This morning I tested the astrometry function with simulators.
I tracked the HD198149 star and try to solve field. The process doesn't converge to a solution. Is it because of the simulators or the process itself ?
I have to mention that I had, sometimes, same behaviors in real situation in some regions of the sky.

The mount model tool did not change the behavior of the Ekos astrometric solver and astrometry.net, it just makes use of them. The simulators and astrometry often solve very well when everything is set up right. I usually use the offline solver and my images solve within a few seconds. The exception is near the poles, things do not solve well at all there. Anywhere else in the sky, it should solve pretty quickly. It just may take a little adjustment.

I have found that a number of factors can affect whether an image gets solved or not. First, are you using the online or offline solver? If you are using the offline one, then you need to have the right index files and need to have the right programs properly installed, if you are using the online solver, you must have a connection to the internet and their servers must be working properly. Second, did you get the right number of stars? You can increase or decrease the exposure time or bin your images more or less to change this. I will usually just play with these two settings when a field doesnt solve. Third, do you have all of the settings correct for the solver such as the telescope equipment, the fits2fits parameter, the coordinates, the search radius, etc. Sometimes, you can just remove or change a parameter that is causing problems and it solves right away.

Hopefully this helps. Thanks

Rob

Haven't had clear weather to try this on real setup yet, but with simulators I noticed that sometimes when running the modeling it starts a new image capture immediately after starting slew to next star. I guess it sees the mount as idle even though it has just issued a slew command. Maybe it should first wait until the mount status changes to busy and then back to idle, though it might be a simulator only issue as well.

Hello Robert,
I assume that it is not your tool that is to incriminate. Maybe I would have to open a new thread for this topic.
For the test I evoke I just used simulators. And all looks fine, I have the good focal and aperture.
But for this object HD198149 the solver never converge and doesn't move the simulator mount. The object is 61° DE which is far from the North Pole.
I tried to do the same for Alderamin in Cepheus and the result was identical.
I have set the telescope focal and aperture to respectively 777 and 111mm. The camera simulator is 1280x960 pixels of 3,75µm. And all the index files covering the computed field (41.3' x 30.9') are presents (index-4208.fits 2Mass and index-4108.fits Tycho2).

Hum, maybe I am silly or Alzheimer. I guess the CCD simulator cannot change the view of the targeted field. So, the process can enter in infinite loop never finding the good solution. If I am right, I am sorry, this topic is not a good one.
But now I understand the philosophy of your tool and YES the automatic mount model is a great improvement and a really good use of astrometry. I hope to test it very soon. Maybe tonight if the weather remain clement.

From your screenshot, it looks like it did solve the image, but you have told it to "slew to target" and you have the accuracy required to be 15. I think its a simulator issue. I have found that the simulator always has a certain amount of error and a sync does not correct it. I would recommend requiring 30 arcsec instead or setting the align module l to sync only and not "slew to target" so that it can actually complete if you want to test the modeling tool with the simulators. The error in the simulator is nice for testing purposes, but you cannot expect its accuracy to improve like a real mount with a pointing model would (unless the simulator is rewritten).

It shouldn't start the capture until it arrives at the star. do you know the circumstances under which this happens or is it random?

Tested it an hour ago. It works fine on 3 stars in Leo. But as for the first iterations the error is huge the colored target becomes very very tiny. So we have to rescale and pan. It would be better to rescale the target while the points are drawn. Maybe it could exists better solutions.

As far as I see it seems to be random, probably a race condition, but quite easy to reproduce.

The target size is determined by the requested accuracy in the solver at the moment, which you can change whenever you like.

The target appears very tiny because you are very far off target. You can delete any points that you don't want to see anymore.

patdut,

here is a demonstration of what I meant about the built in error in the telescope simulator, a series of 20 solutions spread across the sky using the simulator and the mount modeling tool using a wizard generated grid pattern. I just ran this a few minutes ago. Interestingly, the pattern is slightly different than one I ran the other day. and one point is even outside of my circle, which is interesting too. Since I made this tool, I have run several variations of patterns like this with the simulator and it produces interesting patterns. What I don't actually know is whether the error I am seeing is error in the simulators, error in the astrometry.net files, or if it is a periodic error simulation in INDI CCD, which Jasem suggested it may be.

here's a couple from two days ago