Local Astrometry.net + solve-field arguments



I have a question regarding the way SGPro interfaces to the local Astrometry.net, specifically the image scale passed to the solve-field utility.

In the Target Settings/Reference Data frame I ‘Browse’ to an image I took of M45 with my QSI690, 1x1 binning. RA/Dec hints are correctly shown, with image scale shown as 1.692, as extracted from the FITS headers.

I now click ‘Solve’ and view the ansvr log file.

The ansvr log shows that solve-field is called with an estimated pixel scale of 4.77143 rather than 1.692. I’ve left ansvr scale error blank in the ansvr config tool so that it defaults to 100% which gives rise to ‘-L 0 -H 9.5428’:

[2015-01-04 23:03:32] POST /api/upload/ HTTP/1.1
[2015-01-04 23:03:32] request_json is: {“session”:“87”,“allow_commercial_use”:“d
[2015-01-04 23:03:32] UPLOAD: session is 87
[2015-01-04 23:03:33] POST /api/submissions/87 HTTP/1.1
scale_args: -u arcsecperpix -L 0 -H 9.54286666666667
UPLOAD: /opt/ansvr/run_solver 600 /usr/bin/solve-field -p -O -U none -B none -R
none -M none -N none -C cancel --crpix-center --sigma 70 -r --objs 100 -u arcs
ecperpix -L 0 -H 9.54286666666667 stars.fit
Using timeout = 600
exec /usr/bin/solve-field -p -O -U none -B none -R none -M none -N none -C cance
l --crpix-center --sigma 70 -r --objs 100 -u arcsecperpix -L 0 -H 9.542866666666
67 stars.fit

So, shouldn’t SGPro be sending the actual pixel scale to ansvr rather than an estimate?

Thanks for any advice!



SGP scales down the image before sending it to astrometry.net (local or remote). The scale estimate passed to ansvr corresponds to the scale of the downsampled image.




Is the SGP scaling factor you refer to related to the ‘Downsample factor’ in the ansvr settings? I guess not, since for my setup I leave this blank i.e. no downsampling.




No, the ansvr setting tells astrometry.net to do an additional downsampling after the image is received by ansvr. The downsampling by SGP is autonomously done by SGP before the image is transmitted to ansvr/astrometry.net.

An example: let’s say your SGP settings indicate your scale is approximately 1.7 arc-sec per pixel. You take an image and SGP internally rescales the raw image to (for example) 4.25 arc-sec/pixel (scaling factor 2.5), then transmits the smaller image to astrometry.net, providing scale the hint “4.25 arc-sec per pixel”. The solver may rescale it again if you use the downsample option in ansvr. Let’s say the solver returns the actual scale as 4.4 arc-sec per pixel to SGP. SGP divides by the known scaling factor (2.5 in this example) to get the scale of the original raw image, 1.76 arc-sec/pixel.




Great explanation, thanks for your help!