issue with nova astrometry when solving a distorted image

[Davide Nitti]

Dear,

I am newbie in plate solving. I just tried to upload an image with a bit of lens distortion.

the solving was successful but the annotated image seems a bit off. the stars are not centered between the green lines. this means that the distortion prevent to have a good solving or that it's just this annotation that is wrong?

is there any documentation for newbies?

thanks!

[Dustin Lang]

Hi,

Could you please send a link to the image?

We do try to solve for distortion, but we don't always get it right!

One thing you can try when you are uploading an image is, under the "Advanced settings", check the "CRPIX center" box -- which will put the center point of the distortion solution in the center of the image.  You can also play with the "Tweak" (distortion) order.

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/e6f41ac1-7c16-49dc-9f06-a57b0bb7d35dn%40googlegroups.com.

[Davide Nitti]

ok thanks! I'll try
this is the image: https://nova.astrometry.net/annotated_full/4711341

do you also know what are those red and green circles? https://nova.astrometry.net/user_images/4145447#redgreen

thanks you very much

[Dustin Lang]

Okay, yeah, for super wide-fields like this you'll probably want higher "tweak order".  The default is a 2nd-order polynomial, which just isn't that much freedom to fit distortion.  (Though honestly, our distortion fitting is a weak area.)

The green circles are stars from the reference catalog (probably Tycho-2 for wide-fields like this), and reds are stars detected in your image.

cheers,

--dustin

[Davide Nitti]

thanks,

I can try to fix the distortion in another way or make smaller field pictures,

my goal is to make a panorama stitching multiple images together or to stack images and I need to register them.

if the solving was successful what does it mean actually? because I need a subpixel registration precision.

how do I know which stars matched and the error? (preferably with python or a linux command line)

thank you!

Davide

[Dustin Lang]

Our solving is a three-step process -- first we match a 3- or 4-star mini-constellation, then predict where others stars would be found and test that prediction to make sure it's correct, then finally try to solve the distortion by iteratively matching up more stars and trying to solve for the distortion.

There's a "corr" file -- "correspondences" -- that lists stars we think matched.

To view this discussion on the web visit https://groups.google.com/d/msgid/astrometry/44a2952c-b300-41e6-9b0e-4ca26a289529n%40googlegroups.com.