Autofocus Run based on change in HFR

While you guys are in the update and improvement mode, it would be fantastic to have an autofocus run based on change in HFR… based on a percent change in some base number. Seems like an easy add, you already have HFR! Thanks for considering this. Time, number of frames, or temperature just are not very reliable for me, but if HFR starts to creep then it is time to refocus.

1 Like

Well… a couple things:

  • 3.1 is locked. We committed to altering the way we calculate fit (focus position) for 3.1 and we will focus on actually capturing better data in 4.0
  • HFR triggers… This has long been requested over the years. Since HFR can tend to vary by altitude for the same star, a low star with increasing altitude will, for the same focus point, exhibit a decrease in HFR (and vice-e-versa). This seems like it might produce a false sense of security since the HFR appears to be stable, but SHOULD actually change. I don’t claim to be an expert with auto focus math… A lot of what I just said is anecdotal, based on my own image history graphs. I’m curious to see other data… especially locations that attain a higher altitude. For obvious reasons, I am hesitant to add a feature that could be misleading or fundamentally flawed. If we can get a consensus across a bunch of folks, that their HFR data is steady-state the entire night, we can consider this for 4.0.

In order to gather appropriate data to prove this out, anyone interested would need to set up a sequence and sacrifice some time by forcing AF every 15 minutes or so. This way we can be relatively confident that the HFR recorded in image history is representative of focus. If this HFR is constant (or relatively constant to a degree that HFR change from temperature makes insignificant), it seems viable to me.

I have image history enabled all the time. Is the HFR information for image history stored in a log file somewhere? If it is and you want some data I can send it to you. My AF runs were every 40m for RGB and every 60m for NB, maybe not fine enough resolution for what you need. From the end of September all through October was excellent for imaging. I had quite a few successful overnight runs using SGPro :). November doesn’t look so good for imaging at the moment.

Regards

Derek

No, but if you open the sequence, you can export it to a file, by clicking the little down arrow on the dockable module.

Hi Ken,
I was looking at some of my HFR data in the image histories and to be honest I would not really know what limit to put on HFR variance trigger. There are several reasons that HFR could vary. I wonder when a refocus triggered by HFR would be required? For example would you want to re-focus if seeing temporarily got worse and then improved again or would you just want the focuser to maintain its position? On the other hand, I have a situation on one of my telescopes where the focuser shifts after a meridian flip. In this case I can see a HFR trigger possibly being helpful but I just choose refocus after centering and the sequence continues in focus anyway.
Anyway I do have data I can share but clicking on the little down arrow in the lockable module for image history does not give me an option to export data:

I just see Image History Detail, Clear All, Clear this Series or Help - Image History Module as shown in the above screen capture.
The image history in the screen capture shown above is interesting to me. The first three points are from around 23:15 then I sequence through the other filters coming back to the Red filter at about 3:40 . By now the seeing is a little worse and my AF has given me a higher HFR and fewer stars than before, that’s OK. All the remaining points on this HFR history plot are at the same focuser position. I can see the HFR getting better then getting worse and then getting better again a couple of times. My question is would an HFR trigger improve this sequence of images?
Anyway if I misunderstood how to export the HFR data to a file please let me know.

Regards

Derek

Honestly if you guys are that worried about it you should take out a 3rd mortgage on your home and buy an on axis guider with focus lock or whatever that new software he has is called. Then you never have to worry about your stars drifting out of focus during a frame. Because that is what you are really talking about right? At the end of a 20 minute frame your HFR is potentially drifted out by X% and that would then trigger a re-focus. Of course you only want to flag it if it gets worse not better. So +X% HFR = Refocus.

How long are your focus runs taking? Can you speed them up? Bin 4x4 for example? Can you crop down your image in such a way that the downloads are faster? I have actually not looked to closely at this but if someone can tell me how that works: See #3 in screenshot. Does this crop before or after it downloads? If before, you could have a rather fast download and focus run. So then you turn on #1 and set it to focus after however many frames you like just to be sure.

I think the reason to have HFR triggered focusing is to save time. You will have no re-focus runs if your HFR is stable all run. You don’t want to focus because it takes time away from collecting photons. But how much time do you actually lose?

Also maybe a high cloud blows through and for that one image your HFR is a bit higher. So it auto focuses and then the cloud is gone and the best focus position is no longer the best focus position. Now you have two frames that are not as great instead of one.

All that being said, if it works that HFR +X%=trigger a auto focus works, then ok I’m in. I’ll always go with whatever will get me the best data.

Just don’t want the developers sidetracked on something that is a waste of time. They can make that call based on user input. I will use whatever gives the best results I won’t use anything that doesn’t. I asked some questions and gave some feedback and am willing to give more.

Regards

Derek

I really don’t see much value in this proposal. First of all, it does not seem very well defined.

What base number? This sounds like it would need to be customized by each user.

Does any other software do this? I am not convinced this would even work in practice. I see too many possibilities for random results. HFR is affected by a lot of factors, only some of which would make sense to act on.

There are too many other guaranteed beneficial enhancements that I would like to see the developers work on.

The current product has a myriad of options for controlling when to focus. Surely some of them will work for you. Maybe there are other issues that you could address first that would improve your focusing.

Does any other software do this?

Yes, and I think it is what is leading to feature requests like this. But just because someone else jumps off a bridge doesn’t mean SGP has to.

I requested ASTAP for example but only after 1) researching it and using it with other software and liking it a lot and 2) knowing it would not be a huge hassle to add it to SGP. I am glad I spoke up. I am sure Jared and Ken would have gotten around to it eventually though. Refocusing with a bad change to HFR though seems like 1) difficult to add and 2) not worth while.