This is totally the type of testing I am looking for!!! I would like to refine what types of settings we need and what their values should be to solve different types of images for my options profiles.

I used a percentage because images can be very different, and I thought a percentage would work more often. Some images can have 10,000 stars and some can have 50 stars. I was thinking a percentage would be better because we wouldn't want to try to get rid of the dimmest 500 stars in an image that has only 35 stars, and we also wouldn't just want to get rid of 500 stars in the 10,000 star image. What do you think? I could provide both options and we could play with them to see what works best?

Han, I'm not sure about this one. I was thinking that the minarea setting in Sextractor takes care of getting rid of the hot pixels, doesn't it? If you set the minarea to 20, then hot pixels, which have an area of 1 pixel should be eliminated before even getting to the star filtering step. I think it would save processing time if we can get rid of the hot pixels before doing all the sextraction steps. The filtering, which is a feature that I added, is applied after sextraction is complete.

Han,

Perhaps you can help me with two issues I have had with ASTAP with the new features I added in the new version. The first question that I have is whether I can get a list of the detected source positions back from ASTAP. I have now gotten all of the other solver methods to give this information back, including the online solver. But there doesn't seem to be an option for getting this information back with ASTAP.

Also, for WCS, ASTAP produces a WCS file just like the other solvers all do and I have gotten WCS data back from all of them. But for ASTAP, CFITSIO seems to have some trouble opening the WCS file. I get this error: "Error opening fits file /private/var/folders/_t/ntxs0hp56b31tsp0_9t3rttm0000gn/T/externalSextractorSolver_1006126838.wcs, second keyword not BITPIX"

When I open the WCS files from ASTAP, I find that they are not the same as the WCS files from the other solvers. There are \n end of line characters in the ASTAP WCS file whereas the other programs don't have the line breaks. I am not 100% sure, but I think this is where CFITSIO is choking on the file. Acting on this theory, I tried removing all the line breaks and resaving the file, but CFITSIO still did not open the file. I don't know why. If you could help with this, I would appreciate it.

Thanks,

Rob

Hello Rob, no problem, I will try to adapt the WCS file. A little strange it requires BIPIX=8. The comment indicates ASCII of bytes array. Do you have an executable using CFITSIO to test the WCS file? I don't use the CFITSIO library and would have problem creating one. My program is written in Object Pascal without any library and C is not my thing.

I could export the detected stars. It will normally limit the number of stars to brightest 500 or as set. Any close double or small galaxy will be ignored very early in the process. The csv format would be easiest to implement. I had a quick look to the .axy files but that seems rather complicated with unnecessary many keywords. I could also write pairs of 4 byte floats big or small-endian but that will be more difficult to read. Same withe HFD and flux value. But how does it help you with testing? The viewer with annotations is much more powerful.

Yes I have no problem to help with the development of your program.

The sep.extract routine seems very powerful. The only area where it missing a very few good stars is in bright galaxies like M31, M81. Secondly it detects close doubles and small galaxies. In ASTAP these objects are filtered out on ovality. Just tested the hot pixels are filtered by the option Min area=20. That looks good. Is it not better then to change the default from 5 to 20?

In an image exposed of a few seconds it is possible to detect a few thousand stars. I have limited ASTAP default to 500 stars maximum but it's adjustable. More doesn't serve any purpose. But it will also work with 30 stars. As said before, the solving speed Astrometry.net seems to increase significantly by limiting the amount of stars. Strange enough I can't remember any command line setting in Astrometry.net to limit the number of stars.

Cheers, Han

The setting Min area=20 could result in problems for little undersampled images. In ASTAP I solve this by a minimum star size expressed in arc seconds. This minimum is default set at 1.5 arc seconds.

Thank you for the help with the WCS files. I don't think the BITPIX needs to actually be 8, it is just looking for the second keyword to be BITPIX. Looking in the WCS file you save, BITPIX is there, it is just on the second line. Due to the way that the WCS file is formatted, it can't read the second keyword, I think that is all it is saying. Yes, I can certainly test WCS files that you send, or a better option might be for me to just release this version of the program, send it to you, and you can test changes to the output of ASTAP until it accepts the WCS file.

So one purpose of my library is certainly to solve the images and get back the solution information for the program using it, but it has other uses. Another purpose is to get back the sextracted sources and HFR in the image for purposes of focusing, guiding, and source information. And another purpose is to that after an image is solved, you can get the RA and DEC of all the stars in your image, or the RA and DEC of where your mouse cursor is in the image. The program can use this information to do photometry and light curves on the data. I am attaching a screenshot so that you can see what I use the information for:



Yes you are probably right that we should change the defaults. Maybe 20 would be a better setting for minarea. Maybe that should be a setting that is determined based on my setFWHM function that I made that already produces a conv filter. The main thing that I need feedback on is what all the defaults for the many many settings we have at our disposal should be. I'm still playing with all of that I think.

I believe astrometry.net automatically limits it to the 500 brightest stars in the image when it does a solve?

Thanks for the help,

Rob

But before you solve an image, how do you know the size an arc second is in your image? or are you going based upon scale parameters, like astrometry.net does with the low and high scales?

Hey guys,

I have been promising this one for a few days. I have just released the newest version of my SexySolver Library. Please note that I renamed the application to SexySolverTester. The reason was because when I turned SexySolver into a single library, the name conflicted with the application name and that could cause problems I think. The new version adds a bunch of features. My favorite one is loading the RA and DEC of all objects in the image.

github.com/rlancaste/sexysolver-tester/releases/tag/1.0

Please keep testing!!

Oh and I did add the feature Han was recommending, to keep only a certain number of stars. That can now be specified in the star filtering options. Here is a screenshot of my using it below. Note that for the first image, the solving time was already less than 0.3 seconds and it didn't affect it all that much. But for the second image, which was a blind solve, the solve time was reduced from over 30 seconds down to about 8 seconds, and if I set downsample 2 as well, it reduced the time for blind solving this image down to just 1.5 seconds! So I do like your option Han, It helped a lot I think.



Thanks,

Rob

I have uploaded a new version ASTAP 0.9.362a to my webpage only:

hnsky.org/astap_amd64.deb


The wcs header is now different:

BITPIX = 8
NAXIS = 0
NAXIS1 is removed
NAXIS2 is removed

If the WCS files produced by this version are readable by CFITSIO can you try with a hex editor if BITPIX=16 also works? I like to change the header as minimal as possible.
There are not carriage returns in the file as it should be. So editing with a text editor will most likely not work. Editing with ASTAP is also not possible.

I will look to your version 1.0

For exporting detected stars, I could do it experimental. Which data would you like to read? X, Y, flux? , snr?, HFD?

Han

Han, Rob,

as I use the WCS Files from astrometry.net as well as from ASTAP: As far as I have seen: Han you are writing a text file with the entries, which are not readable from CITSIO and astrometry.net writes a FITS file with header entries only.

Michel

I will try again. An problem I notice is that ASTAP doesn't write the WCS files with a multiply of 2880 bytes. That is easy to fix.

Han

I'm having trouble getting this image to solve with the new sextractor. ASTAP could do it in 45s, but I let the other solvers try for several minutes before stopping them. I don't know if it's the number or stars, or that there's some CA in the image. nova.astrometry.net solved it pretty quickly. It's a fairly large file from the new ASI2600. It' also not a single frame, but a final image so no header info.