×
INDI Library v1.8.5 Released (19 Apr 2020)

April 2020 release of INDI Library v1.8.5 introduces new drivers while providing fixes and improvements to existing devices and core framework.

New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed

1 month 2 weeks ago
Docerland
Junior Boarder
Junior Boarder
Posts: 31
More
New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed #54386
Hi!
For what its worth. I have tried sexysolver 1.4 out of the box on my new Macbook Pro on different pictures taken with my Canon on a SW MN190 and ASI1600 on a 152mm refracror and they all solve in between 0,4-1,1 seconds and havent failed once. It works superbly!
Erland

Please Log in or Create an account to join the conversation.

1 month 2 weeks ago
rlancaste
Platinum Boarder
Platinum Boarder
Posts: 2264
Karma: 22
More
Topic Author
New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed #54398
Thank you very much, it is good to know I didn't mess anything up too bad with such a big set of changes. But of course, as always, I already did find a little bug.

I found that the star filters that I created were only being applied for the internal sextractor, they were not being applied to the external sextractor. That made the Ext SextractorSolver method not work too well as well. I will push a change later today that will try to fix that. There are still some differences between the stars found with the internal and external sextractor though. I think that Han alluded to this earlier, it seems to pick up a lot more noise than the internal sextractor. I don't think the star filter settings were the whole story. I will see if I can work on that.

Other than that, keep testing everybody and let me know if there is anything else.

Please Log in or Create an account to join the conversation.

1 month 2 weeks ago 1 month 2 weeks ago by han.k.
han.k
Gold Boarder
Gold Boarder
Posts: 303
Karma: 6
More
New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed #54399
Rob,

This version 1.4 works well. The Min size for stars works well for filtering the M108 image of Ron ( ufile.io/cqoibgcy ) I noted the SExtractor pick up more blogs (some oval) in the image then ASTAP. So it works very well.

I have uploaded and test image:
ufile.io/2yy1vv9p
This is an very faint image. An 0.25 second exposure with some extra artificial noise added. Both Sexysolver and ASTAP can solve it. But for SExtrator I have to set in the star filter the minimum size at 0.8 or lower. The stars are real stars and should have a shape close to Gaussian and a HFD of maybe 2. They are spread out over a few pixels. I don't know what star size is reported by SExtractor but I assume the SExtractor is measuring/reporting the HFR, half flux radius?

Going back to M108 image. For the hot pixel filtering, I have set the Min size at 0.9 or higher. So what I see is that hot pixel filtering could in extreme cases filter out faint stars. The original image without noise is not a problem for SExysolver with a Min size=1.

So not problem, just an observation at extreme conditions.

For Ron, I have released ASTAP 0.9.365 for Windows, Mac and all Linux versions. This version has a new hot pixel filtering which will work much better with your images. f they don't solve in ASTAP, expose longer. There is less need now for darks.

Han

Please Log in or Create an account to join the conversation.

1 month 2 weeks ago
CapnRon
Gold Boarder
Gold Boarder
Posts: 179
More
New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed #54403
I downloaded and installed the new build of SexySolver-Tester. I tried running it with the smallscaleParallel option in the solver (IntSEPInteralSolver). Three of my 4 images solved in under a minute (26, 45, 57 seconds). The fourth image did not solve and after it failed the program closed/crashed?

I am using main option of scale but not position. if I just run the extractor, the AllStars option finds too many stars that are dark pixels. If I run parllelSmallScale it finds only 50 stars (assume that is a default because it does not show in the options.)

I am trying to repeat my first success (15 min ago) with this build, but now it is failing and closing? with just running the solve option of INtSEPInt Solve - parallelSMallScale.

Ron

Please Log in or Create an account to join the conversation.

1 month 2 weeks ago
CapnRon
Gold Boarder
Gold Boarder
Posts: 179
More
New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed #54404
Well it seems to depend on the order I run the images, because I just had 2 successful runs (just using solve as above) and the program is still on the screen and running. I will try the third image.

Please Log in or Create an account to join the conversation.

1 month 2 weeks ago
CapnRon
Gold Boarder
Gold Boarder
Posts: 179
More
New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed #54405
The third image just solved ok and no crash. The 4th image is Light_lumen...003 and now the 4th image solved in 10 seconds!? where it failed and crashed before? But right now it seems to working fine and this option is solving all four images with just 50 stars. It is disabling the inParallel option because I don't have enough RAM for all the index files, but it is running multiple child solvers.

Ron

Please Log in or Create an account to join the conversation.

1 month 2 weeks ago
CapnRon
Gold Boarder
Gold Boarder
Posts: 179
More
New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed #54406
Han,
I have upgraded to the latest version 80.9.365 of ASTAP.

thanks,
Ron

Please Log in or Create an account to join the conversation.

1 month 2 weeks ago
CapnRon
Gold Boarder
Gold Boarder
Posts: 179
More
New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed #54408
the M108 image solved in 9 seconds as is., the Tetrahedrons really find all the stars in this one. The other images seem to be finding far too few stars and the tetrahedrons don't match up with bulk of the stars. I will try adding the dark/flat.
Ron

Please Log in or Create an account to join the conversation.

1 month 1 week ago 1 month 1 week ago by han.k.
han.k
Gold Boarder
Gold Boarder
Posts: 303
Karma: 6
More
New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed #54605
I found an image which seems a problem for the Sexysolver. The Source-Extractor seems to work well for this image but the solver not. Size is 4x2.7 degrees so nothing special. Maybe my indexes?. It solves with "Int Sep Ext. ASTAP". Download:

ufile.io/kvy3vg88

Inspired by the Source extractor performance, I have rewritten the star detection for ASTAP 0.9.367 (development version). It's now more tolerant to unround star shapes:

www.hnsky.org/astap_setup.exe

It will now also detect small galaxies but since stars are abundant this should be not a problem.

Han

Please Log in or Create an account to join the conversation.

1 month 1 week ago
rlancaste
Platinum Boarder
Platinum Boarder
Posts: 2264
Karma: 22
More
Topic Author
New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed #54608
I'm not sure, it could be the indexes. I just tried it and it solved in less than a second. But it did say it solved with index 4111, which you might not have.

Attachments:

Please Log in or Create an account to join the conversation.

1 month 1 week ago 1 month 1 week ago by han.k.
han.k
Gold Boarder
Gold Boarder
Posts: 303
Karma: 6
More
New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed #54614
I have here the 4200 indexes (2MASSS based) including the 4211. The 4111 is based on the Tycho2.

I tried also the classic Astrometry.net and that one also didn't solve it, so the behavior of the Sexysolver is as the expected.

You would expect with 500+ stars it would find enough reference points. A strange technical glitch.

Then I downloaded more indexes the 4112, 4213, 4214 , 4215 using the prgram "All sky plate solver" and SExysolver was hanging itself in the new *.txt and *dat files. See below. That should be fixed. After removing these files from the index directory, still no successful solve using the Classic and SexySolver. Not so important. I downloaded this image some time ago from nova.astrometry.net

Han



Attachments:

Please Log in or Create an account to join the conversation.

1 month 1 week ago
han.k
Gold Boarder
Gold Boarder
Posts: 303
Karma: 6
More
New Internal Solver for Mac, Windows, and Linux -- Testing/ Experiments needed #54617
Downloaded the 4100 index fixes solving the previous file.

Then I tried to solve this 20 x13.3 degrees image:

ufile.io/ii1uyok4

SExysolver hangs after some time. See attached log. Classic Astrometry.net didn't solve either but completed after 400 seconds or so. ASTAP 0.9.367 has no problem solving this image.

File Attachment:

File Name: logsexysolver.txt
File Size:17 KB


Han
Attachments:

Please Log in or Create an account to join the conversation.

Time to create page: 1.572 seconds