Unable to solve locally

[bernd kopi]

Hi all,

I thought I give a local installation a try. I've an Ubuntu system running and installed astrometry.net via sudo apt install astrometry.net

Then I downloaded some index files following the recommendations here http://astrometry.net/doc/readme.html#getting-index-files.

I solved an original uncropped image of my setup at nova.astrometry.net which gave me

 - Size: 58.1 x 38.7 arcmin
 - Radius: 0.582 deg
 - Pixel scale: 0.67 arcsec/pixel

As this fits perfectly to the example of the 1-degree images, I downloaded the fits 4205, 4206 and 4207 and put them in the default folder at /usr/share/astrometry.

After enabling the inparallel option in the /etc/astrometry.cfg I tried solving my first image.

$ solve-field --overwrite --scale-units arcsecperpix --scale-low 0.6 --scale-high 0.7 m104-stretched.jpg

What I got was

Reading input file 1 of 1: "m104-stretched.jpg"...

jpegtopnm: WRITING PPM FILE

Read file stdin: 5202 x 3464 pixels x 1 color(s); maxval 255

Using 8-bit output

Extracting sources...

simplexy: found 150 sources.

Solving...

Reading file "./m104-stretched.axy"...

CPU time limit reached!

Field 1 did not solve (index index-4207-11.fits).

Field: m104-stretched.jpg

Did not solve (or no WCS file was written).

I checked the config which is set to

# Maximum CPU time to spend on a field, in seconds:

# default is 600 (ten minutes), which is probably way overkill.

cpulimit 300

Have I forgotten anything? Do I need other/more index files?

And I noticed that even although inparallel is set, only one CPU is working. Is this correct?

Thank you,

Bernd

[Dustin Lang]

Hi,

Your setup looks fine.  Could you check on the nova site and see which index file it used to get your solution?

The web service uses "--downsample 2" by default, so you could try that to match.  It's often useful to check the plots, if you've got the plotting code going.  The *-objs.png file will show you the stars we found.

cheers,

--dustin

--
You received this message because you are subscribed to the Google Groups "astrometry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to astrometry+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/astrometry/2a98bdbf-1fa6-4948-9f36-f22cb48ecc78n%40googlegroups.com.

[bernd kopi]

Hello dustin,

sorry for the late answer, I always got an error trying to post it in the last couple of days.

Only after removing the *.fits attachments I'm able to post :/

thanks for the downsample param :) Adding this one solved the image in seconds. Tried it with another one, that didn't finish with my initial params. But after I dropped the scale params it worked and also did on a third image.

If I read that correctly, nova used index id 208 while mine used 4206.

Unfortunately, this lead to some new questions :/

When I solved it at nova, it came up with wcs-nova.fits (http://nova.astrometry.net/user_images/4742416)

but solving the same picture on my local installation gave me wcs-local.fits

The values don't match :/ Is this normal? Is this because of the different index files?

And while the commandline says

RA,Dec = (189.985,-11.6217), pixel scale 0.670113 arcsec/pix.

I can't find that RA or Dec value in any of the generated files.

Further, I do get these two lines while processing

Creating index object overlay plot...

Creating annotation plot...

While the -indx.png has stars marked and a green box, the -objs.png has marked stars as well but the -ngc (which should be the annotated one I think) is just the original image without any overlay.

On nova.astrometry.net there is a list of objects in the frame. Does the local installation provide this information as well in some form?

Thank you, Bernd

[bernd kopi]

Since I couldn't attach the fits files, here is a small excerpt.

From nova:

CRVAL1 = 190.044192699 / RA  of reference point

CRVAL2 = -11.6179231264 / DEC of reference point

CRPIX1 = 2396.98947144 / X reference pixel

CRPIX2 = 1494.30242157 / Y reference pixel

COMMENT stars: 114663,114786,114666,114657

From local:

CRVAL1 = 190.121456043 / RA  of reference point

CRVAL2 = -11.556529838 / DEC of reference point

CRPIX1 = 2344.92910767 / X reference pixel

CRPIX2 = 973.739250183 / Y reference pixel

COMMENT stars: 95364,95388,95366,95398

[Dustin Lang]

Hi,

If you add the --crpix-center command-line argument to the local version, and check the "CRPIX in center" option on nova, it will make the center pixel of your image the reference pixel.  (In the WCS standard, the CRPIX are pixel values and CRVAL are RA,Dec values, so CRPIX 1,2 maps to CRVAL 1,2).

You can try using the plotann.py program to get different kinds of annotations on your images.  For the *-ngc.png plot, different layers are turned on depending on the image scale, so sometimes there is just nothing to show.

cheers,

--dustin

--
You received this message because you are subscribed to the Google Groups "astrometry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to astrometry+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/astrometry/5192a275-02d7-4052-9dc3-913d52a66992n%40googlegroups.com.