Can't get Astrometry local to plate solve?


#1

I installed SGP last night and all is working well.

I want to have a plate solving function that doesnt need an internet connection.

I installed local astrometry.net solver from here - https://adgsoftware.com/ansvr/ and followed the instructions to the letter. Including telling SGP which plate solving option to use.

I installed the data files to cover 20% of my narrowest view (8’) to the maximum 1º

If I open an image that I want to add more data to and right click and choose plate solve - blind solve it works away but always reaches about 6 minutes before reporting image cannot be solved.

I have tried several different images - same result.

I have opened ‘watch ansvr log’ and data is whizzing up the screen.

Any ideas ???


#2

If you go into the settings menu on blind solve and click that does another screen open up?


#3

Have you tried Platesolve2? I’ve used it for a long time from within SGP and very satisfied with it, if the position of the FOV and the actual coordinates aren’t too far it solves very fast, i have rarely had a problem with it not solving and in that case there has always been a large difference between the search area and actual FOV.


#4

Hello Sara

When I right click on the image and choose Blind Solve it immediately goes to astrometry.net and tries to solve. I have also tried Pinpoint and Platesolve2 (I have installed the catalogues and pointed the programme at them etc but SGP just wont look at anything except astrometry.net !!

I have noticed that sometimes if I point the solver from within Control Panel it doesn’t show up in Equipment Profile manager and vice versa - I must admit I don’t understand the relationship between these two ??

Thanks in advance.


#5

@Skipper, could you send me your ansvr log file and a FITS image that you are attempting to solve, and please tell me your image scale (arc-seconds per pixel). You can send them to andy.galasso@gmail.com. The ansvr log file is in C:\Users\YOUR_USERNAME\AppData\Local\cygwin_ansvr\var\tmp\platesolve\ansvr-DATE-TIME.log
Andy


#6

Will do - many thanks - its driving me mad !!!


#7

Andy - logs and image sent just now.

David (aka Skipper Billy)


#8

Profile manager is where you save settings for later loading to make use when making a new sequence, control panel is where you can edit the settings for your current sequence, if you save the sequence for later use your control panel settings will be loaded with it.
Getting control of how this works is crucial to be able to use SGP fully :slight_smile:

Also to be able to get a succesfull plate solve you need to input the correct scale and pixels under camera> settings and specs.

I hope this helps in your use of SGP :slight_smile:


#9

Following up from our offline conversation in case anybody else is searching for a similar solution.

There were two settings that needed to be changed to get ansvr working:

  1. The expected image scale was off by more than the default 5% tolerance. A good way to get your accurate image scale is to upload and solve an image with http://nova.astrometry.net/upload Astrometry.net will give the image scale which you then enter into the SGP Control Panel, Camera tab.

  2. Update the ansvr max objects setting to 100:


#10

I thought that Ansvr didn’t care about plate scale, it’s usually how I determine the scale.


#11

Hi Chris,

Good point. Sometimes it does and sometimes it doesn’t. Here’s how it works:

  • if SGP has a scale available (entered in the control panel or entered manually in the Plate Solve window), it will pass the scale to ansvr.
  • when ansvr receives a scale hint from SGP it will set lower and upper bounds on the scale using the “Scale error estimate” ansvr setting and pass those bounds to the solver.
  • when ansvr receives no scale hint from SGP it will attempt to solve using all installed indexes which can be very slow

If the scale is passed along but the actual image scale is outside the bounds, then the solve will fail. If the scale is passed and is correct (within the bounds) the solve will be faster (often significantly faster) than when no scale is passed.

Hope that makes sense…
Andy


#12

Thanks Andy, that makes sense.

I’m tinkering with a mount control program that will control the mount motors directly, handling all the slewing and mount modelling. I’m acquiring an image and plate solving it to get alignment data using PS2 and/or Ansvr as the plate solver. So far I’m not providing any hint data to ansvr but I’ll try adding that except for blind solves.

Would providing a scale with a large scale error estimate give a rapid solve if the scale is correct and a slow but successful solve if it isn’t? I’m using either a 16mm lens or an 80mm scope giving scales of 47 or 2.4 arc secs per pixel.


#13

If the scale from SGP is passed to ANSVR but it is outside the bounds, wouldn’t it make more sense to then revert to a true blind solve instead of just failing? That way it will still solve even if it does take a lot longer to solve.


#14

A large error estimate would result in searching more indexes, and the speed depends on how many indexes are searched, especially the smaller scale indexes which are huge and do not fit in RAM.

It’s a relatively easy one-time operation to get your true image scale, and once you know that scale there’s no benefit of allowing a large error tolerance… it would only slow it down.


#15

That would be up to SGP if it wanted to try again without the scale hint. But ordinarily SGP does know the image scale and the failure is not likely to be due to having the wrong image scale. In other words, it would still fail if you removed the hint, it would just take longer to fail.


www.mainsequencesoftware.com