After ASTAP version 2025.09.12, ARM 64 bit on MacOS, all versions are not...
Stacking program, astrometric (plate) solver, fits file viewer
Brought to you by:
han59
Menu
▾
▴
I have 9 FITS files which I can solve reliably 9/9 for the above version, but all later versions only solve 4/9 of the files. I can swap versions and this is 100% repeatable, the above version solves all files but the later versions do not.
This is keeping all settngs identical. The file quality varies, due to intermittant cloud, and the higher quality files all solve but the only one of the lower quality files does - but only with the recent versions.
Hi Paul,
That should not happen. The reliability should go up, not down. That what I'm seeing with my test files. I'm guessing, but I assume the new version pick up something like hot pixels. So I like to investigate more. Can you share the 9 images?
cs Han
Last edit: han.k 2025-10-05
Hi Paul,
Can you share the nine images for testing?
Han
Hi Paul,
Can you share the nine images for testing?
You can not reply to the email directly. You have to log in Sourceforge
Han
Han,
Sorry, I did not realise the old method of sending was no longer working.
Please try this link and let me know if successful.
https://www.dropbox.com/scl/fi/vyk03rf2hy5eyufcx7njq/New-Folder-With-Items.zip?rlkey=p0jlwy4vkpxo8s0qxezhi5s8n&dl=0
Hi Paul,
I found a bug and fixed it.
M/silicon processors:
https://sourceforge.net/projects/astap-program/files/macOS%20installer/astap_command-line_version_macOS_M1.zip/download
X86_64:
https://sourceforge.net/projects/astap-program/files/macOS%20installer/astap_command-line_version_macOS_x86_64.zip/download
There was a floating variable in the 3th level of star detection (of 4) which was overlooked. It only showed up in specific cases of star population density. Tell me it this works for you.
Cs, Han
Last edit: han.k 2025-10-11
Han,
I tried the M1 link but it does not download an installer - just astap_cli, which I have run but the existing download still has the issue, and the normal downloads are still oldfer versions, so I cannot test it.
Astap for Windows solved all your images. So I original thought you had a problem with astap_cli. There I did found coincidentally a bug, which is now fixed.
I will test the Mac version.
Last edit: han.k 2025-10-11
Okay with the MacOS X86_64 version all images are solved. So now the question is, is this only related to the M1/Silicon version or to your settings?
1) Can you make a screenshot of your settings in tab alignment? Compare it with my screenshot.
2) Can you install/run ASTAP type X86_64 version and test the solving.? At the moment I have no access to a Mac with a M/silicon processor.
Han
Han,
As requested.
I noticed after doing this that I had some settings differently to those I normally use as I was trying them to work out what was going on, which is something I do for older images. I have retried with the same settings, other than D50 as I have D80 loaded. The problem still exists. I have tested this running the X86 version and the problem persists.
Last edit: Paul Homer 2025-10-11
Han,
I have investigated this further. I went and tried this on my 2012 MacBook Pro Intel and found it only failed to solve one of the images - ...003. whereas on M4 it failed to solve 005, 007, 008, 009 and 010. Below are the calculation differences - other columns are identical.
Quality Intel M4 HFD M4 Stars M4
001 44 19 3.4 3.9 498 295
002 45 23 3.5 3.9 540 341
003 50 28 3.4 3.6 563 368
005 80 4 2.2 6.7 379 184
006 80 4 2.1 6.9 366 172
007 72 4 2.2 7.0 343 171
008 90 4 2.0 6.8 357 179
009 94 4 2.0 6.6 366 169
010 85 3 2.0 6.8 338 158
Hi Paul,
I have temporary a virtual M/Silicon MAC running. What I now understand is that you have problems with stacking and for some files it reports in the column result "No solution". So the stacking is skipping some files. Not to confuse find finding astrometric plate solution which works.
This problem occurs in all version, so MacOS and Windows versions of ASTAP.
I tested older ASTAP version from 2024 and 2023 and the problem is for those versions even worse.
I will look into this. But what would you could do for the moment is to select for alignment the option "Astrometric alignment" instead of "star alignment". That will work better for your files.
Han
Han,
What confused me was the different results for some columns between Intel and M4 Macs, my plan is to keep using the older version from early September, however, it is interesting that you have found something more fundamental.
Last edit: Paul Homer 2025-10-14
Now the problem is clear. It depends on the number of stars/quads used by the reference image. You try to stack images of different quality. The maximum number of stars to use for alignment is typically set at 500. Your best image contains about 430 detectable stars which will create about 300 quads to compare. But some images have only about 170 detectable stars (and a much larger HFD). The quads created by less stars will be very different compared with the reference image. So the stack routine can not find a match. See attached. The reference image is marked with a crown.
As soon you set in tab alignment the maximum number of stars to the minimum found so about 170 star, the stacking will combine all files.
The reason you see a different behavior in the X86_64 Mac is likely by having the image different sorted. If you press stack button without analysing the images first, then ASTAP will take the first image as reference image which could have a lower amount of stars like 250 resulting in a different matching result. For the images which indicate "No solution" I will change the message to "No match" to avoid confusion. Note that I have been very busy with modifying the solving algorithm so I initially was focusing on a solving problem.
So the solution is to set in tab alignment the maximum number of stars to the 170.
Creating an automatic fix is not so easy. For the moment I have changed the routine to limit the number of "stars to use" based on the number of stars found in the reference image. That works if the other images have more detectable stars but not if they have less. You do not want to limit the number of stars based on the worst image. Maybe an option to set the reference image manually could help.
There should be no difference between the X86_64 and M/Silicon versions.
Han
All 9 files are stacked when the "maximum nr of stars to use" is set between 170 and 350. So there is some flexibility but not unlimited.
I have added an option in the popup menu of the stack menu tab lights to force a reference image. See attached.
It is not ideal but at least you can force an image with a lower star count as reference.
https://sourceforge.net/projects/astap-program/files/macOS%20installer/astap_M1.pkg/download
Does this work for you?
Han