×

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

Bi-monthly release with minor bug fixes and improvements

Focus is not good

  • Posts: 335
  • Thank you received: 19

Focus is not good was created by MORELLI

I use on my last night the autofocus. I set up polynomial algotrithm. The process ended up with success but the curves red and green didn't match at all !
So as result, the stars on capture are not enough fine.

???
11 months 3 weeks ago #91735

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

  • Posts: 593
  • Thank you received: 276

Replied by John on topic Focus is not good

Hi Morelli,

I'd recommend the Linear 1 Pass algorithm. It should be able to give better results as its more robust. If you're not familiar with it, here is some documentation:
docs.kde.org/trunk5/en/kstars/kstars/tool-ekos.html#ekos-focus

It does require a somewhat consistent focuser. By this I mean if you move to position X then move away and back to X the focuser needs to return the focus tube to the same spot (backlash can be compensated for in a couple of ways).

If you are able to, its worth playing with it on the simulator to get a feel for what needs to be setup at the start of an imaging session.

Hope this helps.
11 months 3 weeks ago #91738

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

  • Posts: 155
  • Thank you received: 12

Replied by Nigel Dunmore on topic Focus is not good

John has the polynomial algorithm changed since 3.6.0? I used to use it quite successfully on astroberry but have now moved to stellarmate to get the latest versions and am finding that it seems to ‘wander’ off to the point it’s going badly out of focus (donuts on my c6) and the hfr starts going down due to SEP not really recognizing donuts or parts of donuts as out of focus stars. In fact SEP seems to ignore quite a few obvious stars and go for ones I cannot/barely see.
11 months 3 weeks ago #91740

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

  • Posts: 593
  • Thank you received: 276

Replied by John on topic Focus is not good

Nobody's developing Polynomial (at least to my knowledge).

I personally would like to decommission it as I think Linear 1 Pass will be better. If there are boundary use-cases where Polynomial does something L1P doesn't do, then I'd be happy to enhance L1P.

I'm currently adding some new focus "measures" to L1P. These are alternatives to HFR. I'm hoping that a couple of them will be more resilient to donuts.
11 months 3 weeks ago #91745

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

  • Posts: 155
  • Thank you received: 12

Replied by Nigel Dunmore on topic Focus is not good

I’m using it at this very moment and it seems to have mostly behaved though I have switched to using gradient rather than sep. It could just be that last night it was actually quite misty plus my having to tweak/poke/fiddle with things from scratch with the new install. I’m now back to getting a solution in 6 iterations which I’m happy with.

Will be very interested in any changes that can handle donuts as they seem to be the main source of problems when focusing an sct.
11 months 3 weeks ago #91746

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

  • Posts: 1208
  • Thank you received: 559

Replied by Hy Murveit on topic Focus is not good

Nigel,

You might try using SEP, choosing one of the preset parameter sets that seem the best fit for your setup, and then possibly adjusting some of the SEP parameters. The thing with SEP is it's very configurable.

I am not suggesting the particular parameters below (they are from my development machine which doesn't have a telescope attached).
Rather just showing you how to find the parameters you might edit.


Hy
11 months 3 weeks ago #91750
Attachments:

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

  • Posts: 593
  • Thank you received: 276

Replied by John on topic Focus is not good

Have a look at the SEP Params section in the L1P Phase 2 doco here... www.indilib.org/forum/development/13108-...elopments.html#91125

I would be interested if the suggestions make any difference. As Hy says, SEP is very configurable so its very likely that by adjusting parameters you'll be able to get a better result.
11 months 3 weeks ago #91754

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

  • Posts: 335
  • Thank you received: 19

Replied by MORELLI on topic Focus is not good

I am aware about documentation. I maintain the french version of it. I use for the first time my focuser last night, during a scheduler session with the result I explain above.
My focuser is a Robert Brown model with temperature sensor. Range from 0 to 9998 with 1/8, initial step 50.
I try the different algorithms, detection mode, etc. Sometimes the result was good, often the process doesn't success. From a session to an another, no reproducibility.
I must say Bathinov is quicker and more precise!
In addition, into the Indi panel, fields title are not completely readable and there is no tooltip. So, it is not easy to understand well the settings.
11 months 3 weeks ago #91769

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

  • Posts: 593
  • Thank you received: 276

Replied by John on topic Focus is not good

Hi Morelli,

As you probably know, there is a section in the doco going through how to setup a Linear 1 Pass run. Hopefully, that will work for you. If not, post some screenshots of the setup and what you're getting and we'll try and help you out (and update the doco if its missing something or incorrect).
11 months 3 weeks ago #91778

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

  • Posts: 155
  • Thank you received: 12

Replied by Nigel Dunmore on topic Focus is not good

Hy, John

I previously tweaked sep will have to have another go. I need to record a set of images and run through them to try and get a good config, it’s not something I want to try out in the cold :)

Anyone thought of building a neural net that you can train with images for your own setup?
11 months 3 weeks ago #91783

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

  • Posts: 335
  • Thank you received: 19

Replied by MORELLI on topic Focus is not good

I'll try this evening. The sky is not enough good for capture, so I can make tries.
Last edit: 11 months 3 weeks ago by MORELLI.
11 months 3 weeks ago #91784

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

  • Posts: 335
  • Thank you received: 19

Replied by MORELLI on topic Focus is not good

I made tries during 3 hours. First I determine the initial step by searching the lenght of focus move to obtain 3 times the HFR from a Bathinov focus. The result was 80.Then I try every algorithm. I move the focuser to external limite (9987) then to 6300. The star is the polar.1) Iterative: SEP, Step 80, 5 frames, tolerance=10%. Succes in 17 step and position 6080. But you can see on the screen copy that the curves do not fit and the neither does the Bathinov.2) Polynomial: with the same settings. Succes in 17 step. The curves do not fit yet. And sure, neither do the Bathinov.3) Linear: Same settings, no success. 4) Linear 1 pass: Gradient, parabol, 10%, 3 frames. Success in 12 and position 7497. One more time, curves and Bathinov are not good.5) Linear 1 pass: On Capella this time. Success in 14 step, position=6138. On a second try, position=6245.6) Bathinov: success in 15 step, position 62327) Linear 1 pass: On Castor. Four successive tries, with moving the focus to position 9987 then 6300. - 15 steps, position= 6001 - 12 steps, position= 6301 - 11 steps, position= 6370 - 10 steps, position= 6476The process is better with gradient then SEP. I have to control that the focuser goes always at the same position on several times. I 'll do that tomorrow and try again MAP tomorrow evening.

I have had a strange issue with the alignment. Ekos says OK, star is into range, but the star is not at all on the image center !
11 months 3 weeks ago #91789

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

Time to create page: 1.132 seconds