Plate solving and goto with AstroTortilla

[Gina]

Hi Gina,

I'm curious as to how you got your netbook to solve so quickly. I'm using an EeePC 1000 and the fastest I've been able to solve is a little over 70s. My setup has an imagescale of 1.79"/pixel and I set the lower limit to 1.78 and 1.8 for the upper limit. I run the solve on my desktop PC first so that I know which fits file the solution is found in and use only that fits file on the netbook so it doesn't have to search through lots of files before solving.

The vast majority of the time is used up by AT doing it's 'writing ppm, using 8-bit output, extracting sources, etc' - this takes well over a minute. The actual solve only takes a few seconds as it has only the correct fits file to check.

So, this leads me to think that your netbook is processing the preliminary stages much more quickly. My netbook is a 1.6GHz Atom with 2GB RAM. Is yours substantially faster? Did you have to tweak some settings to get through the preliminary stages more quickly?

Mine's a 1.6GHz Atom with 1GB RAM. Here's a screnshot of my Asus info and another of my AT window.

The image I used was about a minute or two test sub of the middle of the Heart Nebula with Atik 314L+ on Evostar ED80 with 0.85x FR and Astrodon 5nm Ha filter. AT found 99 objects with the settings above. I'm about to try it with some of my other images with different FL optics.

I should add that AT has the full set of index files I downloaded ie. 4005 - 4019. I think it solved with 4008 but I'll check (if I still have the log file).

[Gina]

No, it was 4006. Here's a screenshot of the log - divided into two overlapping parts.

[Gina]

I have some more results. I used another image this time without the FR - a 2m Ha sub of the middle of the Rosette Nebula. Knowing this to be 15% smaller aps I set the lower limit to 2 and the higher to 3 and ran AT. This solved in 32.5s. I checked the resulting app value, which was 2.22 app, so set L to 2.2 and H to 2.3. This time it solved in 19.2s.

[Tzetze]

Thanks for providing the details Gina.

I just did some further testing with image resolutions. At full res off the 450D, 4272 px on x axis, AT gets through the preliminary stages before checking through fits files in ~50s.

I resized the image to 2000px on x axis and AT got through the preliminary stages in ~30s.

Down to 1392px to get close to the res of the Atik 314L+ and AT got through the prelim stages in just under 20s.

Another good reason to upgrade to CCD in the future!

[Gina]

All the image processing is very much faster with the smaller Atik sensor.

[Gina]

I've now been trying the other extreme - a 55mm FL lens. As the ED80 + FR is 510mm and app 2.5 I would estimate the 55mm FL to give 25aps so I set L to 10 and H to 50 initially. With --sigma still at 100 I got 1999 targets so I aborted and set --sigma 200 and got 950 odd. Solved in 101s.

23.4 aps - so next go with 23.4 - 23.5 aps.

[Gina]

Slightly slower solve than previous images - 22s

[Tzetze]

I recently went through my saved subs and produced a jpg for different targets (only 16 so far).

Ran them through AT on my desktop machine and after they'd solved, I renamed the files to include image scale and corresponding fits file for solution, e.g. M35 1.79 4007-05.jpg

So, if during a session I want to go to M35 I move all the fits files except for index-4007-05.fits into a sub-folder so that AT can only search the correct file. This greatly reduces the total solution time especially when set to repeat until within 1 arcmin.

[russellhq]

Nice idea! I can't really use this feature in AT because of the backlash in my mount. When AT makes the adjustment to the mount after the first solve, the mount is still catching up with it'self when it takes the second image, the resultant capture is then just star trails.

[Gina]

I recently went through my saved subs and produced a jpg for different targets (only 16 so far).

Ran them through AT on my desktop machine and after they'd solved, I renamed the files to include image scale and corresponding fits file for solution, e.g. M35 1.79 4007-05.jpg

So, if during a session I want to go to M35 I move all the fits files except for index-4007-05.fits into a sub-folder so that AT can only search the correct file. This greatly reduces the total solution time especially when set to repeat until within 1 arcmin.

I wonder if that's worth the trouble - it might be. Won't really know until the weather permits some stargazing!

I'm just taking an example of each imaging combination I've used so far. Now covered all the 314L+ image sizes. 55mm, 200mm, 510mm and 600mm. The last one was the 200mm lens with a 1s sub of M42. That took 33s with L=6.6 and H=6.7 app. With --sigma 200 I got 35 hits. When it comes to real live data I imaging the sigma value will need adjusting for the conditions.

[Gina]

AT is definitely looking promising as far as solving is concerned - I think I've got that part cracked Connecting to the mount should be alright unless I get a clash between CdC and AT with the ASCOM driver.

I've yet to see how I choose an object. There seems to be two ways of doing it - use an image already captured and use CdC to choose the object.

Then there's how to use the 314L+. APT CCD mode (which I haven't tried) looks favourite. Or capture an image and use AT camera file dialog (fiddly). Then I've read of screen capture from the live picture but I couldnt see how that worked. It seemed you needed to input the image window margins from the capture software window. I found it would use the QHY5 guide camera (but couldn't plate solve the clouds!! ) Another possibility would be the DSLR finder camera with APT (don't like that idea).

[Tzetze]

Nice idea! I can't really use this feature in AT because of the backlash in my mount. When AT makes the adjustment to the mount after the first solve, the mount is still catching up with it'self when it takes the second image, the resultant capture is then just star trails.

From the config file...

[Telescope-ASCOMTelescope]

syncaccuracy = 1.0

lastselection = ScopeSim.Telescope

syncmaxwait = 2.0

propertyagelimit = 1.0

slewsettletime = 0.0

Looks like increasing the slewsettletime should help in that regard, Russell.

I wonder if that's worth the trouble - it might be. Won't really know until the weather permits some stargazing!

I'm just taking an example of each imaging combination I've used so far. Now covered all the 314L+ image sizes. 55mm, 200mm, 510mm and 600mm. The last one was the 200mm lens with a 1s sub of M42. That took 33s with L=6.6 and H=6.7 app. With --sigma 200 I got 35 hits. When it comes to real live data I imaging the sigma value will need adjusting for the conditions.

Well, it was certainly worth the trouble in my case Gina.

Suppose the image was to be found in index 4007. That would mean searching in 4007-00 through 4007-11. If it doesn't match in those files through fields 1-10, it will go through them again for fields 11-20. I'm not sure how many fields are in each file but I've seen it reach fields 110-120. In my experience it can take 7-10 minutes to find a match through that large a combination of files and fields. If there are going to be up to 5 iterations of this to get target centred to within 1 arcmin it will add up to a very long time where it's questionable if it's even worth using AT.

[Tzetze]

I've yet to see how I choose an object. There seems to be two ways of doing it - use an image already captured and use CdC to choose the object.

I simply use Tools > Goto Image. If you don't have an image of your own, use one of Hubble's.

[russellhq]

You can also use AT's bookmark function to return to a previous position instead of using an image.

Because it takes the mount about 25 seconds to settle if it slews the "wrong" direction, I gave up on slewsettletime and just manually slew it in after the first goto.

[Gina]

Well, it was certainly worth the trouble in my case Gina.

Suppose the image was to be found in index 4007. That would mean searching in 4007-00 through 4007-11. If it doesn't match in those files through fields 1-10, it will go through them again for fields 11-20. I'm not sure how many fields are in each file but I've seen it reach fields 110-120. In my experience it can take 7-10 minutes to find a match through that large a combination of files and fields. If there are going to be up to 5 iterations of this to get target centred to within 1 arcmin it will add up to a very long time where it's questionable if it's even worth using AT.

I see what you mean And if you move the files before it gets dark that ime isn't taken out of good imaging time. I guess the modus-operandum would be to keep all the files in the "other" directory and move just the one (or ones) you know it wants into the active directory before imaging. Drag-n-drop if a pretty quick operation

I would think you can also get the scope pointing in the right direction before dark - aving yet more imaging time.

[Gina]

I simply use Tools > Goto Image. If you don't have an image of your own, use one of Hubble's.

Thanks Yes, I guess solving a published image should work just as well as a simgle short sub of your own. I'll try that

[AlistairHowie]

Yesterday I spent about 45 minutes going through all the posts on this thread, and it struck me that there is some good information to be had within the thread .... but it is time consuming! I've got AT solving my 15s subs in 6-30s. I think it is by accident more than design! I do have a few quick questions for the cognoscenti ...

Is there somewhere that explains what all the solve parameters are in an easy to understand way? I'm not thick, but equally I'm not a programmer. I like the AT manual which is short, simple and, to my mind, well written, but I must admit to getting confused with all the solving options that I could use. I know there's this section here ... http://sourceforge.net/p/astrotortilla/home/Configuration/ ... but I got lost very early on with that page! OK, maybe I am thick, or perhaps I am expecting too much for free software that is, to be fair, in its early stages.

Secondly, and Kirkster brought this up some posts back, when setting the min and max arcseconds per pixel, these numbers seem to get reset after each solve. Is there a way to "hard code" in the relevant numbers and keep them there? Is that the -L, -H settings?

Also, in words of one syllable, can someone tell me what the downscale option means?

And lastly, how would you go about setting up a screen capture for other capture applications like Artemis? Gina raised that question above. I think that's all about inputting window margins, but how would you calculate those ... and would it depend on where that window is on your desktop, since these windows tend to be moveable ... at least they are in Artemis. My thought is that the screen capture window might be best since it could eliminate potential conflicts between various imaging programs and AT that some people have been reporting?

All thoughts gratefully appreciated.

[Gina]

You can also use AT's bookmark function to return to a previous position instead of using an image.

Thanks Yes, that looks very useful. I imagine you can go through all the objects you might want to image (as well as all the ones already done that you want to go back to) and bookmark them. Then when out in the obsy, pick the object you want, slew to it roughly from the bookmark, take an image and solve it and slew to exactly the right place. Great!

[Gina]

I've found that with a lot of software, the manual and help files are written by the developers and less easy for users unfamiliar with programing to understand. I had this problem myself when I was producing software for public consumption but I was fortunate to have a friend who was intelligent and used to computers but not into programing. She helped me see how the average user would see things and I developed manual writing for users rather than software writers. In later years this experience has helped me to write user guides for other people's software that are easier for the beginner to understand. Once I get familiar with a piece of software I can sometimes write simplified guides and bridge the gap between the software developer and beginner user. (Sorry if this is "blowing my own trumpet")

[IanL]

I've yet to see how I choose an object. There seems to be two ways of doing it - use an image already captured and use CdC to choose the object.

If using EQMOD, make sure the 'Append on Sync' option is selected. Now use your planetarium to slew to a target in the usual manner. Then go to AT, make sure all three 'After Solve' checkboxes are ticked, and click the 'capture and solve' button. AT will snap an image, solve it, do a sync (which updates the pointing model in EQMOD) and then repeats the capture, solve, sync until it gets within your selected distance of the target. On my setup, it takes two captures (first and one refinement) to get on target.

Once you are there, use the bookmarks option in AT. You can also do a solve of a saved image during the day and bookmark that if you prefer. Once you have some bookmarks, the planetarium becomes unecessary unless you plan to image targets that you haven't previously bookmarked.

[Gina]

If using EQMOD, make sure the 'Append on Sync' option is selected. Now use your planetarium to slew to a target in the usual manner. Then go to AT, make sure all three 'After Solve' checkboxes are ticked, and click the 'capture and solve' button. AT will snap an image, solve it, do a sync (which updates the pointing model in EQMOD) and then repeats the capture, solve, sync until it gets within your selected distance of the target. On my setup, it takes two captures (first and one refinement) to get on target.

Once you are there, use the bookmarks option in AT. You can also do a solve of a saved image during the day and bookmark that if you prefer. Once you have some bookmarks, the planetarium becomes unecessary unless you plan to image targets that you haven't previously bookmarked.

Thank you Ian - that's very helpful

[AlistairHowie]

(Sorry if this is "blowing my own trumpet")

No no, blow away please. I look forward to your AT numpty guide in due course

[Gina]

Strange thing... I added a Bookmark yesterday and today I've just added another from the current solve (for my next imaging session) but the first one has disappeared - only the second one showing What could I have done wrong?

[IanL]

I don't know where the bookmarks are stored. Most of the configuration is held in text files so have a hunt around. Might be you have loaded a different config and the first bookmark is in a different file?

[Gina]

Thanks Ian I used Save Settings but I've just checked Load Settings and it makes no difference which Settings file I use, it doesn't change the Bookmarks. That's all I've altered.