Unfortunately, AIJ will not currently work with the local astrometry.net
server (AIJ is hard coded to look for the nova server at
astrometry.net). However, I was not aware that this package is
available, and I don't know enough yet to say if I can adapt AIJ to work
with it. However, this capability would be a very good addition to AIJ,
so I will follow up and see if it is possible. Unfortunately I am
swamped right now, so it will be a couple of weeks before I can look at
it in depth. Thanks for pointing out the local server package!
I have added an option to enter a custom/local server name in the
astrometry setup window. Let me know how your testing goes. If you find
a solution that works, I would appreciate it if you could list any
details of the setup here.
One thing to note is that AIJ does not send the image to the
astrometry.net server, only a list of X,Y coordinates. Hopefully the
local server supports the X,Y coordinate list mode like nova.astrometry.net.
I believe the local server address should default to the proper value
for the default ansvr settings (http://127.0.0.1:8080).
Note: do not include /api since that is added internally by AIJ.
Thanks to the added option and Andy Galasso's instant help when I told him about my unsuccessful initial tests, AstroImageJ is now able to use a local astrometry.net server through ansvr v0.17! This is great! (Thank you Andy for adding the capability to process x,y lists!)
I noticed that AstroImageJ has a hard coded server time-out of 30s. If the local server does not respond within 30s with a solution as it did not finish a solve yet (i.e. slow machine, or a difficult narrow field), AstroImageJ resubmits the image, probably assuming the server was not responding in the first place. Although 30s is often enough to solve a field when you know about the properties of the image at hand, it would be good if you could add another option that allows the user to set the server time-out by himself, based on typical solve times on the respective machine.
The timeout happens because ansvr uses a slightly different processing
sequence than nova. I'll not go into those details since they are pretty
much hidden from the user, but the latest daily build now allows up to
10 minutes during the part of the handshake where ansvr does the
solving. I can consider adding an option to allow the user to set the
timeout if the 10 minutes doesn't work for the general case.
By the way, my simple field solves were taking about a minute. I just
found out with Andy's help that if I remove all the index files except
the ones from ~20% to ~150% of my typical FOVs, the same solves now take
only 5 seconds!
You need to actually move/remove the unused index files from:
And yes, a BIG thanks to Andy Galasso for quickly making the ansvr
changes to support the AIJ interface!
Thanks Karen! Locating apertures via RA, Dec just got easier on big data sets. :-)
Your comment about removing index files with improper scale is interesting, but to my experience, defining a range for the plate scale at hand just has the same effect, as this limits the search to the respective index files?
Either the plate scale range is not working (I tried that first), or ansvr loads all the index files before it makes note of the limited plate scale. The files are so huge, I expect it would take a long time to load them even from cache. We can try to figure
this out over time, but for now the easy fix is to remove the index files you don't need.
I just have installed ansvr according to the instructions given at http://adgsoftware.com/ansvr/ and updated my AstroImageJ (AIJ) to the latest daily build. All the settings look fine. Unfortunately, the plate solving doesn't work. The ansvr log stops at "serve notfound uri =". Nothing more happens. Do you have any ideas why it fails?
Your wrote that you tested ansvr v0.17 successfully with AIJ. However, the latest version fo ansvr seems to be v0.16. Is this a typo or do you have indeed a newer version? If the latter is the case: Where did you get it from?