Predistortion; SIP coefficients

[Toni Šarić]

Hi Dustin, everyone.

Brief recap of the problem I am trying to solve.
My images are partially obstructed, and I am trying to determine the sky coordinates of the one position that is, unfortunately on that obstructed part of the image. Since it is at the end of the image, distortion effects are pretty strong.
I still get average accuracy of starfield solving of around 5 arcsec, and my scale is 18.56 arcsecperpix, but the solution at the end of the image is bad.
Therefore, I moved my camera to see stars with the full FoV(not exactly full, but 90%) to correctly determine distortion coefficients.

I solved the image with the tweak order = 3 and obtain SIP coefficients. The average accuracy on the 500 stars is 3arcsec.

After that, I tried to solve the same image but using the previously obtained wcs solution and predistort option, and no tweak order, since position should be predistorted.
But, the solution gets even worse, by much.
Now, my accuracy is 45 arcsec.
I attached both .corr files (_tw3 is first solving, _predistort is after using predistort option) and also solved images(-indx).
It clearly can be seen that index stars(green circles) moved further from the sources.
Additionally, I tried to predistort it manually by first solving starfield with the augment option, then manually correcting the xy position of a source using obtained SIP coefficients, and finally solving it using those corrected source positions. And now the average accuracy is 23arcsec.

Does anyone know what seems to be a problem?

BR

Toni

[Toni Šarić]

Hi Dustin,

as you suggested for predistortion, I have tried a new version of Astrometry, 0.93.

However, I am not able to succesfully run it. These are the steps I performed:

- Run solve field with tw=3 on a file in "Solve5" directory to obtain SIP coefficients. Save that solved fits file and .wcs in different directory(named sip5) .

- Run solve field on a new fits file (the same image as before, again in directory "Solve5") with predistort option and giving it .wcs or .fits from the "sip5" directory and tweak order=0(since now we are moving field stars, index stars should remain where they were).

But I got following error:

solve-field.c:565:after_solved Failed to read WCS header from file Solve5/1669432234.wcs sip_qfits.c:272:read_header_file Failed to parse WCS header from file "Solve5/1669432234.wcs" extension 0 sip_qfits.c:380:sip_read_header SIP: failed to read TAN header sip_qfits.c:496:tan_read_header TAN header: missing or invalid value for "CRVAL1"

And I get empty .corr file. Verbose astrometry output is attached(tw0_afterpredistort.txt).

I don't understand why it is searching for header in "Solve5/1669432234.wcs".

I also tried to run with predistort and tw=3, although this should be wrong, and in my opinion, this will correct for the distortion twice(once because of the predistort option - correcting field stars, and second time because of the tweak order by correcting index stars). This way I get the results, but now the precision is even worse.

Cheers,

Toni

[Toni Šarić]

I managed to run solve field with predistort by applying tweak_order=0 instead of no-tweak. But still it seems to give incorrect solution, as output is telling me that tw=2 is applied with predistortion(check snippet below).

I also attached image of output from three .corr files for the same star,

- tw3.corr is normal solving with tw=3,

- tw0.corr also normal solving with tw=0,

- tw0_after_predistort is solving with predistort option and tw=0.

I hope attached image will better explain my concern.

Cheers

Toni