17 February 2022 at 11:46 pm #575166
I ve been following tutorials on the web, and notes for exoplanet and variable star projects. Only one reference mentions the use of ansvr, the Astrometry.net local server, The notes are here on the BAA. I decided i would rather use this than the well known on-line solver. My problem is that no plate solution is found for my images with AIJ. Here is what i have done, and this may prove helpful as a summary of the operation thus far.
Downloaded and installed ansvr local server (ansvr-setup-0.22) and ran the index downloader. I installed the indexes from 0.1 up to 7 degrees. The ansvr program and batch files are in All programs. I put the indexes in a folder on the C drive (this is optional).
Start AIJ and drop an image onto the tool bar. This opens and WCS is then selected from the menu bar. Use “plate solving with options”. Select the check box to use Custom Server http://127.0.0.1:8080. Also check the box to auto save and the box to show the result log.
Now go into All Programs and select “start ansvr”. Go back to WCS and click plate solve. AIJ finds the stars and puts a blue ring around them. The default number of stars is 50. Ive use 70 and 100, but no change.
I have not been able to input the field size and orientation ( I thought the solver was blind)
So the solver doesnt seem to work and the solved file is not written – any suggestions please as what i should try next?
One other observation if i load a file with WCS positions then to solver logs the fact, so i know the file is being read by AIJ. Im sure there has to be a setting somewhere that ive not found.18 February 2022 at 7:45 am #585241
Tim, I use ANSVR on my TCS in La Palma. Most of the time it works extremely well but can be a bit finicky to set up. As chance happens I’m flying out to La Palma later today, in part to get the observatory cleaned of volcanic ash. So – no time to give a considered answer but I will try to help you out in a day or two if you still need assistance.18 February 2022 at 12:14 pm #585243
Thanks Xilman. i believe i have the answer now. The secret is on the AIJ WCS settings page, where there is a tick box to constrain the search area to RA-Dec and the radius. Turn this off. Also uncheck the tick box to specify the pixel size. One should be able to specify this exactly, but turning it off works. I also entered the correct RA-DEC centre and constrained the search. The solves is lot quicker, as one might suppose. Hopefully im now in a position to solve any field in AIJ
Another hint is to start “Watch ansvr.log”. This runs in a cmd window. From this i was able to see the progress of the solve. Im sure you’r already up to speed on this, but i note it here for anyone else ascending the same incline !
Cheers…18 February 2022 at 3:23 pm #585244
Astrometry.net is a blind solver, but it runs much faster if you give it a guess at the field centre RA/Dec or have an idea how big your field of view is.
I used to use ansvr, but I realised that the version of astometry.net implemented is quite out of date compared to the Linux version and was sometimes failing to find a solution when a friend’s more recent Linux version succeeded. However, when I had a problem the ansvr author was quite helpful – as were the guys who wrote astrometry.net.
Could you test ansvr from a command line?
If you are able to use Python – or similar – its worth installing the W10 WSL2 option and running a Linux astometry.net session from W10. Its how I now work – the start up time for WSL2 sessions is quite small (will be even better in W11).19 February 2022 at 10:47 am #585248
This is what i have found Grant. The local server version of ANSVR on my PC is controlled by AIJ. It can be blind, or restricted to the sky area of interest. I’ve not used it much so far but seems pretty fast when constrained to one area.
Im hoping to use AIJ for asteroid photometry, but i dont know if it can be used successful for moving objects. Perhaps someone has been down this route and can advise me ?
Thanks for the tip about Linux on W10. I will investigate this.19 February 2022 at 1:08 pm #585251
I was finding astrometry.net could take up to 30s on a frame if it had no idea of your field size nor where it was pointing.
However, for a 2700×2200 sensor its taking about 8 seconds per frame (using a Celeron 2 core 2.1GHz) if it knows the approx field location and size. But I was supplying it with coordinates for up to 2000 stars – looking for anything vaguely star shaped with an SNR>10. With a Lodestar that dropped to 2.5s.
A lot of pieces of software don’t seem to consider a slowly moving target.
Silly question, would it be possible to edit the index files so that (say) all declinations further south than -20 were not included. That would improve the search time for blind searches. Has anyone done that?19 February 2022 at 5:47 pm #585252
Not a silly question at all. It has been recommended on the astronomy.net mailing list, including by the author.
Just don’t store any index files which contain areas of the sky which you will never want to solve. Saves disk space too!20 February 2022 at 2:01 pm #585253
That sounds hopeful. I shall have to read round that a bit. Presumably, the files are each for a declination swathe then?
I had forgotten there was an astometry.net group. Thanks for the reminder.
EDIT: I was wrong. Its not as simple as I had envisaged. These links appear to show the declination versus index file relationship…
Realistically, looking at the layout, a better speed increase could be achieved by running multiple instances of astometry.net simultaneously, so I shall take that route instead. Theres no point having multi-core/threaded machines if you don’t redline them occasionally. 🙂20 February 2022 at 3:42 pm #585254
Realistically, looking at the layout, a better speed increase could be achieved by running multiple instances of astometry.net simultaneously, so I shall take that route instead.
As well as, not instead of. You can still flush the index files for useless healpixes and save space and some time.
- You must be logged in to reply to this topic.