DSLR Astrometry Upload Job Keeps Failing

[Ram Hari]

Hi everyone,

Complete newbie here, I've been using Astrometry to help me identify stars in my DSLR astrophotography. I have been trying to identify the stars in this picture (see attached) and have uploaded a cropped version of the night sky I want to ID. However, I haven't been getting much success. I would appreciate any help/tips. Please see attached and feel free to try running it.

Thank you,

Ram

[Bryan]

Ram

This looks like a very wide field, esp if cropped.  What index files did you use?

Bryan

[Ram Hari]

Hi Bryan,

I just uploaded the .jpg image onto http://nova.astrometry.net/ 
I did adjust the advanced settings as Wide Field and Very Wide Field but the processing still failed. I am not familiar with index files so I didn't use any. 

Ram

[Bryan]

Ram

I am thinking that this image has a lot of hot pixels, esp since it is a DSLR image.  Hence, AN cannot solve it because it cannot find a recognizable pattern of stars.  There are a couple approaches to getting a solution, but require more effort.  Perhaps Dustin has a simpler solution.

1.  Take a shorter exposure for fewer stars.  I use a local install of AN.  Your image gets almost 3000 'stars' in its identification pass, which happens before it tries any pattern matching.  AN can solve with less than 100.  When I am plate-solving DSLR images, I use 2-4 s exposures at ISO 1600.  This will allow AN to plate-solve and annotate the brighter stars without being confused by hot pixels.  Obviously, this approach requires starting over.

2.  Eliminate hot pixels by using a dark calibration frame.  You could use this on the present image but it is not a simple process.  Cover the DSLR and take 10-20 exposures of the same length as the original.   If you have a processing package that you like, stack these into a master dark frame.  Subtract the dark frame from the original exposure and solve the result.  If you have the original raw file, that would be better to work with than a JPG.  Canon raw files are CR2 or CR3, Nikon are NEF.  Depending upon the processing package, you may have to convert these to TIF or FIT format first.

Bryan

[Ram Hari]

Thank you for the tips Bryan, I will explore these other approaches the next time I shoot!

Ram

[Bryan]

Ram

 Since, by definition, you are starting from scratch "the next time," I would try the short exposure route first. The only reason for the second approach is to offer a means to solve the existing image.  Going through darks, stacking, etc just to plate solve is burdensome and, based on my experience, unnecessary.

Bryan

[Dustin Lang]

Hi,

I'm actually pretty surprised this doesn't work!  It didn't look to me like there were single-pixel hot pixels, and when I set downsample=4, the detected stars look good.  From the EXIF headers, you have a 15mm lens.  Normally these show a lot of distortion, and sometimes that causes trouble for us.  The usual suggestion there is to pull out a region close to the middle of the image and run that, but I'm trying that and still not getting a solution!  As suggested, you may want to try submitting a shorter exposure, just to see if you can get a solution.  Usually we need 20 or so stars to get a solution.  Having more isn't a problem, but when they're saturated they become harder to centroid correctly.

The other thing I was thinking was the possibility that the pixels are rectangular (we assume they're square), but looking at the specs for this camera, the pixels are supposed to be square.

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/804928e1-1e6d-4925-b432-5d808dd33b67n%40googlegroups.com.

[Ram Hari]

Thank you for the tips Dustin! Yeah, it was a 25 second exposure on a 15mm lens at f/2.4 at ISO 1600.  Maybe a shorter exposure will do the trick.