Problems plate solving

[cuivenion]

Hi, I took one picture of the failed solve fortunately, and it does work at astrometry.net.

http://nova.astrometry.net/user_images/ ... #annotated

Although the sensor is small (GPCAM v1), it was being used with a 135mm lens so it had a 2.09 x 1.57 degree field of view.

[admin]

That's cool - do you have a UI for your local installed version that you can run that frame through too?

cheers,

Robin

[cuivenion]

Sorry do you mean the original .fits capture.

[admin]

Yes,

not sure about AnSvr, but ASPS and AstroTortilla both have interactive UI applications you can use to solve an already saved frame. In AstroTortilla you set the 'Camera' to be the 'File Open Dialog' and then press 'Capture and Solve' and you get to browse for your fits file. Can't remember how for ASPS and don't have that installed right now.

Basically we now know that the captured frame is solvable, but need to check that your locally installed solver is OK too (on its own without SharpCap) - if you see the same message from that then there's something fishy going on in the plate solver. The next version of SharpCap 3.1 will have the extra logging to capture the error message which may help too.

cheers,

Robin

[cuivenion]

I did a snapshot at the same exposure and gain I was trying to solve with, this is it attached. Let me know if thats not what you meant.

[cuivenion]

This is local solve vs online solve.

011.jpg (312.76 KiB) Viewed 2619 times

[admin]

Ok, so your local server looks to be in good shape too

Grab the next build (out later this evening) and retry - hopefully we will see the error in the logs. You can use the 'Test Camera 1' in SharpCap to load up your existing FITS file, set the exposure to 1s (or it might be dimmer/brighter than the original file) and then plate solve - no need to actually get the telescope out or wait for a clear night.

cheers,

Robin

[cuivenion]

Will do, thanks Robin.

[cuivenion]

Hi, Installed the new version. I wasn't sure how to to use my image with the test camera, let me know if this is right. I converted it to png and renamed it m42_dim the same as your sample file and then replaced the sample file with it.

That seemed to work and my image came up and I attempted to plate solve. It took a couple of minutes but it did solve. However when I checked the log the solve co-ordinates, pixel scale and field size are very different from the online and ASPS solve.

[admin]

Ok, I see what is going on with the different solutions - the last log that you sent shows that the plate solve has come upon a relatively poor match for the field at

Code: Select all

RA,Dec = (234.596,6.41545), pixel scale 7.38199 arcsec/pix

.

The way the plate solver behind AstroTortilla, ASPS etc works is that it accepts the first match that meets it's "good enough" threshold and this one does, although not by a big margin (

Code: Select all

log-odds ratio 29.6165 (7.28237e+12), 6 match, 0 conflict, 14 distractors, 28 index

). The other solution from Astrometry.Net and ASPS UI is massively better in terms of quality of match (

Code: Select all

log-odds ratio 459.569 (3.87485e+199), 98 match, 0 conflict, 78 distractors, 122 index

), but never gets found because the weak match gets found first.

Now, I can get the correct plate solve to come out of SharpCap by changing the 'Sigma' parameter in the plate solve settings - this parameter changes how sensitive the plate solver is to stars in the image - the higher it is, the brighter a star has to be to get used. Set this parameter to 10 instead of 50 and then the right results come out. Finding the right value of this parameter is a bit of a balancing act - set it too low and hot pixels or noise get detected as stars and the results are rubbish. Set it too high and you don't get enough stars to plate solve at all. The optimum value also will change somewhat for different exposures, gains, etc. The value of 50 that SharpCap uses by default seems to work fairly well in most cases.

Anyway, different positions explained, but still no real idea about why it failed when tested under the stars - looks like the updated logging is working though so hopefully if it breaks again we will see all the info needed in the SharpCap log.

cheers,

Robin