The shaded area on mine is the non viewable area.
I had another go and it all works. I had previously done exactly the process that you use and kept getting the error. Now it works!
I also have had this exact problem in the past. I have 1 machine where it works well (Odroid running ubuntu) but I have failed to get it to work on Windows PC running KStars. Like you, the invalid polygon error message pops up despite the data points being exactly the same as the one that works. I gave up on it 6 months ago so have not checked again if it has been fixed. It would also be nice to have some way to transfer the point or the model between machines to save the time taken to add each point manually.
Hopefully someone who has cracked this will reply
Try pushing up your exposure time to about 3 secs (0.5 is a bit fast), increase your step size to at least 60-100. If you use 30 the autofocus routine may not be able to get a reliable change in focus to indicate which direction best focus lies. This can change due to poor seeing and all sorts of other things. Using 60-100 just tells the autofocus to use that for its first step. After that it will move back or forward depending on the change in focus dependant on the algorithm used. Generally it will start of with largish steps and get smaller as it gets closer to best focus. Occasionally it gets confused and does that massive move, loses the star and then goes back to the starting point with a focus failed. I think that this is more likely if you use a small first step size.
I have used SEP with some success (although back using gradient at the moment) and I find polynomial is a little quicker. Use 1000 for max travel as Rob has suggested and 3% tolerance although I don't think that this is critical.
Hope this helps.
I have just checked the nightly build for RPi and reinstalled KStars for Windows and everything now works as intended.
Wow that was a quick reply Rob,
I was just scribbling my resolution of the problem and posted it when I saw your reply. Thanks.
I will check on the nightly build.
Ok I was a bit premature with this post. I had inadvertantly changed the API and forgot the to put the http in front as per http://127.0.0.1:8080. That stopped the unknown protocol bit.
The next thing was to click the tick box for internal astrometry.cfg
Now it will solve a loaded FITS!
The corrupted or missing astrometry.cfg pop-up still comes up but this does not seem to matter.
It may be worth stopping this pop-up in the Windows version in this situation as it is a bit confusing.
Anyway all sorted
Is anyone else having this problem?
I occasionally use KStars for windows with the my RPi running remotely at the scope. Previously (before the latest KStars for Windows upgrade 3.3.6) the ANSVR worked well as the "on-line"astrometry solver. The offline solver on the Pi also works well running remotely from the Windows version so this is not a major problem.
Since the upgrade whenever I start up KStars for windows I get this pop-up. (See attached screenshot).
There are some new options in the alignment section but they appear to be more related to KStars on the Pi rather than Windows. I have tried many different combinations and am getting nowhere.
Is there supposed to be a Windows based config file that I paste in the location in the Astrometry.net panel in the config section ? I am unsure of what to do with the new tick box next to config in this situation. (This is in the Windows EKOS options panels)
Also when I try a trial solve with Load and Slew of a previous FITS file I get a "unknown protocol" flag in the debug section of the solver.
I have tried uninstalling and reinstalling the ANSVR program and this has not helped.
Any advice gratefully received
My guiding used to look just like yours, probably worse. What mount are you using and are you using the internal guider?
The internal guider is fairly simple and no where near as configurable as PhD2. It works well for well behaved mounts but less useful for unruly ones. My stars were affected by this type of guiding so I had to find an alternative. For me the change to PhD2 dramatically changed the distribution of my guiding plot so that it was more even and stars improved. If you are using an NEQ6Pro with EQMod I would recommend the Rowan belt mod as it improved my guiding out of sight. PhD2 does take a bit of setting up if the default settings do not work well but it worked well for me.