Jump to content


Photo

AstroTortilla - how does it work?

  • Please log in to reply
104 replies to this topic

#26 ldesign1

ldesign1

    Viking 1

  • *****
  • Posts: 613
  • Joined: 17 Sep 2009
  • Loc: Northern Illinois

Posted 02 May 2013 - 05:16 PM

How much buffer should I give myself when setting the Scale Minimum and Scale Maximum ranges? For example, my imaging setup's FOV (per Astrotortilla) is 20.62(X)27.73 arcmins. What do you recommend I put as the Min and Max?

It looks like you're imaging close to 2500mm. I get 18.82' x 28.27' arcmins with my 10" LX200GPS and DSLR combination. Try my setting below. I typically go 5-10 above the maximum and below the minimum. As long as your setup doesn't change, the FOV should always remain roughly the same. I don't know how your screen captures affect the final size if you are cropping the image or adding extra border information. The scale is 100% accurate only with your actual image file from the camera.

Lastly, how do I turn off the Config setting which automatically determines the range search of the next plate solve I will do? It seems to store the range of the previous successful plate solve, and then automatically begins with this range for the next plate solve I try to do. I would like to turn this feature off so that I can just set a permanent range for it to search within (e.g. the FOV of my current imaging setup +/- some buffer).

You must be talking about the Default setting. You are able to load, modify and save new settings from the file menu. The default configuration file [astrotortilla.cfg] is located in your local user folder under Windows 7. It may be somewhere else with another operating system.

C:\Users\YOUR USER\AppData\Local\astrotortilla.sf.net\AstroTortilla

You can either overwrite this file or create a new file which you will have to load each time you what to solve at setting other than the default. I've created six different setting for my various setups.


You will need to experiment with the SIGMA value to get the shortest amount of solve time. It all depends on the level of noise in your image.

Attached Files



#27 Phil Sherman

Phil Sherman

    Surveyor 1

  • -----
  • Posts: 1505
  • Joined: 07 Dec 2010
  • Loc: Cleveland, Ohio

Posted 03 May 2013 - 08:12 PM

I've just started evaluating AstroTortilla (AT) for my own use and also discovered that it wouldn't solve any of my existing images. After some experimentation, I discovered that AT won't solve images with stars down to mag 19 or greater. It wouldn't even solve an image of M45!

I took one of my images and edited it to show only the brightest stars and it solved it in 90 seconds. That's great time for a virtual machine XP system with only 1g of memory.

My biggest problem with AT is that I need to run it in multiple environments. I want to do this with only one shared copy of the astrometric data files, which is supported. Unfortunately, the cygwin installation included with AT does not include any text editor or the terminal utility to run it in. A text editor is needed to alter the astrometric configuration files to move the plate solving data to a shared location. Fortunately, I have a Windows editor that will edit Linux text files.

I'm now altering my imaging procedures to include a 5-10 second exposure of any field I intend to image. This will allow me to use AT's "reposition the scope" feature to continue imaging on a second or third night.

Phil

#28 ldesign1

ldesign1

    Viking 1

  • *****
  • Posts: 613
  • Joined: 17 Sep 2009
  • Loc: Northern Illinois

Posted 03 May 2013 - 10:36 PM

I've just started evaluating AstroTortilla (AT) for my own use and also discovered that it wouldn't solve any of my existing images. After some experimentation, I discovered that AT won't solve images with stars down to mag 19 or greater. It wouldn't even soylve an image of M45!


It solves both my uneditted JPG and my processed file of M45 in 38 seconds and less. You may need to fiddle with the settings a liitle more. Which index files are you using? My success comes from using the 200 series.

#29 Footbag

Footbag

    Fly Me to the Moon

  • *****
  • Posts: 5755
  • Joined: 13 Apr 2009
  • Loc: Scranton, PA

Posted 05 May 2013 - 11:48 AM

I don't want to hijack this thread, but I've been very interested in getting AT working, but have had no luck.

I have everything hooked up and downloaded, but when I click capture and solve, I get an error in the log that says something like "Error Traceroute last call".

I figure it's something easy, but I'm stuck.

Thanks in advance.

#30 fmhill

fmhill

    Messenger

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

Posted 05 May 2013 - 01:11 PM

Adam,
That is not an error I have heard of or seen before.

First thing I have to ask, have you been able to get Astrotortilla to resolve any image? Is this a very first attempt?

Please explain/describe your set up starting with the computer and operating system you are using, the camera and OTA, and what series and range index files you have installed...

This is all information needed to understand what might be happening...

#31 Footbag

Footbag

    Fly Me to the Moon

  • *****
  • Posts: 5755
  • Joined: 13 Apr 2009
  • Loc: Scranton, PA

Posted 05 May 2013 - 01:31 PM

Adam,
That is not an error I have heard of or seen before.

First thing I have to ask, have you been able to get Astrotortilla to resolve any image? Is this a very first attempt?

Please explain/describe your set up starting with the computer and operating system you are using, the camera and OTA, and what series and range index files you have installed...

This is all information needed to understand what might be happening...


First, I'm typing all of this from memory. I'll be by my computer later and will correct any verbiage mistakes. It quite likely I mis started a message.

I haven't been able to get it to solve anything. I have tried a few times in the past, but it always stops immediately after clicking capture and solve. The bottom of the page says something about the image not being solved "in 0.0s" And I get the Traceroute error.

Computer is a Gateway Core 2 Duo laptop running vista. I downloaded the 22~ arcmin files and down.

Camera is a Canon XS, OTA is an Edge HD 800. I have tried manually selecting the images, but when I click OK, nothing happens.

EDIT: I just installed it on a desktop to test it and it seems to be working fine. I may have to reinstall on my laptop.

#32 fmhill

fmhill

    Messenger

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

Posted 05 May 2013 - 02:10 PM

Adam,
It sounds like your setup on the laptop has a configuration error...

I ran into this when I switched to Version 0.4 on my HP DV6T laptop. For reasons I have never been able to figure out, the AT installer was not putting the index files in the astrometrey/data directory.

Whether it was a folder creation error or a malfunction of the installer I was never able to figure out. As I also have a desktop with AT installed and working, I simply copied the astrometry/data area complete with index files to a USB drive and manually installed it using cut and paste to the correct location and AT has worked fine on the laptop ever since...

#33 Footbag

Footbag

    Fly Me to the Moon

  • *****
  • Posts: 5755
  • Joined: 13 Apr 2009
  • Loc: Scranton, PA

Posted 05 May 2013 - 05:27 PM

I uninstalled it and reinstalled it, but it crashed the computer while installing cygwin. So that wasn't setup correctly. Maybe I can manually install it?

#34 fmhill

fmhill

    Messenger

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

Posted 05 May 2013 - 08:00 PM

I uninstalled it and reinstalled it, but it crashed the computer while installing cygwin. So that wasn't setup correctly. Maybe I can manually install it?


Yes, Make sure you have enough disk space on the Laptop and you can use the Windows file manager to cut and past from a USB drive or possibly you can link the two computers with an Ethernet cable if they both have Ethernet ports...

There is nothing special about the CYGWIN, it can be copied from the other machine as long as you put it in its own folder complete with sub directories on the Laptop C: drive... This is essential so AT can find it...

As long as CGYWIN is a carbon copy of what is on the Desktop, it should work...

#35 LoveChina61

LoveChina61

    Viking 1

  • *****
  • Posts: 910
  • Joined: 19 Jun 2009

Posted 05 May 2013 - 08:52 PM

I am now having better success doing plate solves. Thanks to everyone for your help!

I am still using screenshots to run the plate solves upon. I mis-spoke earlier on how I do that. My observatory computer displays the starry photo that I want to plate solve, and I use the Host computer in my home to take a screenshot of what is displaying on my observatory's computer. I then run this photo through Astrotortilla for plate solving on my Host computer. It works well as long as I am using a 30-second shot that has had a Dark frame subtracted. This is much more convenient for me than sending the original 3.75mg FITS file back home to my Host computer and then plate-solving that original FITS file. My observatory computer is an ancient dinosaur and I don't want to load the entire Astrotortilla program on there.

I have been playing around with Sigma values. Another fellow had mentioned that he sets the Sigma value so that he gets between 20-60 "Simplexy". Less than 20 Simplexy and he might not have enough stars to do a plate solve, and many more Simplexy than 60 and it can take a long time to plate-solve. So I have been using my mouse to click on the Sigma value in the Astrotortilla startup screen and change the value to get the correct number of Simplexy. I put a Sigma of 80 (for example: --sigma 80 --no-plots -N none) and then click Capture and Solve to load my screenshot into Astrotortilla. As it starts to solve, I look at the bottom line and within the first 5-7 seconds it will flash the number of Simplexy. My setup is a bit more flexible so if the Simplexy falls within the range of about 15-80, then I let it continue to plate solve. However, if it falls outside of this range then I quickly Abort (Aborting works well at this point but can freeze up Astrotortilla on my computer if Aborted at other times), set a new Sigma value, and then try to solve all over again. If I don't get enough Simplexy, then I lower the Sigma value (e.g. from 80 down to 40. For example: --sigma 40 --no-plots -N none), and if my previous Simplexy number was too high than I raise the Sigma value (e.g. from 80 to 100) and try again.

Thanks again to everyone for your help!

Mike

#36 Footbag

Footbag

    Fly Me to the Moon

  • *****
  • Posts: 5755
  • Joined: 13 Apr 2009
  • Loc: Scranton, PA

Posted 06 May 2013 - 01:00 PM

I uninstalled it and reinstalled it, but it crashed the computer while installing cygwin. So that wasn't setup correctly. Maybe I can manually install it?


Yes, Make sure you have enough disk space on the Laptop and you can use the Windows file manager to cut and past from a USB drive or possibly you can link the two computers with an Ethernet cable if they both have Ethernet ports...

There is nothing special about the CYGWIN, it can be copied from the other machine as long as you put it in its own folder complete with sub directories on the Laptop C: drive... This is essential so AT can find it...

As long as CGYWIN is a carbon copy of what is on the Desktop, it should work...


That worked. Now just working on getting it to solve faster.

#37 pfile

pfile

    Gemini

  • -----
  • Posts: 3163
  • Joined: 14 Jun 2009

Posted 06 May 2013 - 05:59 PM

I've given up on AstroTortilla for now and have downloaded a trial for Sequence Generator Pro, which uses Elbrus and astrometry.net. I have been sucessful with Elbrus before. Last Saturday, I tried AstroTortilla but could not get it to open Nebulosity. It kept waiting on the camera, but the camera was working fine. I'm sure that it was my mistake, but I will wait for a more mature version.

Regards,

Charlie B


astrometry.net and astrotortilla use the same engine - astrometry's solve-field.

elbrus is supposedly a royal pain in the butt to install and get working, so there's no free lunch...

why don't you post some images so someone can help you with the parameters? i have been using astrometry.net's solver for years now with great success. astrotortilla really brings it to the next level with solve & sync.

#38 Charlie B

Charlie B

    Viking 1

  • -----
  • Posts: 958
  • Joined: 22 Mar 2008
  • Loc: Sterling, Virginia

Posted 06 May 2013 - 06:46 PM

astrometry.net and astrotortilla use the same engine - astrometry's solve-field.

elbrus is supposedly a royal pain in the butt to install and get working, so there's no free lunch...

why don't you post some images so someone can help you with the parameters? i have been using astrometry.net's solver for years now with great success. astrotortilla really brings it to the next level with solve & sync.



When astrotortilla worked, it solved fine with the parameters that I used on images that I had. I could not get it to open and use Nebulosity when actually trying to observe. I admit that it was likely my problem, but it seems to be finicky and overly sensitive to parameters.

My last excursion with astrometry.net was on Flickr, where I submitted some images to be solved. The returned solutions, at that time, were not correct. See here!

Elbrus is easily installed and reasonably easy to use, but does need a good scale and angle to work properly, and you need to know if the picture is inverted or not.

I've been plate solving with PixInsight mostly, but I wanted to automate the process, hence the trials of Sequence Generator Pro and Astrotortilla. I tried Astrotortilla first because it was free. I may return to it for this reason, but SGP has other benefits besides plate solving.

Regards,

Charlie B

#39 pfile

pfile

    Gemini

  • -----
  • Posts: 3163
  • Joined: 14 Jun 2009

Posted 06 May 2013 - 09:39 PM

i'll bet that both ImageSolver/AnnotateImage and astrometry are both wrong due to lens distortions. or there could be a bug in the flickr interface. i've never seen such a discrepancy between the two.

in fact i find ImageSolver to be very finicky and require a lot more tuning before it can find a solution. mainly if the # of stars detected is too many then StarAlignment just goes out to lunch trying to match stars.

i'm not using a camera control package which is supported by AT so i just use the open file dialog. i usually run with -z 2 and -H and -L, and that's it. solves every time though sometimes it takes a couple of minutes.

i actually want to switch to SGP but for now i'm stuck because i have the STT and so can't use PhD to guide while something else is connected to the camera via USB. one of these days i'm finally going to get tired of this and try to add the http support into phd2. just need to find the time.

the only thing that bothers me about SGP is that they require elbus, and only use astrometry as a backup. so far you can't promote astrometry to the main solver. added to that they wrote an interface for astrometry's web-based solver service... so if you want to run solve-field locally you have to install this hack which emulates nova.astrometry.net, which is really kind of a rube goldberg solution. it would be way more straightforward for them to just call solve-field and be done with it.

#40 LoveChina61

LoveChina61

    Viking 1

  • *****
  • Posts: 910
  • Joined: 19 Jun 2009

Posted 06 May 2013 - 11:14 PM

i usually run with -z 2 and -H and -L, and that's it. solves every time though sometimes it takes a couple of minutes.


I sure would like to be able to say that Astrotortilla solves everytime that I try to use it. Where do you place the settings you mention in your quote above?

You can see an example of the settings that I use at the very top of this thread's second page. I usually start with the setting of "--sigma 100 --no-plots -N none". Should I just place your suggested string of settings at the end of this sentence so that my final setting string is "--sigma 100 --no-plots -N none -z 2 -H -L" ? I don't mind waiting a few minutes for it to solve as long as it will eventually be able to successfully complete the plate solve.

Mike

#41 Footbag

Footbag

    Fly Me to the Moon

  • *****
  • Posts: 5755
  • Joined: 13 Apr 2009
  • Loc: Scranton, PA

Posted 07 May 2013 - 07:23 AM

OK. I got one to solve properly in 284s, but one improperly solved in 360s.

I need to work on getting it to solve faster and more accurately.

I am using 300s exposures to test with. Would shorter exposures be better? That is probably the way it would work in the field.

My sigma is set to 1, shold I change that? What exactly does sigma do?

Is the a place for more info on this besides this thread?

#42 LoveChina61

LoveChina61

    Viking 1

  • *****
  • Posts: 910
  • Joined: 19 Jun 2009

Posted 07 May 2013 - 08:17 AM

I just tried inserting the string "--sigma 100 --no-plots -N none -z 2 -H -L". Didn't work at all. When you get the chance, please let me know how to insert the extra parameters you suggested. I am always open to new settings that can enable the plate solving to work better and better.

I am still getting a higher "successful plate solve" percentage when just using the online version of nova.astrometry.net. However, when Astrotortilla is able to solve it always gives me a bit more accurate plate solve than the online version.

#43 Charlie B

Charlie B

    Viking 1

  • -----
  • Posts: 958
  • Joined: 22 Mar 2008
  • Loc: Sterling, Virginia

Posted 07 May 2013 - 10:14 AM

i'll bet that both ImageSolver/AnnotateImage and astrometry are both wrong due to lens distortions. or there could be a bug in the flickr interface. i've never seen such a discrepancy between the two.



You may be right, but ImageSolve matched the scale expected from the camera/lens combination, but it was not a blind solver. I haven't a clue why astrometry got it's scale value. However, astrometry correctly identified the objects in the image.


So far, I'm not sure that SGP is worth the effort, as I don't have an observatory. I may keep the free version for its scripting capability and forget its plate solving. Likely, I will give astrotortilla another shot after it matures more (and people like you work the bugs out and get a good handle on what parameters to use and when).


Regards,

Charlie B

#44 pfile

pfile

    Gemini

  • -----
  • Posts: 3163
  • Joined: 14 Jun 2009

Posted 07 May 2013 - 12:56 PM

-H and -L take arguments, image scale in degrees (by default). i know from prior blind solves that my camera/OTA combo yields a FOV of about 30 arcminutes by 20 arcminutes (0.5 degrees by 0.33 degrees) so i use -H 0.6 -L 0.2 just to give it some wiggle room.

#45 Phil Sherman

Phil Sherman

    Surveyor 1

  • -----
  • Posts: 1505
  • Joined: 07 Dec 2010
  • Loc: Cleveland, Ohio

Posted 07 May 2013 - 01:42 PM

I downloaded AT and also discovered that it wouldn't plate solve any of my existing images. The clue to getting AT to solve images is in manual where it states that you should be using a 5-10 second image at the camera's highest ISO. The critical piece of information is the 5-10 seconds - a very short exposure.

I took an unsolvable image and edited it to eliminate all but the 20 or so brightest stars in it. When I fed this into AT, it solved it in around 90 seconds, I considered this to be excellent time because AT was running in a WinXP virtual machine with restricted memory and a single CPU available to it. Observation of the computer during the analysis run indicated that the plate solve time was a function of disk I/O speeds. I expect that moving the astrometric data files to my SSD will significantly improve this.

I'm also changing my imaging procedures to include a short exposure of any field I image. This will be used to reposition the scope if I need to continue the imaging of that field the following night. AT will do this if it has an image it can resolve.

Phil

#46 pfile

pfile

    Gemini

  • -----
  • Posts: 3163
  • Joined: 14 Jun 2009

Posted 07 May 2013 - 04:23 PM

i'll bet that both ImageSolver/AnnotateImage and astrometry are both wrong due to lens distortions. or there could be a bug in the flickr interface. i've never seen such a discrepancy between the two.



You may be right, but ImageSolve matched the scale expected from the camera/lens combination, but it was not a blind solver. I haven't a clue why astrometry got it's scale value. However, astrometry correctly identified the objects in the image.


So far, I'm not sure that SGP is worth the effort, as I don't have an observatory. I may keep the free version for its scripting capability and forget its plate solving. Likely, I will give astrotortilla another shot after it matures more (and people like you work the bugs out and get a good handle on what parameters to use and when).


Regards,

Charlie B


it could be that they have downsampled the image to solve it but then not upscaled the result that they paste into flickr. there's no way they don't know the correct image scale because as you say the object identification would be completely wrong.

for me, AT is super useful to sync the scope. even small amounts of polar alignment error can cause large pointing errors if there's no model in the mount computer. with AT i just point the scope near where i want to image, solve and sync, and then do another goto, and my target is centered perfectly. no reason to build a model if you can periodically re-sync the mount.

imho it's of the greatest benefit if you don't have an observatory, cause you'd have to build a new model every time you set up.

#47 Footbag

Footbag

    Fly Me to the Moon

  • *****
  • Posts: 5755
  • Joined: 13 Apr 2009
  • Loc: Scranton, PA

Posted 07 May 2013 - 05:44 PM

OK. Once I got it to solve one of my images, I used the reported scale to adjust my arcminwidth settings accordingly. Now it solves most images very quickly.

What does the search radius do? If I know my mount is pointing close to my intended target, can I reduce this setting? Would it further speed it up?

#48 Charlie B

Charlie B

    Viking 1

  • -----
  • Posts: 958
  • Joined: 22 Mar 2008
  • Loc: Sterling, Virginia

Posted 08 May 2013 - 06:34 AM

imho it's of the greatest benefit if you don't have an observatory, cause you'd have to build a new model every time you set up.



True, but building a 4-star model is not very difficult and I can sync to a close by star to get very good pointing, which I've been doing for years. I don't truly need plate solving, but would like to have it.

Charlie B

#49 pfile

pfile

    Gemini

  • -----
  • Posts: 3163
  • Joined: 14 Jun 2009

Posted 08 May 2013 - 12:19 PM

i just had no end of problems with my G2 trying to model, plus having to build models on both sides of the meridian is tough for my location.

you can use AT to build up the model if you configure the ascom driver to do additional aligns on sync. that way you don't have to worry about centering stars or whatever, just sync/align where you happen to be pointing.

of course YMMV. for me, doing repeated syncs (not aligns) near each other caused the G2 to completely corrupt it's model.

#50 Charlie B

Charlie B

    Viking 1

  • -----
  • Posts: 958
  • Joined: 22 Mar 2008
  • Loc: Sterling, Virginia

Posted 08 May 2013 - 04:48 PM

i just had no end of problems with my G2 trying to model, plus having to build models on both sides of the meridian is tough for my location.



I do not have problems building models, but do have problems at the meridian flip. The newer motors stick out further than the older ones and I have to stop and wait until it's safe to flip. One good thing about plate solving is that I can return to the exact position after the flip.

Charlie B






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics