×

INDI Library v2.0.7 is Released (01 Apr 2024)

Bi-monthly release with minor bug fixes and improvements

Announcing StellarSolver 2.0

  • Posts: 2885
  • Thank you received: 819
It looks to me like the nightly version of the PPA built kstars-bleeding just fine. launchpad.net/~mutlaqja/+archive/ubuntu/...d.name_filter=kstars

I think the issue is just the stable version because the stable PPA is building the old version of KStars that doesn't have the kstars updates to use the new StellarSolver API that I added yesterday. 
2 years 2 months ago #81072

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

  • Posts: 2885
  • Thank you received: 819
Sorry about the multiple posts, I think there is a forum bug.
2 years 2 months ago #81073

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

  • Posts: 2885
  • Thank you received: 819
I made a patch for Jasem that he can use to update 3.5.7 to use the new StellarSolver and I sent it to him.  Hopefully that helps with the PPA.
2 years 2 months ago #81076

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

  • Posts: 55
  • Thank you received: 8

Replied by Paolo on topic Announcing StellarSolver 2.0

Could you please post this patch somewhere please?
I'd like to apply it to Kstars 3.5.7 in openSUSE too! Thanks :)
2 years 2 months ago #81080

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

  • Posts: 11
  • Thank you received: 1

Rendering Error in layout Message/Item: array_keys(): Argument #1 ($array) must be of type array, null given. Please enable debug mode for more information.

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

  • Posts: 11
  • Thank you received: 1
Forum error, causing duplicates
Last edit: 2 years 2 months ago by Jussi Saarivirta.
2 years 2 months ago #81091

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

  • Posts: 219
  • Thank you received: 41
As you have been recently working on Watney I've a request :). I'm currently using a fast Newtonian (f4). This kind of telescope is know for showing donut images almost every time. They only shows something like a star very near the focus position. The problem appears when we are trying to run the autofocus process. If you go too far from the perfect focus spot, instead of detecting the center of the unfocused start (the donut), every algorithm used until now, detects the brighter region of this ring... that give us a wrong HFR.

After reading the way you are using the extract the sources, could it be able to cope with this kind of images? I mean you talk about contiguous pixels over the background. Do you think it will be able to cope with this kind of images?

Thanks
2 years 2 months ago #81093

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

  • Posts: 2885
  • Thank you received: 819
Thank you Jussi for the feedback, I believe the current plan is to release the next version of KStars in a week or two, so any information like this is very helpful.  Now that I think the API revision is done I can go back to focusing on any bugs like this.  I will look into what you mentioned

Paolo, I sent the patch to Jasem and I would like him to take a look and make sure it works ok.  If there is something else in KStars that needs to be changed that I missed such as something Jussi mentioned, I should probably add that to the patch before I post it.  So yes, but let't make sure the patch is good first.

Also, about the question about Reflectors and focusing.  I also would like to see this improved since I have reflectors not refractors and it is very important to me.  I was looking into it recently, but I have been working hard on getting StellarSolver 2.0 up and running.  But I did make some progress. I added some convolution filters and options for that to the new StellarSolver and it should be able to detect the out of focus star easily in the star extraction.  I found that the Mexican Hat algorithm seems to be better at that than a gaussian.  I was also experimenting with other shapes of convolution filters too.  The biggest thing though is increase the fwhm of the convolution filter and it easily detects the whole star no matter what the algorithm.  I would like to build that function into the KStars focusing algorithm in the future, but I have to finish the stellarsolver integrations first.
The following user(s) said Thank You: Rafa Barberá
2 years 2 months ago #81094

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

  • Posts: 11
  • Thank you received: 1

@rbarberac
Hey, if you have some sample images I could look at, that would be nice - theoretically small rings could work out just fine. The Watney built-in algorithm fairly crude, but works

- Jussi
The following user(s) said Thank You: Rafa Barberá
Last edit: 2 years 2 months ago by Jussi Saarivirta.
2 years 2 months ago #81095

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

  • Posts: 300
  • Thank you received: 57
ROB LANCASTER WROTE:
"I think the issue is just the stable version because the stable PPA is building the old version of KStars that doesn't have the kstars updates to use the new StellarSolver API that I added yesterday. "

Exactly. It's a mismatch in the ubuntu ppa between the "stable" Kstars and StellarSolver.

I was on stable Kstars ppa but my stellarsolver updated anyway and killed Kstars completely.

Once I updated to Kstars nightly ppa, I can run again.

But there's no point to having a "stable" Kstars that can't run!
Last edit: 2 years 2 months ago by Scott Denning.
2 years 2 months ago #81096

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

  • Posts: 219
  • Thank you received: 41
Not at hand right now. Next clear night I'll generate some of them :)
2 years 2 months ago #81097

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

  • Posts: 2885
  • Thank you received: 819
Jussi, Saarvirta, please check to see that this fixes what you were observing in terms of the typos, file paths, and tooltips:

invent.kde.org/education/kstars/-/merge_requests/559/diffs
2 years 2 months ago #81103

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

Time to create page: 0.673 seconds