problems other than just a deteriorating focus (like guiding issues for instance). The important thing IMO is to become aware of such problems as soon as possible (certainly before doing image processing the day after as it happens now) in order to avoid taking useless light frames and wasting precious observing time. The focus module could be left totally unchanged and do re-focusing as it does now. In addition to it, the "global" HFR value would be calculated and displayed (with or without graph) whenever a light frame is received. It would be 100% up to the user to perform re-focusing, check guiding, or take other measures in order to improve image quality then.

Yes, this is what I currently do. I run an external program that watches for new files, analyzes some (preselected) stars by fitting a 2D moffat model and then plots FWHM and ellipticity. (It also aligns images and gives a live stack view). Ellipticity usually points at guiding errors and/or wind, FWHM obvious focus errors, but also seeing quality. I do get a very good correlation between the FWHM variations and seeing measurements using a DIMM (Differential Image Motion Monitor).

The latter is one reason I don't use the automatic monitoring by EKOS anymore. There is no real way to decide if the focus drifts or the seeing gets worse, unless you also include the temperature plot (unless of course your focus system has slip). So a general monitoring module definitely should also include the temperature (but I think that is already on the wishlist, is it?).

As for monitoring in EKOS: How much work would it be to grab the (full field) star selection and HFR fitting from the focus module, and put it into the FITS viewer? I know it even already has a 'HFR' slot in the statistics which is always -1. So it seems the evaluation part is (still) missing...