Plate solve hint problem with ANSVR


#1

Hi, Ken,

I am practicing plate solve at home to get myself more familiar with it.
I am using ANSVR as solver. PinPoint seems to work only with FITS with a bunch of headers that provide some extra info.
I loaded a JPG into SGPro.
I found that putting, say, no matter what I put in the RA/Dec hint, it had no impact on solving and I checked the ANSVR log. The json request does not contain any ra/dec information at all
The log entry looked the same however i change the ra/dec input.
I tried to put M42 (which is not the object but for the sake of testing shouldn’t matter, right?) into the object field. It also had no impact on the request sent to ANSVR. In none of the cases, the request json contained center_ra, center_dec, or radius parameters. Same goes for FITS files that have good information in the header. The all looked same as the following.

[2016-06-26 17:14:24] POST /api/upload/ HTTP/1.1
[2016-06-26 17:14:24] request_json is: {“session”:“79”,“allow_commercial_use”:“d”,“allow_modifications”:“d”,“publicly_visible”:“y”,“scale_units”:“arcsecperpix”,“scale_type”:“ev”,“scale_est”:4.9416666666666664,“scale_err”:5.0}
[2016-06-26 17:14:24] UPLOAD: session is 79
[2016-06-26 17:14:24] upload type is: FITS
[2016-06-26 17:14:25] POST /api/submissions/79 HTTP/1.1
[2016-06-26 17:14:24] forcing scale_err to 5 (was 5)
scale_args: -u arcsecperpix -L 4.69458333333333 -H 5.18875
UPLOAD: /opt/ansvr/run_solver 480 /usr/bin/solve-field -p -O -U none -B none -R none -M none -N none -C cancel --crpix-center -z 2 -u arcsecperpix -L 4.69458333333333 -H 5.18875 stars.fit
Using timeout = 480
exec /usr/bin/solve-field -p -O -U none -B none -R none -M none -N none -C cancel --crpix-center -z 2 -u arcsecperpix -L 4.69458333333333 -H 5.18875 stars.fit
Reading input file 1 of 1: “stars.fit”…


#2

ANSVR is a “blind” solver, meaning that it does not need nor accept RA/DEC hints. It essentially solves the reference image against the whole sky.

I highly recommend that you try the free PlateSolve 2. It does solve using RA/DEC hints, and is very, very fast - when it works. If your hints are close, it works very well.

Tim


#3

Actually ANSVR is based on astrometry.net, and it does accept RA/DEC hint and search radius if you provide them (center_ra, center_dec, and radius)

astrometry.net is very fast when it works too. However my current experience is that the downsampling parameter plays a substantial role in success of solving and speed of solving (between 1 second success and not solvable). I think the star extraction algorithm might have some dependence on how bloated the stars are.

I have compared with PlateSolve2 a while ago. It was hit and miss between the two.

Anyway, if it is not too hard, I’d prefer to see the astrometry.net hint used by SGPro too. But now that I know that the hint is not sent, I wouldn’t try to put them in until they are supported.


www.mainsequencesoftware.com