Image Solver script won't solve?

[Josh Lake]

Here at NEAIC, we've been showing off the great functionality of Image Solver and Annotate Image to generate sky maps and add WCS headers to FITS. I've used it very successfully on a large number of my images, but when I went to run it on some Astrobin data, it failed no matter what I changed.

The image is this shot of M101: http://www.astrobin.com/18295/  I just pulled in the large JPG, but I have access to the final FITS as well.

I set the focal length to 2048 mm and the pixel size to 6.8, both from the equipment detail on the Astrobin page. From there, I've tried every setting of star sensitivity, searching, binning... and it always fails to Solve.

Any idea which parameter could be wrong? I've run IS on plenty of final stretched JPGs with success, so I'm not sure what's going on with this one.

[oldwexi]

Hi Jake!
The challenge with this image is that it  is to much cropped or resampled.
If you use the original data of 2048mm focal length you dont get there because the Resolution than is
0.6 arsec per pixel.

So you have to change the Resolution to 0.343 arcsec per Pixel.
This gets you with the 4288 Pixel of the Image to a fake focal length of
3606mm

So i used in ImageSolver:
M101
Resolution 0.343 arcsec/Pixel
Pixel Size 6my
Linit Magnitude 16
Star sensitive 0.54
Iterations 4

this worked very fast.
http://www.werbeagentur.org/oldwexi/fotos/m101_anno.jpg
With such foreign images i first measure the real width of the Image in arcseconds
and divide by the Pixels the image has - this gets the correct arcsec resolution
Good luck

Gerald

[Josh Lake]

Ah, very smart Gerald, thank you! This makes much more sense, and while I had a feeling that the cropping would affect the solving, I did not realize it would be that sensitive. I'm very much looking forward to the Blind Solver script that is in the works.

I'm glad you responded to this thread because I wanted to tell you that my presentation in general, and my description of PixelMath functions in particular, were very well received at the NEAIC conference. I demonstrated your 'image profile' code and people were impressed at the novel use of some more advanced PM functions beyond just Max() and basic frame blending. Thanks again for your permission to use it, it spiced up my talk!

[Andres.Pozo]

ImageSolver works much better with linear images. The first step of the process tries to match a synthetic star field with the image to solve. This is done using StarAlignment. This process works really well for the purpose for it was created. However it has problems with nonlinear images and when trying to register images with different resolution.

[Josh Lake]

That makes sense. It's such a great script, and when combined with Annotate Image, it got a lot of impressed comments at NEAIC.

[wnourse]

I was one of the ones so impressed!  So I guess, from Andres' comments, it would make sense to try to solve the image early in the workflow, to embed the WCS into the FITS header, then use the annotate script at the end?

Josh, thanks for doing a couple of great sessions!

Will

[Josh Lake]

I was one of the ones so impressed!  So I guess, from Andres' comments, it would make sense to try to solve the image early in the workflow, to embed the WCS into the FITS header, then use the annotate script at the end?

Josh, thanks for doing a couple of great sessions!

My pleasure, I really enjoyed the whole Pro-PI spirit at the conference this year. Between Jon and I doing 'official' PI talks and Yuri, Joe, and Jon Davis prominently using PI in their talks, I think it was a win for team Pleiades this year in general.

And yes, when the raw data is available, it's certainly the way to go for Image Solver. Joe and I were just trying to figure out what to do with already-processed data with no linear data available. Measuring the frame for the real arcsec/pixel ratio is the best way. But I'm hoping that a new Blind image solver will eventually make PI that much more powerful.

[Walcho]

ImageSolver works much better with linear images. The first step of the process tries to match a synthetic star field with the image to solve. This is done using StarAlignment. This process works really well for the purpose for it was created. However it has problems with nonlinear images and when trying to register images with different resolution.

I was confronted with the problem and was ready to ask the question.
But the answer is there! Thank you for this really useful information.