I've been battling with focus in Ekos recently. I'm using the linear algorithm (my focuser has some backlash so this should help out) and that seems to be behaving. But i'm having problems with the star detection algorithm picking up stars that are not stars, getting crazy low HFR values and it keeps going, finding "stars" when I can see the actual star is way out of focus. I've tried this both full field (when it detects the fake stars theres way fewer of them) as well as subframe. Subframe you can clearly see the star is defocused yet its picking up things.

I'm using the SEP algorithm with the default profile, I've tried averaging it over 2 frames, it does the same thing on Lum or Ha or any filter. I'm running 1600mm on an 8" f/3.9, lum I use 1s exposures and Ha 3-5s. I'm more or less sure i've done something wrong but I'm not sure how to fix it properly. Any ideas what I could tweak to help it only pick up the proper stars?
 

The average feature does not average the frames, but averages the mean HFR over a certain number of frames. If your focus is disturbed by noise, I suggest you bin to 2x2 (and reduce exposure) and/or use a median filter. You may also change the detection profile to see how the procedure behaves.

-Eric

Thanks will do when the skies are clear again. Where do I find the median filter?

Are you taking dark frames for focusing? This will eliminate "fake stars" from being selected for focus. It really improved my focus accuracy.

Ron

I haven't, It seemed to not be saving files off to the dark library and kept asking to cover the scope. I'll try again, i might have done something wrong

Sorry it posted multiples times

I'd also say that you need to start autofocus somewhat near good focus. It doesn't do a good job of detecting stars when you're way out-of-focus. So, if the ideal will have an HFR in the range of, say HFR=1.5, then starting with an HFR of 5-10 will likely not work out. If you start at 2-3, things will work much better. 

You also need to adjust the step size for your focuser. Not sure what to say, but you don't want to use very small steps, where there are 5 inside the critical focus zone, nor very large, where a few steps and you're out of the range where stars can be detected well.

Have you tried the "Dark Library" feature in the Capture tab? It makes it really easy to take all of your dark frames in a range of exposures and bins.

Ron

Thanks, I typically start close to perfect focus. Typically get 1.8 HFR and I start 2-2.5 sort of range. I had a step size of 50 which throws it WAY out, I'll try making that smaller as well.

I'll try this again later tonight. Didn't have luck capturing darks when I did it before but will try again

My 2 cents as I faced such problem before (for a whole night), use 2x2 binning, start close to good focus with L filter, make sure your mechanics is suitable for your focuser (in my case max step was 1000, and each step is 100), set the exposure time for Ha, O3 and S2 filters to 15 seconds and most importantly in my case was enabling the coil power, as I discovered later that the optical train was heavy and always change position by gravity.

Since I made the above changes, my focuser works fine since then

hope this helps

Thanks for the help guys. Its been working reliably so far. I found that adding the median filter, 5s exposures for the narrowband filters, and changing the step size down and max steps out have helped a ton. WIll keep trying it out to see if its really 100% reliable.