Jump to content


Photo

AstroTortilla vs UniMap for plate solving

  • Please log in to reply
37 replies to this topic

#1 Moromete

Moromete

    Viking 1

  • *****
  • Posts: 684
  • Joined: 15 Feb 2012
  • Loc: Romania

Posted 18 October 2013 - 03:31 PM

Which one of these softwares is better for fast and accurate GOTOs with plate solving done with a 50mm finder guider and QHY5L-II guide cam?


I see both softwares as a free alternative to Celestron StarSense and probably with a better performance too.

I have played with both only inhouse and I like more Unimap.
Here is Unimap http://larryo.org/as...imap/index.html .
Unimap works without an Internet connection.

I haven't understood yet how to use AtroTortilla.
Here it is http://sourceforge.n...illa/home/Home/

#2 telfish

telfish

    Messenger

  • *****
  • Posts: 493
  • Joined: 17 Nov 2010
  • Loc: Adirondack Mountains NY

Posted 18 October 2013 - 03:54 PM

I use AT with my T3I, C11 and hyperstar. I love it. Once you have it set up it will solve in 20 scs and realign your scope to exactly center your target.

Not sure how it would work with the hardware you have listed.

Terry

#3 Moromete

Moromete

    Viking 1

  • *****
  • Posts: 684
  • Joined: 15 Feb 2012
  • Loc: Romania

Posted 19 October 2013 - 12:18 AM

I found a very nice tutorial for using AstroTortilla herehttp://lightvortexastronomy.blogspot.ro/2013/08/tutorial-imaging-setting-... .

#4 Moromete

Moromete

    Viking 1

  • *****
  • Posts: 684
  • Joined: 15 Feb 2012
  • Loc: Romania

Posted 19 October 2013 - 12:21 AM

Sorry, that's the correct link http://lightvortexas...setting-up-a...

Does AtroTortilla work with Sharpcap or Firecapture?

#5 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5724
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 19 October 2013 - 04:26 AM

astrotortilla will work with any software that outputs FITS, JPEG, and can even capture from the screen. internally it uses the astrometry.net blind solver, which has been proven rigorously in research to be a very robust solver.

it is not as fast as PinPoint or Elbrus but is completely blind - you don't need to provide the camera orientation, pixel scale, or even hints as to the approximate RA/DEC where it's solving.

#6 Raginar

Raginar

    Fly Me to the Moon

  • *****
  • Posts: 6138
  • Joined: 19 Oct 2010
  • Loc: Rapid CIty, SD

Posted 19 October 2013 - 09:32 AM

It's not too bad, I use server to back up my SGP. If it has a hint, it's pretty fast.

#7 Gib007

Gib007

    Sputnik

  • -----
  • Posts: 28
  • Joined: 19 Sep 2013
  • Loc: Gibraltar

Posted 19 October 2013 - 10:04 AM

Thanks for linking to my tutorial Moromete. I'm glad you find it useful! :)
AstroTortilla is fantastic software and I would gladly have paid for it.

#8 Moromete

Moromete

    Viking 1

  • *****
  • Posts: 684
  • Joined: 15 Feb 2012
  • Loc: Romania

Posted 19 October 2013 - 10:08 AM

With a 50mm F4 finder guider + QHY5L-II how accurate is Astrotortilla/Ebrus/Unimap for GOTOs when the main scope is a C11(F6)+DSLR?

I haven't played with Elbrus but Unimap looks much nicer than Astrtortilla.

Without hints both Astrotortilla and Unimap fail to solve or need very much time to solve an image, as I experimented inhouse.

#9 Moromete

Moromete

    Viking 1

  • *****
  • Posts: 684
  • Joined: 15 Feb 2012
  • Loc: Romania

Posted 19 October 2013 - 10:10 AM

Thanks for your Astrotortilla tutorial, it's very well done! :)

#10 Raginar

Raginar

    Fly Me to the Moon

  • *****
  • Posts: 6138
  • Joined: 19 Oct 2010
  • Loc: Rapid CIty, SD

Posted 19 October 2013 - 11:37 AM

Why would you not just use plate solves with your C11? There shouldn't be any issues doing that. I get my images within 50 pixels of my target easily and it's usually around 8 pixels.

The main problem you have is that by doing the aiming with a different scope that isn't perfectly aligned.... Well you can figure it out. Just try the software. Or, try SGP. It has the option for a localized astrometry server too.

#11 Tonk

Tonk

    Fly Me to the Moon

  • *****
  • Posts: 7148
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 19 October 2013 - 12:42 PM

Unimap works without an Internet connection.


Thanks for highlighting that fact. My observatory is remote and has no internet access so this sounds ideal.

#12 Vostok

Vostok

    Lift Off

  • -----
  • Posts: 13
  • Joined: 23 Mar 2008
  • Loc: Finland

Posted 19 October 2013 - 01:40 PM

AstroTortilla runs without an internet connection too. :)

#13 telfish

telfish

    Messenger

  • *****
  • Posts: 493
  • Joined: 17 Nov 2010
  • Loc: Adirondack Mountains NY

Posted 20 October 2013 - 07:56 AM

Why would you not just use plate solves with your C11? There shouldn't be any issues doing that. I get my images within 50 pixels of my target easily and it's usually around 8 pixels.


The main problem you have is that by doing the aiming with a different scope that isn't perfectly aligned.... Well you can figure it out. Just try the software. Or, try SGP. It has the option for a localized astrometry server too.


The only reason I can think of is for visual use. If you wish to use it as a means to get really accurate goto's then use the main scope.

It's an interesting idea.

#14 Phillip Easton

Phillip Easton

    Ranger 4

  • -----
  • Posts: 323
  • Joined: 24 Dec 2010
  • Loc: DFW

Posted 20 October 2013 - 10:27 AM

Unimap doesn't appear to be in development anymore. That download file is pretty old. I had trouble getting it to solve sometimes.

In my experience AT is much better plus if has mount integration. With AT connected to the mount (AT gets hints for RA and DEC) and the right parameters I can get a plate solve in under 10s. Then it will re-sync the scope.

My next attempt will be to use the databases it installs with SGP.

Cheers!

#15 Coastal

Coastal

    Mariner 2

  • -----
  • Posts: 207
  • Joined: 23 Apr 2009
  • Loc: Vancouver island, Canada

Posted 20 October 2013 - 11:57 AM

astrotortilla will work with any software that outputs FITS, JPEG, and can even capture from the screen. internally it uses the astrometry.net blind solver, which has been proven rigorously in research to be a very robust solver.

it is not as fast as PinPoint or Elbrus but is completely blind - you don't need to provide the camera orientation, pixel scale, or even hints as to the approximate RA/DEC where it's solving.


AT not as fast as Pinpoint? On my system it totally smokes Pinpoint and Elbrus. If you are doing every solve as a blind search with no pixel scale or centre coordinates then indeed it is slower. With the right command line arguments to the astrometry.net engine you'll be impressed with the speed. A SSD is a recommended location for the index files. Better yet, if you have a lot of ram, having them in a ram disk is fantastic! I never have to wait for solves now.

#16 Wmacky

Wmacky

    Surveyor 1

  • *****
  • Posts: 1933
  • Joined: 24 Nov 2007
  • Loc: Florida

Posted 20 October 2013 - 01:30 PM

That tutorial came at just the right time as this weekend was for getting AT setup. Is sync to scope only for EQMOD? I'm using a CGE mount with Stellarium.

#17 Coastal

Coastal

    Mariner 2

  • -----
  • Posts: 207
  • Joined: 23 Apr 2009
  • Loc: Vancouver island, Canada

Posted 20 October 2013 - 01:40 PM

Sync to scope should work with any ASCOM driver that supports sync (I seem to recall a few drivers not supporting sync, possibly early Ioptron ones).

Sean

#18 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5724
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 20 October 2013 - 07:39 PM

Sean, if you can pass on your settings this would be much appreciated.

On my system solves take about 20-30 seconds. It is getting RA/DEC hints, and constrained to a 30-degree area around the reported RA/DEC. I have also pruned down the indexes.

#19 Coastal

Coastal

    Mariner 2

  • -----
  • Posts: 207
  • Joined: 23 Apr 2009
  • Loc: Vancouver island, Canada

Posted 20 October 2013 - 08:02 PM

Sean, if you can pass on your settings this would be much appreciated.

On my system solves take about 20-30 seconds. It is getting RA/DEC hints, and constrained to a 30-degree area around the reported RA/DEC. I have also pruned down the indexes.


What is your imaging scale, and do you have that properly set in AT? Have you tried reducing the search radius as well? The site listed in this thread has a lot of details that can really help. I tend to omit the -sigma 1 as my ccd image is pretty clean. Anyway if you save a fits file and for example place it in your cygwin home directory you can play around with the arguments (use --overwrite) to see what works best with your setup. I might be seeing a slight boost because I compiled a 64 bit recent snapshot of astrometry.net versus the much older one that is included in the AT package.

Sean

#20 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5724
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 21 October 2013 - 12:42 AM

Yup as I mentioned I limit the search radius to 30 degrees. I have set a lower and upper bound for the image scale that matches my sensor (my sensor is about 2x1.5 degrees and I limit to 1-2 degrees).

It's possible that a large fraction of my solve time is due to file I/O. My fits are 35mb. But that camera is gone. Guess I'll try Binning with my new camera... The old one was an OSC and didn't bin well.

#21 anat

anat

    Viking 1

  • -----
  • Posts: 717
  • Joined: 03 Jun 2004

Posted 21 October 2013 - 01:25 AM

I use the Astrometry.net engine in Linux for solving images with plate sizes of about 1 deg or slightly less. It takes less than 5 sec on PC. And it takes about 15 sec or less on Arm Linux devices.

I always use downsample = 2 or 3. And I put the minimal number of index files having the largest sky diameter down to the sky diameter of about a half of the camera coverage (30 arcsec). If you put all index files in the index folder, the computation will be slow even though scale-low is specified. The engine still tries to use some of index files with small sky diameters.

#22 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5724
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 21 October 2013 - 03:11 AM

hi Anat,

yup i have also pruned my indexes. What i noticed was even with downsampling the solve time was still high. i think downsampling itself consumes CPU so it would be better to bin.

#23 anat

anat

    Viking 1

  • -----
  • Posts: 717
  • Joined: 03 Jun 2004

Posted 21 October 2013 - 03:29 AM

If you convert your original image file to a compressed format (JPG) first, it may help. Plate-solving doesn't need a big raw file :)

#24 Phillip Easton

Phillip Easton

    Ranger 4

  • -----
  • Posts: 323
  • Joined: 24 Dec 2010
  • Loc: DFW

Posted 21 October 2013 - 07:42 AM

hi Anat,

yup i have also pruned my indexes. What i noticed was even with downsampling the solve time was still high. i think downsampling itself consumes CPU so it would be better to bin.


Have you tried adjusting sigma? I set mine to around 7 and get around 20-40 stars to search. Much quicker than if you are searching on several hundred.

Cheers!

#25 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5724
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 21 October 2013 - 08:38 AM

i just checked and i was at sigma 30! :tonofbricks:

scale minimum 0, maximum 200 (arc-minutes). well i'll try again with tighter limits.






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics