Plate solving and goto with AstroTortilla

[RogerTheDodger]

A few days ago I came across a utility called AstroTortilla - henceforth abbreviated to "AT".

AT is a windows wrapper utility that combines cygwin, the astrometry.net plate solver, ASCOM mount control and external camera/imaging software into a one-stop utility that wraps together software used in astrophotography to easily achieve:

sub-arcminute GoTo accuracy,

fully automatic blind GoTo calibration and

fast, quantitative polar alignment

by using plate solver software to work out the coordinates of a stellar photo. (to quote the sourceforge website https://sourceforge....illa/home/Home/ )

Installation has a couple of gotchas, on the link above it tells you to READ THE GETTING STARTED MANUAL - if you don't, it won't install correctly.

The second gotcha is Windows 7 - use a specific install dir for AT (like c:\astrotortilla) - there is a problem with permissions if it installs to Program Files.

Just let cygwin install to the default.

You will need to request access to the index files from Astrometry.net, the guide gives the email address and I got access in about 4 hours. There is a lot of them, you don't necessarily need them all - more info on selecting which files is given here http://trac.astromet...trometry/README in the Getting Index Files section.

The third gotcha is ASCOM communication - all programs need to use the same user level for them to communicate with a single EQMOD instance - the easy way I found was to run everything as Administrator (I'm using CdC, PHD, AT and APT).

Once installed and the indeces are in the correct place you are ready to try a solve. Start AT, and open the log viewer so you can see what is happening. leave the mount disconnected, and in the camera section just use file open and hit Capture and Solve, give it an image file and let it run.

Solve times will be dependent on the image size, FOV and what indeces you are using. For my setup, imaging with a canon 1100d on a SW200PDS I am using the 205,206 and 207 indeces, downscaling = 2 or 0, scaleminimum = 1, scalemaximum = 2 and sigma = 11 and I'm getting reliable solves from a 5 sec 6400ISO image in under 25 seconds.

So how does the GoTo perform? Last night I set up on my pier and started up the software - no star alignment - and told CdC to slew to M33, the initial unaligned slew was inaccurate as youd expect - almost an hour out in RA and a couple of degrees in DEC. So, with the three lower right tick boxes selected I hit Capture and solve, AT calls APT to fire off a 5 second image, reads it solves it, re-slews the scope to it's calculated position and triggers off a second image to check the slew and there is M33 bang in the center of the image, ready to start imaging... total time from unpark was less than two minutes.

I didn't have time to try the polar alignment part, too many clouds.

This tool massively shortens the time to get the scope on target for imaging, I was cloud dodging last night and every slew to a new patch of clear sky got me on target in a couple of minutes, I was truly impressed so a big big thanks to the team developing this

You can use the astrometry.net software from the command line in cygwin from this install if you wish to do actual solves and get the annotated output files.

[Breakintheclouds]

This sounds very exciting. I've been looking for a way to improve my goto accuracy without spending big money on TPoint or MaxPoint. I'll give this a go. Thanks for sharing!

[chrisshillito]

Is AstroTortilla creating and managing some kind of pointing model itself that results in the improved goto?, or is it just providing a sync that EQMOD then uses in the usual way to to build its own internal model?

Chris.

[Breakintheclouds]

I'm installing it right now, and checking out the documentation. From what I've read, you do your GOTO as usual, with your preferred planetarium/control software. Once the slew stops, AT offers more than one option. The first is to take a photo, work out exactly where the scope is pointing, and issue a sync command to EQMOD as you suggest. That's pretty good, but perhaps even more usefully, it can then use the information about where the scope landed after the GOTO to issue a command to the mount to slew to the RIGHT place. It then checks again whether the scope is now pointing where you told it to point and corrects again if necessary. You can choose how accurate you want this to be - e.g., you can say "Keep going until you're less than an arcminute from where I told the scope to go"

I am *so* excited to try this out tonight. Get lost clouds!

[themos]

We have been discussing doing the same thing with cygwin + astrometry.net + homemade scripts here:

http://stargazerslounge.com/topic/143540-star-hopping-with-on-the-fly-astrometry/

[RogerTheDodger]

I didn't actually look at EQMOD to see if the sync was adding them but I suspect it must be as it is updating CdC. on the first slew from parked it looks to see where it is actually pointing and computes a new slew based on the difference between where it is and where you have requested, and then moves to that point, it doesn't triangulate. Moving to a new point in the sky last night was always one slew from CdC then one correction from AT.

[RogerTheDodger]

Ian, make sure you get version 0.3 which came out today. If you have 0.2.10 you can just install 0.3 over the astrotortilla part, seems to work from my daylight test.

[Breakintheclouds]

Thanks, Roger - as it happens I did notice version 0.3 saying it was less than four hours old and took a chance by downloading it. So AT is now on the computer and I'm just waiting for all the indices to download right now so that I can test it...

[chrisshillito]

I didn't actually look at EQMOD to see if the sync was adding them but I suspect it must be as it is updating CdC. on the first slew from parked it looks to see where it is actually pointing and computes a new slew based on the difference between where it is and where you have requested, and then moves to that point, it doesn't triangulate. Moving to a new point in the sky last night was always one slew from CdC then one correction from AT.

Hi Roger,

It might be worth checking the EQMOD points list/point count just to be sure of who is doing what. If an ASCOM sync is being issued then I would doubt the need for astrotorilla to perform any goto computations. It just has to read the RA/DEC coords prior to the plate solve and then issue a new slew to the exact same coords once the sync has been applied. The accuracy of goto would remain EQMODs responsibility with astrotorilla providing a more convenient method of building/improving the EQMOD alignment model and automatically retrying gotos should a plate solve show a positioning error (some time ago I wrote a little client application for elbrus to do the same sort of thing).

There are several approaches to buiding and using the EQMOD pointing model - some folks with permamently sited mounts will create an all sky model that can be reused over many sessions, however many who setup each night, and don't intend to tour the sky, will simply add points close to their intended target as and when they visit them. The advantage of using a plate solver for noth methods is not only speed but also that you can sync closer to your target (as you're not needing bright stars to sync on) and there is no possibility of accidentally synching on the wrong object.

Chris.

[RogerTheDodger]

It's even easier than that, you tell it where you want to be and it puts you there - start point is irrelevant, all you need is PA on the mount. as I have to dismount from my pier after each use it's invaluable to me as it messes up the park point as I re-balance the scope every time I re-mount it.

[skybadger]

I have been looking at the SAO WCS tools for similar reasons, including manual extraction of stars and brightness readings for widefield astrometry and photometry purposes. I looked at cygwin/tortilla a while ago but I could really do without cygwin....

Mike

[RogerTheDodger]

you don't have to touch cygwin at all - this installs it for you and everything is done from a windows gui

[themos]

That looks good, but what happens if you already have cygwin installed?

[Breakintheclouds]

There's an option not to install it again

[Breakintheclouds]

So to report back... I installed AstroTortilla yesterday, and following the instructions it was really easy. I tested it last I night, when gaps in the cloud allowed. Even with EQASCOM, my gotos are normally not ideal - perhaps because creating a good enough all-sky map takes too long. Although it failed on one or two plate solves last night, most of the time AT was superb, getting my CCD image *exactly* like SkyTools predicted it should look with one minute. When I think that it's currently only alpha software, this is definitely one to watch.

I'll do some more experimentation to see if I can pin down the couple of failed solves. It's possible it was thin cloud throwing things off, or perhaps I need to add the next most detailed set of index files

[ncjunk]

So to report back... I installed AstroTortilla yesterday, and following the instructions it was really easy. I tested it last I night, when gaps in the cloud allowed. Even with EQASCOM, my gotos are normally not ideal - perhaps because creating a good enough all-sky map takes too long. Although it failed on one or two plate solves last night, most of the time AT was superb, getting my CCD image *exactly* like SkyTools predicted it should look with one minute. When I think that it's currently only alpha software, this is definitely one to watch.

I'll do some more experimentation to see if I can pin down the couple of failed solves. It's possible it was thin cloud throwing things off, or perhaps I need to add the next most detailed set of index files

I have acp and it has a script that uses the astrometry.net api. Sometimes the astrometry server goes off line which can cause failures.

It is amazing that astrometry.net have produced such a usefull api...its fantastic

[RogerTheDodger]

Try adjusting the sigma value in the table bottom left, that will affect the number of stars found, value for sigma is between 5-200. What does the field size come out as for your cam? (middle right on AT)

[Breakintheclouds]

My field size is approx 22'x16', and AT calculates this properly

[RogerTheDodger]

Try scalemin 0 scalemax 1 or 2.

[Breakintheclouds]

I've set it to work in arcminutes, rather than degrees, so I tried 0 and 60 (which is basically what you suggested). However, I also set the refinement parameter to 1. This turns out to be really useful - after the first solve it then automatically adjusted scale_min and scale_max to what they ought to be - which in my case was about half the width of my frame (11) and double the width (44). Subsequent plates are now solving much faster. Perhaps give that a go?

[RogerTheDodger]

Ooh that's a good one, I hadn't spotted that.

[Breakintheclouds]

You and I seem to have become the resident SGL AstroTortilla fan club, Roger!

[RogerTheDodger]

Heh, the other imagers will eventually come to love it too

[ianaiken]

Sorry to go off topic slightly. I use Sequence Generator Pro for plate solving. The latest Beta also handles meridian flip so you don't need to intervene it just flips the mount, centers the frame, resumes PHD guiding and away it goes taking more subs. Well worth a look.

[RogerTheDodger]

All info is good info That looks like a good piece of software.