Looking for beta testers of new Polar Alignment utility.

[perdrix]

Where do I send the files please?

Dave

[perdrix]

It looks like something went "odd" since I first installed PPA. Possibly related to a Windows 10 install.

Where's the tutorial please?  I was hunting for it and failed.

Dave

[shamantanthew]

This software is great! I had a chance to test it out last week in the field but my local solves weren't working fast enough. I need help getting AstroTortilla to do local solves faster! The Nova website is fairly fast, around 10 seconds each, but I won't always have internet access out in the field. Sometimes AstroTortilla won't solve at all depending on my settings.

Vertical: http://img.photobucket.com/albums/v245/Nathan49456/IMG_3891.jpg

Horizontal: http://img.photobucket.com/albums/v245/Nathan49456/IMG_3890.jpg

I'm using a Canon 6D with a 400 mm f/5.6 lens which makes for 3.45 arcsec per pixel. I'm using the following AstroTortilla settings and indices:

Astrometry.net uses the 4112 and 4113 indices to solve these two images, so I when I move all the others out of the folder and force AT to just use these two indices it still fails to solve!

 

Edited April 3, 2016 by shamantanthew

[themos]

Hello Nathan (is that right?)

I always find that an initial solve using Nova provides a reliable solve and gives me an image scale I can use to fine tune local solves. I sometimes use Ron Wodaski's CCD calculator to get an estimate of image scale but people sometimes use different resolutions on their DSLRs. In this case, the CCD calculator gives a very much different image scale than a Nova solve. Nova gives me 18.44 arcseconds per pixel but the CCD calculator says more like 4. The difference is that you have posted reduced resolution JPGs (1024x683). So, try a local solve (through AstroTortilla if you like) with image scale limits 17.5 to 19.5 arcseconds per pixel and then use PPA to do the same and you might get this.

(added in edit: the Polar Alignment error is about 10 arcminutes)

 

 

Edited April 3, 2016 by themos

[shamantanthew]

Thanks Themos! I wonder if you using the smaller resolution files is what's causing our different results.

Here's a link to the full-size files:

vertical: https://onedrive.live.com/redir?resid=2ECF5A87B0942D0D!14831&authkey=!AFDpiv7s9DvGKbc&v=3&ithint=photo%2cJPG

horizontal: https://onedrive.live.com/redir?resid=2ECF5A87B0942D0D!14832&authkey=!AJ6-B4kkii2EkgU&v=3&ithint=photo%2cJPG

And PPA gives me the following polar alignment error (using the full size files and Nova to solve):

[shamantanthew]

I may have stumbled across the "golden setting," so to speak... downscaling. I set downsclaing to 2 instead of 0 and now it solves in 15 seconds. Gonna keep playing with this.

[themos]

6 hours ago, shamantanthew said:

I may have stumbled across the "golden setting," so to speak... downscaling. I set downsclaing to 2 instead of 0 and now it solves in 15 seconds. Gonna keep playing with this.

The only other significant setting I use (apart from downscale=2) is the "configfile" in PPA's settings. I've made an astrometry.net config that searches only the 4100 series indices. For your high-res images, the limits can be  -L 3.29 -H 3.63, and  I then get a local solve, showing 10 arcminutes error again.

[daz]

Themos - just landed here from the AlignMaster question....

Do I understand that provided I can get my local astrometry server to solve an image from my DSLR, PPA should be fine?

[themos]

That's right, Daz!

 

[daz]

OK cool. Off to grab it now.

And I need Anaconda?

Will that break my existing Python 2.7 install? Not overly worried, as I can just reconfigure my Eclipse IDE to point to the Anaconda installation if it does...

[themos]

If you already have a 2.7 python installed, it might be easier for you to install the necessary python packages instead of getting the Anaconda bundle. The probably complete list is ujson, astropy, PIL, numpy, scipy

 

 

[daz]

Ok have installed Anaconda and the PPA software.

Have registered with Nova so I have an API key. I also configured the local server settings as well...

I selected the V image and hit Nova but it is just sat there saying 'solving image'. I went onto Nova and uploaded the image and it solved in a few seconds...

Have I missed something? 

 

Apologies for brevity. Am on the phone 4g cos my broadband is playing up!

[daz]

OK, more testing....

I have run PPA.py from the cmd line to see what the output looks like, and it looks like I have some issues!!

When I try and locally I get an error about not being able to load tables from the index files - so I need to check into that..

When I try and solve via Nova, I get an error than ujson cannot be found by TKinter... I might do a re-install of Anaconda just to be sure....

 

 

 

[shamantanthew]

57 minutes ago, daz said:

OK, more testing....

I have run PPA.py from the cmd line to see what the output looks like, and it looks like I have some issues!!

When I try and locally I get an error about not being able to load tables from the index files - so I need to check into that..

When I try and solve via Nova, I get an error than ujson cannot be found by TKinter... I might do a re-install of Anaconda just to be sure....

 

 

 

Hey Daz, I ran into the same problem with Nova and the ujson error. This occurred despite me installing the Anaconda 2.7 version (I think I installed x64 and not x86, not sure if it matters though). I fixed this by following Themos' instructions on page 4 of this thread: open a cmd window, change your directory to your python installation, e.g.  "cd c:\users\mathe\anaconda2" in my case, then use the command "pip install ujson"... In my case it said I had to update pip so I followed the commands to do that, then I re-entered the command pip install ujson and it installed ujson at that point. Restarted ppa.py and now Nova works.

[daz]

Ah...... I didn't get that far back in the thread

I had errors with the 4200 and 4201 index files - so I removed them and that error went away! I could then solve the test images

However, with my own images I could not get a solve. In the end I uploaded to Nova and solved them, then downloaded the WCS files and finally got a solution. I need to play a lot more and see what the right settings are for the DSLR/Lens (I was using my 600D/18-55 lens and tried at 18mm and somewhere near to 30mm). Even with the solved file on Nova so I could get the image scale, I still could not get a local solve to work...

Still, progress

[themos]

Thanks for the user support! It's been a while since I installed it and I had forgotten all the hoops. I'll make a todo-list now. So, use the "pip install stuff" to bring in the needed python modules.

[themos]

daz, I woud stick to the long focal length end of that range, 55mm should be fine. I've downloaded the 4100 series of indices, myself. Make sure the downscale option is set to 2. There's a bunch of extra flags for turning extra file generation (and annotations) off. There's  screenshot in the thread somewhere...I am not on my own laptop at the moment so can't paste my settings. 

[themos]

Ok, just finished a PPA session, the extra settings are -p -B none -M none  -P none -R none -U none -N none -O and I use the 4100 series, my image scale is  8 arcsecs per pixel. My reported error is under 10 arcSECONDS in both alt/az.

[daz]

Thanks Themos

I found the post with those settings, and had the down sampling set to 2.

I need to go and check the Nova site again to see what my calculated image scale was - I've just thought that I might have made an error in the unit of measurement!

[themos]

been there, done that. I always stick to arcseconds per pixel!

 

[daz]

2 minutes ago, themos said:

been there, done that. I always stick to arcseconds per pixel!

 

[daz]

Sorry Themos - work and weather conspire against me!

Saturday looks like the next opportunity for me to test, so in the meantime, I'll Python updated with ujson and get the Nova plate solving working...

One thing I meant to say - with the ujson error, the app just sits there with the 'solving image' message, rather than registering the error and reporting back...

[themos]

1 hour ago, daz said:

Sorry Themos - work and weather conspire against me!

Saturday looks like the next opportunity for me to test, so in the meantime, I'll Python updated with ujson and get the Nova plate solving working...

One thing I meant to say - with the ujson error, the app just sits there with the 'solving image' message, rather than registering the error and reporting back...

I will do something about that in the next release, as it's pretty frustrating!

 

[daz]

13 hours ago, themos said:

I will do something about that in the next release, as it's pretty frustrating!

 

The weather?

Work?

 

Oh wait..... 

[shamantanthew]

Forecast calls for a clear but breezy night... Looking forward to giving PPA another shot now that I have local solves working!