Lunar image registration and stacking issues

I’m not having much luck registering and stacking a set of 4 lunar FITS images that frame only a portion of the first quarter Moon (a quarter degree field centered on the disk’s center), so the full disk isn’t visible.

I tried both lunar and planetary registration methods, Image Pattern Alignment (planetary - full disk) and Enhanced Correlation Coefficient (planetary - surfaces), but the stack ends up off like this:

Any suggestions for options or processes that can help Siril register the images automatically? I may try manual registration but I’m not sure I understand how it works.

I’m using Siril 0.99.10 on Crostini Linux and Chrome OS Stable 91.0.4472.114. Here’s the console log:

15:50:30: Welcome to siril v0.99.10
15:50:31: Supported file types: BMP images, PIC images (IRIS), PGM and PPM binary images, RAW images, FITS-CFA images, Films, SER sequences, TIFF images, JPG images, PNG images, HEIF images.
15:50:31: Loading init file: '/home/paoloamoroso/.config/siril/siril.config'
15:50:31: Parallel processing enabled: Using 8 logical processors.
15:50:31: Setting CWD (Current Working Directory) to '/home/paoloamoroso/moon/moon20210617canary4'
15:50:31: Loading registration method: One Star Registration (deep-sky)
15:50:31: Loading registration method: Two or Three Stars Registration (deep-sky)
15:50:31: Loading registration method: Global Star Alignment (deep-sky)
15:50:31: Loading registration method: Image Pattern Alignment (planetary - full disk)
15:50:31: Loading registration method: Enhanced Correlation Coefficient (planetary - surfaces)
15:50:31: Loading registration method: Comet/Asteroid Registration
15:50:31: Default FITS extension is set to .fit
15:50:59: Conversion: processing 4 files...
15:50:59: Conversion succeeded, 4 file(s) created for 4 input file(s) (4 image(s) converted, 0 failed)
15:50:59: Reading FITS: file, 1 layer(s), 3358x2536 pixels
15:50:59: Sequence loaded: firstquartermoon_20210617_ (1->4)
15:50:59: Execution time: 188.09 ms.
15:51:33: Registration: processing using method: Enhanced Correlation Coefficient (planetary - surfaces)
15:51:33: Reading FITS: file, 1 layer(s), 3358x2536 pixels
15:51:33: Reading FITS: file, 1 layer(s), 3358x2536 pixels
15:51:33: Registration finished.
15:51:33: Best frame: #1.
15:52:06: Stacking will use registration data of layer 0 if some exist.
15:52:06: Stacking: processing...
15:52:06: Processing all images in the sequence (4)
15:52:06: Stacking result will be stored as a 32-bit image
15:52:06: Computing normalization...
15:52:06: With the current memory and thread (8) limits, up to 8 thread(s) can be used for sequence normalization
15:52:06: Reading FITS: file, 1 layer(s), 3358x2536 pixels
15:52:06: Reading FITS: file, 1 layer(s), 3358x2536 pixels
15:52:06: Reading FITS: file, 1 layer(s), 3358x2536 pixels
15:52:06: Reading FITS: file, 1 layer(s), 3358x2536 pixels
15:52:06: Using 13083 MB memory maximum for stacking
15:52:06: We have 8 parallel blocks of size 317 (+0) for stacking.
15:52:06: Starting stacking...
15:52:06: Median stacking complete. 4 images have been stacked.
15:52:06: Integration of 4 images:
15:52:06: Pixel combination ......... median
15:52:06: Normalization ............. additive + scaling
15:52:06: Pixel rejection ........... none
15:52:06: Rejection parameters ...... none
15:52:06: Background noise value (channel: #0): 42.449 (6.477e-04)
15:52:06: Saving FITS: file, 1 layer(s), 3358x2536 pixels
15:52:07: Execution time: 911.14 ms.

Hello. I’m sorry but siril is really not dedicated (yet) for this kind of image.
Try Planetary System Stacker which is free and open-source.

Thanks, that’s understandable.

Planetary stacking actually works pretty well for full-disk images of the Moon and planets. Would manual registration be practical for lunar surface images like the ones I’m trying to process?

PSS works well with your images.
Just use the surface algorithm in preferences.

Someday siril will do it… Someday… :slight_smile:

Isn’t it @vinvin :slight_smile:

What do you mean by PSS?

Planetary System Stacker

Got it, thanks. Planetary System Stacker looks great but my environment doesn’t meet the system requirements.

I’d like to try manual registration in Siril but so far I haven’t figured how to get started.

Maybe, one day…
Manual registration is probably not documented, but it’s full-image registration like other automatic registration methods of siril, while for planetary images, partial image registration is preferred, like with multiple points of alignment because the atmospheric turbulence is not coherent in the full image.
Anyway, the so-called planetary registration methods of siril will work with your images, but you must not use median stacking. Use sum stacking or average/rejection stacking, otherwise the registration data is not used.

1 Like

Thanks for the suggestion. I tried sum stacking, average stacking with rejection, and all the other options in the Methods dropdown menu but the result is similar to this screenshot.

Maybe the registration didn’t work then, do you see shifts in x and y in the images list of the sequence? It can be opened by clicking on the lowest-right button of the main window.

1 Like

No but with this kind of image Siril does not work well.
This is known and this is why someday we would like to develop the MPP branch.

What’s strange is that, after registering with the Enhanced Correlation Coefficient (planetary - surfaces) method, there is no shift in x or y. If the stacked image is so off I’d expect some displacement.

Because the algorithm fails.
As I said. Siril is really not good for that.

It is not got at it but it should still work, I have not tested this in months, maybe it’s broken…

On this kind of image, with low contrast it returns 0 because it fails to align.
Someday MPP will save us :).

But of course worth checking.

OK …

My bad. There is a bug…
I will fix it in master in a few minute.


Try PIPP & Autostakert.

Thanks but both tools require Wine, which may not be practical on Chrome OS.

Then PSS is the only and, for sure, the best solution for you :).