Downloaded and compiled on RPi4 the master branch of StellarSolver. Running StellarSolverTester and opening a FITS file, shows the next warnings

When I press START it crashed. In the terminal I can read:

Can you share this file you tried to solve?

Yes, of course!. Is a stack generated with the last version of SiriL You can download it from my drive crash on RP4

Rob, I did some testing. It is hard tinkering because I do not know
the underlying pattern. What is SEP's function and in what way differs
StellarSolver from Local Astronomy? Why are the results on the same
light frames different? Is this a matter of the fine-grained settings?
Is it only my subjective impression that the parallel solver is faster
but not necessarily more succesfull? Some more information could be helpful.

SEP | StellarSolver | 1-FastSolver (no Parallel Algorithm)

- M66 ok

- Iris Nebula ok

- M57 failed

SEP | StellarSolver | 4-ParallelSmallScale

- M66 ok

- Iris Nebula failed

- M57 failed

SEP | StellarSolver | 7-Midsized-Stars (with Parallels set to
Automatic)

- M66 ok

- Iris Nebula failed

- M57 failed

SEP | StellarSolver | 7-Midsized-Stars (not set to Parallels)

- M65 ok (named M65 and not M66)

- Iris Nebula failed

- M57 failed

SEP | Local Astrometry | 1-FastSolving (no Parallel Algorithm)

- M65 ok

- Iris Nebula ok

- M57 failed

SEP | Local Astrometry | 4-ParallelSmallScale

- M65 ok

- Iris Nebula failed

- M 57 failed

rbarberac,

That was an interesting image.

I loaded it up and printed the first 2 messages you sent, but that just means the header info didn't have the scale and position info it would need to do an informed solve, it would need to do a blind solve. Then I ran some tests. The Internal SEP did not return any stars when I tried to extract them, but the external sextractor did return stars, so we can use that for solving. We can investigate why SEP didn't return stars. So for solving I tried this combination:

External Sextractor, Local Astrometry, Parallel Small Scale Profile. It solved with this combination of settings in about 10 seconds.

I will look into why it didn't get any stars in SEP.

Thanks,

Rob

Cerro Torre and others who are testing,

Thank you very much for all the testing you are doing, it is very helpful! I can provide a little info. Sorry I haven't written documentation yet. I have not had much time. Maybe I can start it right here:

For plate solving there are several parts in StellarSolver that are important:

______________________________________
Source Extraction: to find the stars in your image in order to solve. In StellarSolver, I have the option for 3 different methods:

- Internal SEP: this requires no external programs, it is the same SEP star extraction algorithm we have used in KStars for Focus and Guiding for awhile now. It is essentially a library version of the method below (though there are some differences which is why they give slightly different results). It is entirely internal to the program, so there are no files saved to disk for the extraction which is great for Raspberry Pis etc.

- External Sextractor: this does require an external program, SExtractor, or the Source Extractor. This is their official standalone program. The drawback is you would need to have sextractor installed and it does save a bunch of files to disk in order to do its operations.

- BuiltIn Sextractor: This uses whatever method of source extraction the solver uses by default. StellarSolver uses SEP, just like the Internal SEP setting. Local astrometry.net uses its own source extraction method which uses a bunch of external resources including python, netpbm and other packages. (All those external dependences caused me huge amounts of headaches years ago when I ported kstars to Mac computers) And finally ASTAP has its own internal source extractor which is pretty good.

Note: Either Internal SEP or External Sextractor should be superior to the built in version of the programs. SExtractor is REALLY good at extracting stars, and that greatly speeds up solving, but it has a LOT of options that we need to perfect.

______________________________________
The Solver: The program that will be used to do the solving of the sources that were found. In StellarSolver, I give 4 options for that

- StellarSolver: This option uses an internal library build of astrometry.net that I worked really hard this spring to produce. It uses no external files like configuration files etc, and saves no files to disk (except 0KB solved and cancel files) which is great for Raspberry Pis. Since this library is entirely internal, no programs have to be installed beyond KStars itself, so if you are going to use this option, you don't need the astrometry.net package at all. This is going to make a world of difference for Windows users who cannot install astrometry.net unless they do it in a compatibility layer.

- Local Astrometry.net: This option uses the good old fashioned local astrometry.net installation many users have used with KStars for years. The only differences in Stellarsolver are that we no longer need the configuration files, we can do parallelization to make it MUCH faster, and we can use Internal SEP or External Sextractor to give it the sources to solve. So I took the same old tool and gave it more uses.

- Local ASTAP: ASTAP was available in KStars previously, but I have implemented more options for using it in StellarSolver as well as giving you the option to use Internal SEP or SExtractor with it. The options for ASTAP are now shared with astrometry so you can just set your options in the profile and it will work fine. ASTAP does NOT support parallelization. I tried.

- Online Astrometry.net: This option was previously available in KStars as well, but I did do a bunch of work on it to make it work better, to use Internal SEP or Sextractor if you like, to use the options in the profiles, and to provide clearer feedback to the user about what is going on. Technically, online Astrometry.net is already using parallelization on their server, so I didn't implement it for this.

______________________________________
The Options Profiles

I found that the number of options for plate solving and source extraction that were available to the user was truly bewildering, but really powerful. So I have been working to consolidate the options and make the same set of options work for all the source extractors and solvers across the board. That way if you perfect a profile, you should be able to use it no matter which solver or source extractor you use. And you should be able to switch back and forth quickly and easily. I am hoping that we can perfect the profiles so that they will be helpful for users using many different setups. But I will seriously need help trying to improve them and get them to that point. To this end, I have made the options profile editor so you can use the profiles I think are good, so that you can make your own profiles, change my profiles, and so you can share good profiles back and forth. The same profile can be used for both source extraction and solving if you develop a truly good profile. I know the editor nor the profiles are perfect yet, but I think it is a far easier way to approach this than having a huge amount of options all in different places.

Here are the profiles that I have developed along with some comments. This could definitely be changed in the future.

Profiles mainly for Solving:
1-FastSolving. -- I developed this before I made parallel algorithms. I had optimize the parameters to solve images fast, but it does NOT do parallel solves.
2-ParallelSolving -- This was my first parallel profile. It can be faster than FastSolving, but does not work nearly as well as the next 2. I might delete it.
3-ParalleLargeScale -- This profile is meant to solve DSLR scale images very fast. It assumes larger image scales to solve faster than the above.
4-ParallelSmallScale -- The DEFAULT for solving. This profile is meant to solve telescopic images quickly. Most users should probably use this one.

Profiles mainly for Source Extraction in Focus and Guide
5 - AllStars -- This profile is meant to detect all the stars in an image, it is meant
6 - SmallSizedStars -- I meant this one to just detect smaller stars and ignore bigger stars
7 - MidSizedStars -- I meant this one to just detect medium sized stars
8 - BigSizedStars -- I meant this one to just detect bigger stars and ignore smaller stars

I would say that my profiles for Solving are much more refined, the Source Extraction ones need work. That being said, some have said that they had better solving with one of the ones I meant for Source Extraction. Maybe they could be the starting point for solving profiles. Hence the reason that I made the profiles universal. If you develop a really good one, you can use it for everything.

Hopefully this helps and was clear,

Rob

Cerro Torre,

Do you have this Iris Nebula image? There was clearly a setting or two in the Fast solving profile that made this one solve it and not the others. Maybe try editing the FastSolving profile and turning on Parallelization and seeing if it still solves?

rbarberac,

I think I just corrected the issue with SEP doing the source extraction on your sample image. Please update and test again. It works on the image you sent me now.

Thanks,

Rob

Hi Rob,

Now, it didn't crash, but it didn't works. I've stoped it after more that 10 minutes trying a blind solve. Then I tried to guide it adding limits to the focal length and then crashed. Also, as you have merged this branch (instead of squashing the changes into a single commit) it's almost imposible to revert the app to a point before the change. I've trapped my self into a corner : I can't revert to the oficial 3.4.3 because there is a problem with the OnStep driver and I can't go to the current nightly because the Solver is not working on the RP4 build.

rbarberac,

I thought you had ran that image in the StellarSolver tester program before and that’s what I thought you were going to test for a good comparison to see if it is fixed. Could you try that? The issue with the image you sent was that it was a float image and the SEP code in StellarSolver handled it slightly differently than the other types. I just corrected it so it does them all the same way. It solved fairly quickly.

It sounds like this time you tried it in kstars. You said it didn’t work, can you provide details of how you tried it? Did you try different ways ?

I didn’t make any kstars changes for this image, so if there was a similar or related bug somewhere in kstars I haven’t corrected it.

But I would need more details to know first if the issue is solved in StellarSolver or not for your image. And then second if there is some sort of issue in kstars, what might it be related to?

Thanks for your help,

Rob

I think the point rbarberac was getting at is we use kstars/ekos as the tool for our imaging and Stellarsolver is now the "way" in kstars so that's how we test it.

@rbarberac go back in time in git to just befor Jasem did the "big" Stellarsolver check-in and down load a zip of those files. Make a ~/Stable directory and
build/install from that if you are doing actual work. It's how I unpainted myself from the same corner as you find yourself.

Hi jackbak,

Sorry, in my post I was addressing the first part of rbarberac's post, not the second part. As for the second part, I definitely apologize if the integration of StellarSolver into the main KStars repo before it was totally ready has caused anybody any issues. I agree that we weren’t really ready to put all this in KStars master right away, especially because people rely on the master branch of the repo for nightly builds, and we needed to test it thoroughly before putting it in master. I was hoping for more of a testing time period where people could test this on my fork or maybe in a different branch in the main repo and we could have a nice discussion about how to improve it before putting it in the official KStars Master repo. But, let's not be too hard on Jasem for that, he is a great guy and works incredibly hard to support KStars and INDI and us. It has caused a little more pressure on me to fix all these problems and get it all integrated, and maybe that was a good thing, because I had mostly finished StellarSolver back in June, and I was not integrating it into KStars very fast at all. This has pretty much lit a fire under me to get it done. We already did complain to him about putting it in the main repo so quickly, but it really has worked to get things up and running. So instead of focusing on that, now that its done, let's work really hard to polish everything up and get it ready for release. Please keep testing and I will keep trying to perfect things with StellarSolver and its integration with KStars for the release. And again I am sorry if the integration has caused anybody any issues because it wasn’t ready yet.

Thanks,
Rob