×

INDI Library v2.0.6 is Released (02 Feb 2024)

Bi-monthly release with minor bug fixes and improvements

Suggestions for Focus module

  • Posts: 527
  • Thank you received: 139
Since 3.0, I’ve been imaging with quite a few sequences, and have seen some repeated issues crop up with focusing.

I’ve seen the Focuser system automatically select stars in the corners multiple times. This is certainly a poor choice if you have any off axis issues like tilt, or if you have coma in the corners that needs to be cropped out. It should try to pick stars within a center circle area of the image.

Secondly It can tend to grab rather large stars. If you’re imaging with an RC and you have diffraction spikes, this creates some issues on the focusing module knowing what is the best FWHM for that star. Alternatively I’ve seen it pick extremely small 1-2 pixel stars, then easily lose the star as the focusing routine starts. Seems the best option is to pick stars between an FWHM of 2-6 and anything outside of that size should be discarded. Of course, this might make things challenging if someone starts the focus routine out of focus. So maybe there are several steps where it picks any star first regardless of FWHM to try to get it in rough focus first, then it moves on to a star with an FWHM of 2-6 to ensure focus is accurate. This to me would be way more reliable.
The following user(s) said Thank You: Alfred
5 years 2 months ago #33438

Please Log in or Create an account to join the conversation.

  • Posts: 435
  • Thank you received: 31

Replied by Paul on topic Suggestions for Focus module

Andrew,
I am glad that someone else has also noticed some unusual focusing behavior.
If I let the software select a star it always selects one near the upper left corner, fairly close to the edge. I would prefer that it selected one about a third of the way from the centre, if possible, to maximise the focus for the entire frame.
I think it is selecting one with an appropriate brightness so I am not be seing the second part of your issue.
Paul
5 years 2 months ago #33445

Please Log in or Create an account to join the conversation.

  • Posts: 1029
  • Thank you received: 301
Alternately, you can use the full frame focusing option. It will smooth out the specificities of the frame by averaging the HFR of the detected sources. However, it is slower eventually (5-10 seconds lock-up on a 1.6GHz monocore for a 1Kx1K frame) and less stable than a single-star procedure. A compromise for an all-round automation perhaps.

-Eric
The following user(s) said Thank You: Derek
Last edit: 5 years 2 months ago by Eric.
5 years 2 months ago #33453

Please Log in or Create an account to join the conversation.

  • Posts: 150
  • Thank you received: 15

Replied by Picard on topic Suggestions for Focus module

+1 for the first proposal.
5 years 2 months ago #33471

Please Log in or Create an account to join the conversation.

  • Posts: 456
  • Thank you received: 76
+1 for the full frame focus option. I get much better results with it. Its a bit slower but not much for me (although I'm using an i5 quadcore desktop)
5 years 2 months ago #33473

Please Log in or Create an account to join the conversation.

  • Posts: 105
  • Thank you received: 30
I would encourage also providing hooks for a third party focus routine to be run during a sequence. I've been working on something and at the moment I just use it with some code I've thrown together for data collection. I tried running it as a script between frames during a sequence but the way EKOS pretty much assumes any BLOBs are for it during a sequence (even when a script is run) when the external autrofocus software started taking frames they were captured and written out as sequence images by EKOS.

I think having EKOS act more as a director of a symphony of tools (astrometry.net, PHD2, external autofocus, etc) would be better than locking up useful tools in the EKOS framework. This has been a real boon for the Windows/ASCOM world as there are tons of programs (ACP, Voyager, SGP, etc) that can be used to automate imaging and each one doesn't have to reinvent all the parts each time.
The following user(s) said Thank You: Eric
5 years 2 months ago #33475

Please Log in or Create an account to join the conversation.

  • Posts: 981
  • Thank you received: 159

Replied by Alfred on topic Suggestions for Focus module

+1 for first proposal which is reasonable and would clearly improve reliability IMO.
The following user(s) said Thank You: Derek
5 years 2 months ago #33476

Please Log in or Create an account to join the conversation.

  • Posts: 435
  • Thank you received: 31
Although there are some advantages with the Full Frame option the download speed is a killer for me - approx 20-40 seconds depending on which camera I have. Having a faster CPU makes very little difference.
If you are attempting focus with Narrowband filters you need significantly longer exposures to have a star, which makes it worse.
Paul
5 years 2 months ago #33477

Please Log in or Create an account to join the conversation.

  • Posts: 3
  • Thank you received: 0
I am in the process of learning to use Kstars/Ekos and was testing autofocus yesterday when I noticed the very behaviour you describe. Auto select star checkbox results in a star being selected in the upper left corner. I am using a rasberry pi 3b+ with the stellarmate os ( downloaded 20190404 - so quite new ;)
My focuser is pegasus astro - dmfc driver
Are there any specific news or settings related to this issue, available?
4 years 11 months ago #37522

Please Log in or Create an account to join the conversation.


I don't mind at all adding support to external tools that can do a better job than the internal ones. I believe the star selection can be further improved if the SEP algorithm can be tuned up further.

All development work for KStars happens on phabricator.kde.org, please join us!
4 years 11 months ago #37524

Please Log in or Create an account to join the conversation.

  • Posts: 328
  • Thank you received: 53
There are major issues with SGP being the conductor of the symphony. Which is why I've spent a month trying to learn this software. SGP does more signaling and listening. If it doesn't her back it can just 'go out to lunch'. Things seem to get hung up... platesolve2 takes 30 seconds to abort. If you do ANYthing out of the ordinary or try to abort a step - FORGET IT. It does a sequence and good luck if you just want to grab a quick sub. If it does seem to lock up it's a reboot of 5-6+ applications and can be a 10-minute process. I DO NOT RECOMMEND THIS CONCEPT. You must do what SGP expects and in proper order. It is not forgiving.
There is a reason after 2 years of SGP I'm HERE!

Kstars/Ekos is more of a single unit. (when compared to SGP) and it multitasks much better. I can send a image to be solved and still grab another image with the main camera while it's solving. I can zip off 4-5 subs in the camera without mapping out an entire sequence. There is no comparison between Kstars/Ekos and SGP. While we could learn some things from the UI of SGP... I SURLY would not recommend running a lot of external programs. But yes, the internal guiding code could use an update to a version of PHD2. (the ASIair has a PHD lite) built into their software. (works great).

We don't need ANOTHER SGP - we need BETTER than SGP. And Kstars/Ekos is already VASTLY SUPERIOR.

I have excellent results with the autofocus. Not at first with full FOV focus, but allowing it to pick a star and crop that star works extremely well.
I'm using a focuslynx / feathertouch focuser.

Sounds to me like you should solve your tilt/edge star problem. I spent 2 weeks working on flattening my field with proper spacing and now it can gab any star for focus. But it shouldn't be hard to make it use the center 40% of the frame.
AP Mach1 / CP4 APCC & PEMpro.
EXP SCI - ED152cf APO - Celestron 11" RASA - Stellarvue 80mm
Baader F2 HS NB filters, Lodestar X2 guide camera / OAG - ZWO 290mm mini
ZWO ASI1600MM Pro / ASI174M (solar) / ASI094MC
NEXDome, CLoudwatcher, AVX mount/ASIair and Stellarmate

The following user(s) said Thank You: Urmas Leming, Richard Roth
Last edit: 4 years 11 months ago by Ron Kramer.
4 years 11 months ago #37536

Please Log in or Create an account to join the conversation.

  • Posts: 1029
  • Thank you received: 301
About the focus module picking a star in the top left corner, this is not entirely true. The first full frame taken has a default location for the green square which is not the chosen subframe, and is misleading. There are several UI elements to improve in this module indeed. So sorry that my list be coming up so slowly to phab.

-Eric
4 years 11 months ago #37546

Please Log in or Create an account to join the conversation.

Time to create page: 0.864 seconds