×

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

Bi-monthly release with minor bug fixes and improvements

New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed

  • Posts: 106
  • Thank you received: 4
Please have a look at class AstrometryParser. Is this a typo?

virtual bool startSovler(const QString &filename, const QStringList &args, bool generated = true) = 0;
Powered by

GNU / Linux
Git
KDE neon
KStars | EKOS | INDI

and some cheap hardware
3 years 4 months ago #62236

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

  • Posts: 2876
  • Thank you received: 809
I assume you mean the "startSovler" part?

I would say yes, somebody meant to write startSolver and goofed. I just took a peek, and it looks like it has been called that for more than 4 years. That is fairly easy to change, but I am not sure we need to do so because:

The good news is that all those astrometryparsers should not really be needed anymore because now all that options parsing is handled internally in StellarSolver to make the code simpler and cleaner. Not only that, the same options profiles will work for all the methods StellarSolver uses, ASTAP, internal solver, local astrometry, and online solver, instead of having many different options for the different methods of solving all in different places and some in configuration files that could be set incorrectly. Now it is just one set of options, all internal in KStars in the one place, so you can'f have a messed up file somewhere preventing your astrometry from working.
The following user(s) said Thank You: Heiko
3 years 4 months ago #62237

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

  • Posts: 1309
  • Thank you received: 226
KStars Built Build: 2020-10-30T03:22:05Z
Last night I compiled StellarSolver and KStars. I was pleased to find importing RAW images is now supported Thank you.
I also tested with a few other files, and noticed one of my JPG images I had successfully been solving does not load properly now. The image itself is now distorted.

The original jpg is also attached.

I also had one image I previously could solve in around 10 seconds but now took at least 80 seconds to do. My trouble is, now that half of the preset profiles have been pruned out, I am not sure what settings made up the difference. I can recall I had some of my best results with the now removed Fast Solver preset.
This image is attached. The Bubble Nebula.

Lastly. I do not believe the RAM check is not reliable at the moment. I have encountered some crashes with it on. I believe I am just on the limit of stable use of Load All Indexes on a Pi with 4GB. It will either:
1. Work really fast with indexes loaded into RAM
2. Will say I do not have enough RAM and work slower
3. Or concerningly it will say I can load into RAM, but then crashes.
I have attached logs that should help, but do note they were from an earlier build, details are included.

Thank you.

Edit: I found that JPGs exported specifically from GIMP are distorted when loaded. Any ideas as to why that might be? The image used to load without issues until only very recently.
Last edit: 3 years 4 months ago by Andrew.
3 years 4 months ago #62263

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

  • Posts: 106
  • Thank you received: 4
This is a post which is slightly off-topic:

I have the impression that we have reached the point where the development is freezed. This will give the development team time to update the user guide.
That's all very well and -no doubt- a lot of work.
But I would like to know when KStars' API Reference will reflect the new code base of the upcoming version 3.5.0?
Will stellarsolver's code base be merged with KStars' code base?
Is the majority of the API Reference automatically generated as soon as version 3.5.0 is a stable release?
I am sure that there will be talented developers in the future who will maintain and improve the programs. But an up-to-date API Reference could pave the way for users who would like to understand a little bit of the underlying source code, which to my eyes is really complex.
The knowledge of the developers is valuable and should be passed on. Maybe they could write a book about coding in Qt and C++ using KStars/EKOS and INDI as the examples. Some UML charts, more explanations.
Sorry for bothering you, just dreaming.
Powered by

GNU / Linux
Git
KDE neon
KStars | EKOS | INDI

and some cheap hardware
3 years 4 months ago #62301

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

  • Posts: 2876
  • Thank you received: 809
Hi Cerro Torre,

I wouldn't say the development is freezed, it never is. The new release of KStars 3.5.0, as its name indicates, is a major release with lots of changes. It is a huge overhaul in how solving and star extraction is done, a huge overhaul in how FITS Files and other Files are handled, and a number of other great changes. It probably looks like the development is done from the outside, but I can assure you, we are still fixing a bunch of little issues.

The plan is as of now to keep StellarSolver in a separate repository. My thought is that it can be used by many different projects for star extraction and for solving. I foresee it being a general use tool for cross platform astronomy applications. I primarily developed it for KStars, but I think it can be more than that. It is of course open source and of course I regularly contribute to KStars as well, so if someday we find we need to integrate it, we can do so. But really, the way SEP and astrometry code were being used in KStars before, it was like a tangled spaghetti. I think having the code separated and making it a dependency simplifies the code in KStars. It encapsulates the functions and related code so that KStars just has to set options and tell it to either solve or extract stars, StellarSolver handles the rest.

I don't know the details about API updates and how that happens, but I do know that we need to update the KStars Manual before we do the release so that people know how to use it. I agree with you that everything should be well documented so that newcomers can figure out what is going on. I have been trying to produce Setup Scripts in repositories that would get new developers on Windows, Mac, and Linux set up to develop for KStars. These are not really ready for general use, since I have been focused on these other priorities, but I hope to get back to them soon.

Thanks,

Rob
3 years 4 months ago #62306

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

  • Posts: 2876
  • Thank you received: 809

I don't know why the Jpeg images would load poorly now. I know Jasem was working to improve the loading of Raw images and other image types, maybe something with that changed it. I can forward this on to him.

For the RAM check. Maybe we need to be more conservative, since I would strongly prefer that astrometry take longer to solve than for a crash to happen because memory is exceeded. Right now, the way I have it set is you can select the option to load all the indexes or not select it, but then if it finds there will not be enough RAM, it will automatically shut off the option. Making it more conservative though, means that someone who might have had just enough RAM to run the indexes in Parallel might not be able to do so since we are being conservative. I will think about this though because maybe it would be best to have three options instead of 2, auto, on, and off. That way, we could make the ram check more conservative, but if somebody wants to risk it, they could turn it on and ignore the check.

So if you are looking for FastSolving, that profile is essentially now SingleThreadSolving. The primary difference between SingleThreadSolving and SmallScaleSolving is that when the SmallScaleSolving profile has to do a partially blind or fully blind solve, it solves using multiple threads in parallel. I tried to keep the best parts of the FastSolving profile in the new ones.

I have included below the main settings that I used to create the old FastSolving profile and the new SingleThreadSolving Profile. Most of the settings are the same like the conv filter and the number of stars to keep. I did change the min width to 0.1 degree from 1 degree, because there are lots of folks who have smaller scales than 1 degree and I wanted it to work for them too. I didn't see much of an increase in solving time for that change, but you can put it back to 1 degree and see if it makes any difference. The downsample setting looks different, but now that there is an autodownsample feature, it should do that automatically if your image is big. You can try turning off autodownsample and setting it back to 2 and see if that makes a difference. I think all the other settings were really the same as the old FastSolving.

3 years 4 months ago #62307
Attachments:

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

  • Posts: 1309
  • Thank you received: 226
Thank you Rob.
I just noticed that my 'original' JPG images were compressed when I posted them. That may change how they load and solves. So I have gone back and attached them within a zip archive to protect them from compression for further testing.
3 years 4 months ago #62308

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

  • Posts: 2876
  • Thank you received: 809
So I tried your two images both in the StellarSolverTester program and in KStars.

The Bubble Nebula image solved in about 2.5 seconds in both KStars and in the tester with the SmallScaleSolving profile. It loaded fine in both programs

The M33 image took less time to solve in the Tester program (I am attaching the results), but just as you observed KStars seems to have an issue with loading this image. I can send it to Jasem.

3 years 4 months ago #62309
Attachments:

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

I just tested with both JPEGs and they both load perfectly fine and solve fine as well. No distortion at all. So I take it it must be something at your end causing this.

3 years 4 months ago #62310
Attachments:

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

  • Posts: 106
  • Thank you received: 4
This makes sense.
I would welcome the opportunity to get access to some kind of training material; count me in - at some point in the future ;-)
Powered by

GNU / Linux
Git
KDE neon
KStars | EKOS | INDI

and some cheap hardware
3 years 4 months ago #62317

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

  • Posts: 1309
  • Thank you received: 226
Although Rob was able to repeat the issue yesterday. After a fresh build today, the distorted image issue did not return.
3 years 4 months ago #62326

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

  • Posts: 2876
  • Thank you received: 809
Yep, I did see the distorted image. But it could have been fixed already. . .
3 years 4 months ago #62330

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

Time to create page: 0.596 seconds