great work on the new solver! I built from git source master branch on ubuntu 20.04 and ran tests
on
- regular subs captured in kstars
- blind solves on sril stacked and processed versions of same targets (32 bit float .fit)
- blind solves of siril stacks further processed in gimp (.jpg)
I can't get the star circle overlays to appear but FWIW attached are my test results.
As expected, blind solves take a bit longer but none stumped the tester. Looking forward
to having this available in SM OS release ... very fast solves on all my subs !!!
Images taken with WO GT71 with FlatA6III FR/FF (336 mm FL) and ASI385MC-cool camera (3.75 um pix size).
It would be cool to be able to click a row in the star table and have that star highlighted either with a
glyph or have the annotation details pop up beside it.

Hi Edgar,

You are 100% correct, now that we have stellarsolver, astrometry.net is no longer absolutely required (though it is still nice to have) so there is no need to check or require the astrometry folders or config file. In fact, even for using external astrometry.net, stellarsolver can create its own config file for the solver to use and so doesn't need astrometry.cfg. I am sorry we haven't gotten to this yet, since we were making sure that there were not crashes and that images were properly being solved first, but we really need to tackle this soon.

Thanks for the reminder,

Rob

dean,

Thank you for testing!

If you use the Sextract function in the tester, the stars will get circled and the star list will appear, and if you solve, you will get all the RA and DEC values appended to the star list. If you just solve and don't sextract separately, it just solves the image. So you can use the sextract feature either before or afterward and then it will show the star list and circles.

Rob

just as you described ... works perfectly thank-you.

Ok, git pull of both StellarSolver and Kstars/Ekos 20 minutes ago. Rpi4 4G astroberry. The StellarSolver Tester works very well.

Ekos, Align module crashes consistently right after capturing / loading the image. Internal SEP | Internal Solver | 4 - ParallelSmallScale

It would be useful if I could find any setting combination (Internal Solver or ASTAP or Astrometry) that could solve. Right now I get crashes or no solutions.

Thanks.

Progress !

Since I have all the external index files and I just noticed the "Load all indexes in Memory" is checked in Options Profile Editor.
I am forcing Ekos to run out of memory -- Crash!

Now that it works on the bench we must see it work in the field.

But Yay, progress.

So we were doing some testing on a pi last night and found that there were 2 things that seemed to really be causing the internal solver to crash on the pi, the load all indexes option and not downsampling enough. We found that downsampling by 4 along with not loading all the indexes worked great. We are thinks a default profile specific to the pi is a good idea.

Totally agree. Thanks for the downsample 4 tip I'll try it - as with just 1;1 it is hit or miss (mostly miss).

Downsample 4 = much better.

It would seem that there is a use for 8Gb of Ram on the new Pi4 after all! If only to load all indexes into memory.

Wouldn't you need a 64-bit OS to get to 8Gb ? I'm waiting for a stable 64-bit for the Rpi4.

Ubuntu 20 makes full use of the 8GB in the new Pi4.