Jump to content


Photo

AstroTortilla 0.4 - easy to install!

  • Please log in to reply
40 replies to this topic

#1 Vostok

Vostok

    Lift Off

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

Posted 22 January 2013 - 07:31 PM

Hi everyone,

I thought you'd like to know that we've just released AstroTortilla 0.4 at http://astrotortilla.sf.net/

If you've heard about AT before, you might remember all the questions on how to get the whole complicated thing working with that Cygwin environment and having to beg for database files via email. Consider all those problems gone - the new installer does everything for you automatically. Also all the known bugs are gone too.

If you haven't heard of AT, here's a few points:

What you need: Computer control of
- An ASCOM mount
- A camera with MaximDL, Nebulosity, APT or almost any other software

What you get:
-GoTo accuracy within the arcsecond
-Blind, automatic GoTo calibration
-Automatic, fast polar alignment error measurement.

Feel free to try it out.

Lauri Kangas
and the rest of the AT team here in Finland

Edit: Added URL.

#2 JoseBorrero

JoseBorrero

    Gemini

  • *****
  • Posts: 3144
  • Joined: 04 Sep 2009
  • Loc: Puerto Rico Island

Posted 22 January 2013 - 09:05 PM

Not to far ago APT developer Ivo announce it. Thanks

I do have maxim and pinpoint, but I also have APT and I wish to learn how to install it and run it with APT.

Thanks and Welcome to CN

#3 ldesign1

ldesign1

    Viking 1

  • *****
  • Posts: 616
  • Joined: 17 Sep 2009
  • Loc: Northern Illinois

Posted 24 January 2013 - 04:01 AM

I'm one of those who had to request the database files. I had downloaded about 20GB of data one file at a time which I did over the course of several hours. I started to install the new version this evening before heading to work and this is much easier. I noticed that there are much more in the database than I thought. I set it to download everything and watched it go for a while before leaving it to run while I was gone. By the time I get home for work, it should all be done. Now it should work with all my combinations of telescopes, focal reducers, and barlow lens. I can't wait to give it a try at 5000mm focal length. :)

#4 Vostok

Vostok

    Lift Off

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

Posted 24 January 2013 - 07:22 AM

Ha ha. Nice solution for anyone with no bandwidth concerns. ;)

Using all the indices might in some cases slow down the solution searching a bit. But it's relatively easy to have all index files elsewhere on disk and only keep those needed for a specific setup at the astrometry directory in Cygwin.

We'll try to put up a guide describing the steps to optimize for as fast solutions as possible. But first we have to figure them out..!

Lauri

#5 ldesign1

ldesign1

    Viking 1

  • *****
  • Posts: 616
  • Joined: 17 Sep 2009
  • Loc: Northern Illinois

Posted 24 January 2013 - 11:58 PM

Hey Lauri:

I noticed that the index files all start in the 4000 range. The ones that I had downloaded from astrometry were in the 200 range. Are there any difference between the two sets of downloads or are they duplicates with different names?

#6 Vostok

Vostok

    Lift Off

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

Posted 25 January 2013 - 09:37 AM

The 200 series are generated from the USNO catalog and are those that you had to ask the permission to download and use and are restricted to pass forward to anyone.

4000 series indices are generated from the 2MASS catalog and are licensed GPL ("free"). That's why we can distribute those along with the program.

They should work identically with 2XX corresponding to 40XX. If you already have those 200-series indices matching your setup, you don't have to get the new ones. Just make sure to place the 200-series files unpacked at /usr/share/astrometry/data in the cygwin hierarchy.

#7 LoveChina61

LoveChina61

    Viking 1

  • *****
  • Posts: 925
  • Joined: 19 Jun 2009

Posted 25 January 2013 - 11:36 AM

Is this updated version fairly straightforward to install and use? In another thread last year you (or one of development partners) had said that they would only recommend that experienced users and (presumably) those with previous Cygwin experience should use the AstroTortilla software. That left me with less than complete confidence although I would absolutely love to use it if there is no chance whatsoever that it will mess up my computer system.

I have no experience whatsoever with Cygwin and am not even sure what it is.

#8 ldesign1

ldesign1

    Viking 1

  • *****
  • Posts: 616
  • Joined: 17 Sep 2009
  • Loc: Northern Illinois

Posted 25 January 2013 - 04:34 PM

LoveChina61:

From my experience, if you are only interested in using AstroTortilla, then you don't have to worry about knowing what goes on with Cygwin. It runs in the background. This new version of AstroTortilla installs everything for you. As for Cygwin, I only use it if I want to plate solve my images with the annotations. I should not mess up your computer. It will either solve the image or tell you that it could not solve the image. If it takes too long, I just force quit the application and everything continues to work fine. You do need to know some command lines to get it to work though. It runs in a Windows command prompt window. That's probably why they recommend that you know what you are doing. But like I said, it will either solve or not solve.

#9 ldesign1

ldesign1

    Viking 1

  • *****
  • Posts: 616
  • Joined: 17 Sep 2009
  • Loc: Northern Illinois

Posted 25 January 2013 - 04:44 PM

Vostok:

Thanks for that information. I ran a test with both versions of indices separately as well as together. I used an image of M42 @ 400mm f/5. I covers approximately 2x3 degrees.

Below is how long it took for AstroTortilla to solve it.

1: 200-series USNO Catalog = 42 seconds

2: 4000-series 2MASS Catalog (UNKNOWN. After 15 minutes, I stopped it.)

3: Both catalogs loaded together = 405 seconds

#10 Vostok

Vostok

    Lift Off

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

Posted 25 January 2013 - 06:47 PM

Is this updated version fairly straightforward to install and use? In another thread last year you (or one of development partners) had said that they would only recommend that experienced users and (presumably) those with previous Cygwin experience should use the AstroTortilla software.


Previously the installation was ridiculously hard. It was possible to do by precisely following instructions but the process involved manually compiling the astrometry.net inside the linux-like cygwin emulator via the command line with really frightening and complex commands etc. Even if you followed instructions, it was very easy to do something wrong if you had know idea what you were doing.

But as mentioned, that's something in the past. The new installer does everything for you automatically and actually you'll never even see Cygwin if you don't want to (for those of you who do, its command line is available just as before).

So if you use an ASCOM controlled mount and a supported camera capture software, go ahead and try it.

Ralph, I've seen reports of the 4000-series indices not working as well as the old ones elsewhere too. Let's se if we find out whether this happens alot.

Lauri

#11 LoveChina61

LoveChina61

    Viking 1

  • *****
  • Posts: 925
  • Joined: 19 Jun 2009

Posted 26 January 2013 - 01:58 AM

Ralph, does it come with a list of the precise command line statement(s) that I must use to plate solve?

That's basically all I want to do as my observatory scope loses its way occasionally. If I just need to type out one command statement at the command line prompt and then press my keyboard's Return key, then I should be able to handle that :)

#12 Vostok

Vostok

    Lift Off

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

Posted 26 January 2013 - 04:51 AM

Ralph, does it come with a list of the precise command line statement(s) that I must use to plate solve?


You don't need the command line to do anything. You can either connect Tortilla to a camera to make an exposure or load the latest exposure manually. The frame is then plate solved and it's coordinates, FOV and rotational angle are shown.

If you're connected to the telescope, you can add a sync point to where you're pointing. You can also make the scope slew back to the correct place you tried to slew to before you got lost.

If you for some reason DO want to use the command line, you can do that. Open up the Cygwin command line and enter "solve-field exposure.jpg". You don't need AstroTortilla for this, but getting Cygwin and solve-field (the astrometry.net command) to work is by far easiest to accomplish using the AsrtoTortilla installer (which by the way has the option to install only Cygwin and astrometry.net but not AstroTortilla itself).

The astrometry.net command line stuff however is not described in the user guide anymore since there's zero need for the command line in AsrtoTotrtilla itself.

Lauri

#13 Aimo

Aimo

    Lift Off

  • -----
  • Posts: 18
  • Joined: 28 Jan 2010
  • Loc: Finland

Posted 26 January 2013 - 05:53 AM

Is this updated version fairly straightforward to install and use? In another thread last year you (or one of development partners) had said that they would only recommend that experienced users and (presumably) those with previous Cygwin experience should use the AstroTortilla software. That left me with less than complete confidence although I would absolutely love to use it if there is no chance whatsoever that it will mess up my computer system.


The current version handles all the system setup, you only need to provide four sets of input if the defaults are not suitable:
  • Where to install AstroTortilla (default C:\program files\astrotortilla),
  • where to install Cygwin (default c:\cygwin) and
  • Where to put Cygwin package cache (default c:\temp\cygcache\, deleting it only slows down upgrades), and
  • which index files to download (default is a rather wide-angle set). This is the only thing you need to figure out yourself.

Other than that, all you need to do is click Next. Optimizing the performance for your setup can be done in day light thru the AstroTortilla GUI: find the right combination for binning, downscaling, exposure time, and the mysterious --sigma N option and running a solve with the FileOpen camera selected. Use a few 2-5 second exposures as test material. Based on experience from a variety of users the fastest solve times seem to come when the log shows 200-300 stars detected from the exposure. I haven't tried the --objs N or --depth N yet to see if cutting the number after detection yields any improvements. Restricting the area of search to around 30 degrees of scope position also gains a significant improvement for scope syncing solves.

If you've downloaded the 200-series index files, you do not need the new ones.

If you use a single, fixed setup, you can set the "scale refinement" to 0.1 for a 10% margin for field size after the first blind solve. For multiple setups simply reset the scale related settings to large enough values, optimize the other parameters and save settings as a new file. Currently it still saves all settings, but you can open the saved config file e.g. into notepad and remove all the settings you don't want to change when loading the setup-specific file. I typically leave in only the exposure time, camera binning and all solver parameters, and when swithing to a different scope I simply load the pre-optimized settings leaving all the other AT parameters as they were.

When uninstalling, AstroTortilla removes itself and tells where you originally installed Cygwin, so you can delete it if it's not needed by any other software. AT uses one registry entry which is fully removed by the uninstall, Cygwin only uses disk space only with no changes to the computer setup. The user specific AT config file is left your application settings directory under astrotortilla.sf.net/AstroTortilla/ and it takes about 1-2kB. I don't want to mess up my computer and as creating a new release package means running the install-uninstall cycle several times to verify all the combinations, I do take care on what's done and where.

I have no experience whatsoever with Cygwin and am not even sure what it is.


As for Cygwin, it has a good write-up in Wikipedia. In short it's an environment for POSIX (i.e. Unix and Linux) software to be compiled against in order to run on Windows. The plate-solver by astrometry.net uses several traditional Unix utilities and is written for the POSIX APIs, so the easiest way to make it run on Windows is to use the Cygwin environment.

If you don't want know anything else about Cygwin, there's no need to. It's prone to breaking when upgrading or adding new features, but we take care of running the recovery commands automatically when upgraded/installed/updated/refreshed thru the AT installer.

Ralph, does it come with a list of the precise command line statement(s) that I must use to plate solve?

That's basically all I want to do as my observatory scope loses its way occasionally. If I just need to type out one command statement at the command line prompt and then press my keyboard's Return key, then I should be able to handle that


AT doesn't have a command line interface for the time being, only a simple point-and-click interface. As soon as the current version reaches sufficient maturity, I'll start writing the AT2 with a better overall architecture enabling a couple of features I'm desperately missing in the current one, which are difficult to fit in the current codebase. As for schedule, this is a free, fully open-source hobby project helping another hobby a bit, so no promises.

-Antti

#14 LoveChina61

LoveChina61

    Viking 1

  • *****
  • Posts: 925
  • Joined: 19 Jun 2009

Posted 26 January 2013 - 08:32 AM

Can it work on Windows Vista? Also, where can I download the 200-series index files? I only readily saw access to the 4000-series files on the AstroTortilla website.

Thanks!

#15 Aimo

Aimo

    Lift Off

  • -----
  • Posts: 18
  • Joined: 28 Jan 2010
  • Loc: Finland

Posted 26 January 2013 - 03:50 PM

Can it work on Windows Vista? Also, where can I download the 200-series index files? I only readily saw access to the 4000-series files on the AstroTortilla website.
Thanks!


It has been tested on Xp and Win7, so I don't see a reason it wouldn't work on Vista. The 200-series indexes are still covered by the same restrictions and access instructions as before, the 2MASS based 4000-series indexes are GPL licenced. They are also more accurate than the 200-series indexes based on USNO-B and they contain at least as many star-quads as the USNO-B based indexes down to the same minimum star-quad features of 2 arcminutes.

-Antti

#16 LoveChina61

LoveChina61

    Viking 1

  • *****
  • Posts: 925
  • Joined: 19 Jun 2009

Posted 28 January 2013 - 11:32 AM

I finally was able to download everything including the massive index files. I got it installed. Out of about 8 tries, it solved twice. It took about 20 minutes to solve each of those two times. The other times it was not able to solve even after 45 minutes so I exited out of the program and tried another picture file instead.

I am not sure if I am doing something wrong or what. I wonder if using the 200-series index files would work better. I mean, were most people getting good solves quickly with those indexes in place? Alternatively, I wonder if you can suggest some other parameters that I can insert into the program?

Thanks for any help you can give! I love to use the astrometry.net online and have had good results with it most of the time. But I do not always have an internet connection available and would love to be able to use it offline.

#17 fmhill

fmhill

    Messenger

  • -----
  • Posts: 436
  • Joined: 17 Jul 2012
  • Loc: Cape Cod, Massachusetts, USA

Posted 28 January 2013 - 01:16 PM

You should be getting plate solves much faster than 20 minutes. I have Astrotortilla running on a LAPTOP computer with an Intel i3 processor and 6GB RAM with Windows 7 Pro 64 bit operating system and resolves take 7 to 20 seconds.

Try setting Sigma to 100, that is what I find works best...

It would help to understand what problem you are having if you would tell us what your computer is, CPU, amount of RAM, and operating system. Also what the FOV and focal length of your telescope.

#18 LoveChina61

LoveChina61

    Viking 1

  • *****
  • Posts: 925
  • Joined: 19 Jun 2009

Posted 28 January 2013 - 08:24 PM

I set the Sigma to 100 but it doesn't appear to be helping at all.

A handful of seconds after it starts trying to do a plate solve, at the bottom of the screen it says "Field 1 did not solve (index index-4000-07.fits, field objects 1-10)" and then keeps sifting through the various index files to no avail.

For testing purposes, I am only using picture files that were already able to be solved by the astrometry.net online site.

My computer system is fairly decent:
Vista Home 32-bit, Intel duo core 2.4gz, 3gig RAM

I am using a SBIG 2000xcm with pixel size 7.4 x 7.4 microns
FOV 28.3 x 21.2 arcmin (per Astrometry.net online plate solve)
2.22 arcsec/pixel

1422mm Focal length from my Meade LX200 10" Classic, native focal length 2540 (@f10) x .56 focal reducer in place.

Any help you can give me would be appreciated :)

#19 fmhill

fmhill

    Messenger

  • -----
  • Posts: 436
  • Joined: 17 Jul 2012
  • Loc: Cape Cod, Massachusetts, USA

Posted 28 January 2013 - 09:21 PM

Vista is an operating system I know nothing about however tests I have done with Astrotortilla on various different Windows computers leads me to think your computer should not be the cause of such long resolve times.

What do you have Scale Miunimum and Scale Maximum set to
?

As I use two different telescopes, I set Scale Min to .2 (0.2) and scale maximum to 2 which is a range both my telescopes FOV fit within...

What range of indexes do you have installed?

What format and size are the image files you are using?


#20 LoveChina61

LoveChina61

    Viking 1

  • *****
  • Posts: 925
  • Joined: 19 Jun 2009

Posted 29 January 2013 - 10:13 AM

The scale_low was previously set to 0 and I changed it to 0.2

The scale_high was set to 179 and I changed it to 2 as you suggested.

I then tried it out again but it did not solve after 25 minutes so I stopped it. I still see the message "Field 1 did not solve.." and then it starts to thumb through the index files.

The pictures I am trying to solve with are JPEG files that are 295x345 in dimension and 28kb in size. FOV = 28.3 x 21.2 arcmin. They are screenshots of the 10 seconds focusing shots my camera was displaying on the computer screen. Each of the pictures was successfully solved by the online version of the Astrometry.net plate-solver.

I downloaded the indexes beginning with the ones that were 20% of the size of my camera's FOV and all other index files on up to 30 arcmins. Altogether it was something like 3.39gigs of index file data.

Thanks for hanging in there with me :)

#21 fmhill

fmhill

    Messenger

  • -----
  • Posts: 436
  • Joined: 17 Jul 2012
  • Loc: Cape Cod, Massachusetts, USA

Posted 29 January 2013 - 11:19 AM

About the index files you have loaded, how you chose and loaded the indexes does not answer the question I asked, I want to know what range of files you actually have, i.e. 4005 to 4012 as an example. However that you downloaded 3.39GB index files causes me to guess you probably have them all...

What I am most concerned with is what the low numbered range you have is, mainly 4006 and below... I would move any index files above 4006 to a storage directory so AT does not see them when it is resolving...

About the "Field 1 did not solve,field objects 1-10" message, you will always get this, it means only that a resolve was not found in that index on the first pass. AT will make a number of passes through the indexes looking for a best fit and having indexes not needed for the FOV of your camera and telescope combination only slows the search for best fit down...

What I am most suspicious about, is the image size and format you are trying to resolve. I resolve JPG images approximately 2400 x 1500 pixels in size, and using a computer with an i3 2.6 Ghz CPU and Windows 7 Pro 64bit operationg system, a plate solve takes about 7 seconds to as long as 45 seconds on tough ones typically...

The key difference I see here from what you describe is the images and how you are getting them that you are attempting to solve... What I have heard from others although I have not run into it myself, is that AT is fussy about the image quality. Often images not directly from the camera will not resolve...

Therefore my suggestion would be to take imaged directly as downloaded from your camera, do a simple file type conversion to JPG without altering size or any other parameter, and see if they resolve...

I'm suspicious the reduced image size you are attempting to solve simply does not hold adequate resolution of data in JPG format at the reduced size... This seems to be the most significant difference between what I am resolving and what you are attempting to resolve...

#22 LoveChina61

LoveChina61

    Viking 1

  • *****
  • Posts: 925
  • Joined: 19 Jun 2009

Posted 29 January 2013 - 11:51 PM

The data file range I downloaded runs from 4002-4007. Would it help if I also downloaded the 4001 index files?

I am now converting directly from the raw camera files to JPEG and trying that out. My .jpeg frames are now a larger 1200x1600. The next time I will try to take them with an even higher resolution if you think that would help. I have the SBIG set to "Auto" image quality, but the next time I can set it to "High".

The Astrotortilla program occasionally freezes up on me and says "Not responding", but if I leave it alone for awhile and don't click on it with my mouse then it often works out the kink on its own and then continues sorting through the index files where it last left off.

It solved my first file in 7 minutes which was a picture of ngc2903 saved as a 2.2mg 1200x1600 jpeg file.

It was not able to solve the second photo of ngc891 which was a 1.89mg 1200x1600 Tiff file. I resaved the same file as a jpeg and it also was not able to solve it.

It then solved the Pleiades in just 46 seconds. I used the original 30sec exposure FITS file for solving.

I then stuck with FITS files for it seemed to love them. It solved the next 3 FITS pictures as follows: M95 in 27 seconds, M66 in 6 minutes 20 seconds, Flame Nebula in 14.2 seconds.

That's great and meets my needs exactly! I have learned how to freeze the time in Starry Night Pro, then set it back to the original time I took the FITS photo (as revealed in the FITS header), then click on "Edit" and then "Center On" and then enter the plate-solved coordinates, then Sync the telescope's gaze to the center of the screen (e.g. the plate-solved coordinates), and then change the time in Starry Night Pro back to the current/present time. I imagine this could be done with any telescope control program. In this way, I do not need to rush whatsoever in receiving the plate-solve results back from Astrometry.net or from the AstroTortilla program.

Thank you all for this excellent program :)

#23 fmhill

fmhill

    Messenger

  • -----
  • Posts: 436
  • Joined: 17 Jul 2012
  • Loc: Cape Cod, Massachusetts, USA

Posted 30 January 2013 - 12:18 AM

By all means, stick with the FITS file format, no need to convert to JPG. You did not explain in your prior posts that RAW files were in FITS format, only that you were using screen capture in JPG format... Converting to JPG format is only necessary if your raw files are in a format AT will not recognize...

As to index file 4001, the way to determine if this is necessary is to use the log view in "Tools" and keep a record of which indexes that resolves are completed with. If you see 4002 being used predominantly, then yes, having 4001 might allow AT to resolve the few that it is failing to resolve...

Unfortunately I know of no method of telling what indexes are necessary for a given FOV, all I know is that lower index numbers, i.e. 4002 resolves smaller FOV vs a higher number as 4007 being used for a larger FOV.

I have seen a post on a different forum where someone came up with a formula for relating FOV to Indexes however I do not have that information nor do I know how accurate it is... I think the best approach is to keep a record of what indexes are used the most with your images and as I suggested before, move the unused indexes to a storage folder/directory so they are not being scanned needlessly wasting time. And if you see the lowest numbered index being used often, then, yes, I would think adding the 4001 index is worth a try...

Glad to hear you are making progress...

#24 ldesign1

ldesign1

    Viking 1

  • *****
  • Posts: 616
  • Joined: 17 Sep 2009
  • Loc: Northern Illinois

Posted 30 January 2013 - 12:52 AM

Unfortunately I know of no method of telling what indexes are necessary for a given FOV, all I know is that lower index numbers, i.e. 4002 resolves smaller FOV vs a higher number as 4007 being used for a larger FOV.


If you go back to the AstroTortilla installation file, and select to install only the Index Files, you are allowed to choose the FOV range in arcminutes that you want to install. Don't go through with reinstalling the files. Just take note on which Index Files Correspond to which FOV.

#25 fmhill

fmhill

    Messenger

  • -----
  • Posts: 436
  • Joined: 17 Jul 2012
  • Loc: Cape Cod, Massachusetts, USA

Posted 30 January 2013 - 01:43 AM

Unfortunately I know of no method of telling what indexes are necessary for a given FOV, all I know is that lower index numbers, i.e. 4002 resolves smaller FOV vs a higher number as 4007 being used for a larger FOV.


If you go back to the AstroTortilla installation file, and select to install only the Index Files, you are allowed to choose the FOV range in arcminutes that you want to install. Don't go through with reinstalling the files. Just take note on which Index Files Correspond to which FOV.


You must have a different version than what I am using, I do not get any listing of files as you describe...

Any chance you could supply a screen shot in JPG mode of what you are suggesting?






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics