Registering 2 night sessions together fails

Software Siril
1

So I took a shot at M31 over 2 nights.
Same camera & lens but the first night I didn’t have calibration frames, the second night I did.

When I process both sessions totally separate, it works to register & stack them.
However, when I copy all the fit files (debayered lights from sessions 1 + debayered calibrated lights from session 2) into 1 sequence and try registering, it only succeeds in registering the fits from session 1.

The lens, focal length, iso, shutter speed were the same over the 2 sessions. The aperture however was f4 vs f5. I assume this is causing the issue?

How can I merge 2 sessions and register them? Or can I somehow stack the 2 sequences that are seperatly succesfully registered?

Mac Studio M1 Max 64GB RAM 4TB SSD
Mac OS 14.6.1 & Siril 1.2.4 (build11551)

Console output at the end of trying to register (translated from Dutch to English):

When starting

13:10:16: Findstar: processing for channel 1...
13:10:16: Execution time: 336.61 ms
13:10:16: 1137 Gaussian profile stars found in image, channel #1 (FWHM 6.787534)
13:10:20: Registration: processing with method: Global Star Alignment (deep-sky)
13:10:20: Interpolation clamping active
13:10:20: Global star registration: with current memory and thread limits, up to 10 thread(s) can be used
13:10:20: Reading FITS: file m31-combo_00001.fit, 3 layer(s), 6984x4660 pixels, 16 bits
13:10:20: Reference image:
13:10:20: Findstar: processing for channel 1...
13:10:20: 1137 stars found in reference, channel #1
13:10:20: FWHMx: 6.78 px
13:10:20: FWHMy: 4.25 px
13:10:20: Global star registration: processing...
13:10:20: Reading FITS: file m31-combo_00001.fit, 3 layers, 6984x4660 pixels, 16 bits
13:10:20: Reading FITS: file m31-combo_00079.fit, 3 layers, 6984x4660 pixels, 16 bits
13:10:20: Findstar: processing for channel 1...
13:10:20: Reading FITS: file m31-combo_00096.fit, 3 layers, 6984x4660 pixels, 16 bits
13:10:20: Findstar: processing for channel 1...
13:10:20: Reading FITS: file m31-combo_00042.fit, 3 layers, 6984x4660 pixels, 16 bits
13:10:20: Findstar: processing for channel 1...
13:10:20: Reading FITS: file m31-combo_00112.fit, 3 layers, 6984x4660 pixels, 16 bits
13:10:20: Findstar: processing for channel 1...
13:10:20: Reading FITS: file m31-combo_00061.fit, 3 layers, 6984x4660 pixels, 16 bits
13:10:20: Findstar: processing for channel 1...
13:10:20: Reading FITS: file m31-combo_00022.fit, 3 layers, 6984x4660 pixels, 16 bits
13:10:20: Findstar: processing for channel 1...
13:10:21: Saving FITS: file r_m31-combo_00001.fit, 3 layers, 6984x4660 pixels, 16 bits
13:10:21: Reading FITS: file m31-combo_00155.fit, 3 layers, 6984x4660 pixels, 32 bits
13:10:21: Findstar: processing for channel 1...
13:10:21: Reading FITS: file m31-combo_00127.fit, 3 layers, 6984x4660 pixels, 32 bits
13:10:21: Findstar: processing for channel 1...
13:10:21: Reading FITS: file m31-combo_00141.fit, 3 layers, 6984x4660 pixels, 32 bits
13:10:21: Findstar: processing for channel 1...
13:10:21: Reading FITS: file m31-combo_00002.fit, 3 layers, 6984x4660 pixels, 16 bits
13:10:21: Findstar: processing for channel 1...
13:10:21: 0 stars in view 141 found, channel #1
13:10:21: Not enough stars. Skipped frame 141
13:10:22: 0 stars found in frame 127, channel #1
13:10:22: Not enough stars. Skipped frame 127
13:10:22: 0 stars found in frame 155, channel #1
13:10:22: Not enough stars. Frame 155 skipped
13:10:22: Reading FITS: file m31-combo_00142.fit, 3 layers, 6984x4660 pixels, 32 bits
13:10:22: Findstar: processing for channel 1...
13:10:22: Reading FITS: file m31-combo_00128.fit, 3 layers, 6984x4660 pixels, 32 bits
13:10:22: Findstar: processing for channel 1...
13:10:22: Reading FITS: file m31-combo_00156.fit, 3 layers, 6984x4660 pixels, 32 bits
13:10:22: Findstar: processing for channel 1...
13:10:22: 1257 stars in view 112 found, channel #1

Then it runs over all files, and ends with the following

13:12:10: Sequence processing partially succeeded, with 310 images failing.
13:12:10: Execution time: 1 min 50 s
13:12:10: Registration complete.
13:12:10: 425 images processed.
13:12:10: Total: 310 failed, 115 registered.
13:12:10: Checking sequences in directory: /Users/<redacted>/Desktop/import/Astro/M31-combo.
13:12:10: Close array m31-combo_
13:12:10: Reading FITS: file r_m31-combo_00072.fit, 3 layer(s), 6984x4660 pixels, 16 bits
13:12:10: Loading array: r_m31-combo_ (1 -> 115)

Session 1 is 115 fit files, session 2 is 310 fit files.
So it appears it only succeeds at registering 1 session.

The first part of the logs say ‘Not enough stars. Skipped frame …’ which is weird because I use the same settings as when I registered both sessions separately and then it worked just fine.

2

More information is needed…
It fails. Ok, but what are the logs?
what is the version, the OS,…
Do the two sessions overlap well?

etc… —> How to report issues — Siril 1.2.5 documentation

3

updated original post

4

Can you share 2 images, one from the uncalibrated sequence and one from the calibrated one?

5

The fit when I put it in a new sequence or the original raw file?

6

The 2 fits that you put in the new sequence, just to test why the second wouldn’t align to the first

7

https://www.dropbox.com/scl/fo/s20hiyu26zjvubwc6fw9m/AMBIbC_5sbdGIGShgw80v4M?rlkey=4fo1aq13ujq5tg87k4zd1rkxl&dl=0

File that ends with 00005 is one from the first night, that successfully gets registered.
File that ends with 00300 is one from the second night, which doesn’t register.

8

ok, the problem is that one image is 16b and the other is 32b.

I guess that for the first sequence, this is the uncalibrated, you’ve converted and debayered the sequence in one go using the debayer checkbox in the Sequence tab right?
Instead you should just convert (without debayer) to a sequence, say lightS1 for “lights of session #1”, and then type the command

calibrate lightS1 -debayer

This will ensure that the images are debayered and converted to 32b. You will get a sequence pp_lightS1

Then, another piece of advice. You should not try to join the 2 registered sequences (so r_pp_lightS1 and r_pp_lightS2) but the 2 “calibrated” sequences instead. That is to say you should convert all pp_lightS1 and pp_lightS2 images to m1combo and take it from there, so register and stack. Otherwise, your images have undergone twice interpolations (from the individual session registration and then again from aligning the sessions together).

Cheers,

C.

9

For the first sessions I didn’t take calibration frames, so I just converted. The second session I had calibration frames so did that as well and took the calibrated version (maybe I debayered those, not sure anymore).

I’ll try again as you say and report back. Thanks for the effort so far!

10

Without calibration, files will remain 16-bit. There is no need to convert them to 32-bit. This is why you have such a difference.

11

Ok so the 16 vs 32 bit is not the issue then?
I tried loading them all in 1 sequence and then choose a frame from the second session as my reference frame. I clearly see stars on the image but opening the Dynamic PSF window, no matter the settings I use, it recognises 0 stars. So something is wrong but I can’t figure out what.

12

Yes it is. I was explaining why the difference.