Stars alignment without tracking

RomainD · October 23, 2021, 6:41pm

Hello folks!
I’m new to astrophotography witthout any proper setup and I’ve been trying some experiments lastly : shooting with a smartphone attached to a pair of 8x42 binoculars.
I’ve taken 600 pictures of 2s exposure with dark and offsets and now trying to stack them using Siril.
All the steps done manually untill the alignment are fine but when trying to do the alignment it does not work : SIRIL does not find the stars to align, almost all frames are rejected (ex on 50 frames, only 11 are taken).
Is there any way to do this alignment knowing that the object is moving through the field of view?
I was hoping that the 1 star alignment with “follow the star movement” will help me out but this does not work as well.
Any help would be appreciated

Thanks!
Romain.

cissou8 · October 24, 2021, 5:56am

Hello Romain,

Several things you can try:

2-3 stars alignment. If your mount isn’t tracking, you will also get field rotation, and 1-star alignment can only detect shifts.
global registration. Select the reference frame as one towards the middle of your series. So that you maximise overlap between this frame and the frames at the start and at the end of the series. You said it would only include a limited number, have you checked that enough stars were detected? Have a look at this page: Siril - Image Analysis

Cécile

RomainD · October 24, 2021, 4:24pm

Hello Cécile and many thanks for your reply !
I’ve tried the different methods you recommended with only 4 frames.
The stars detection is good with about 70 to 90 detection and no rejected frames.
Though I end-up with duplicated stars in the stacked image as seen below.

alignment-stars

I’ve tried the 2-3 stars technique as well as the global with all the diffrent algorythms for the alignment.

Also, the 600 pictures are taken by batch of 50s then re-centered again so I don’t follow a regular rotation.

Any ideas?

cissou8 · October 24, 2021, 8:06pm

Didn’t realize you were recentering every now and then, so not sure the 2-3 star algorithm can work. But as long as global registration catches 70 to 90 stars, it should be enough.

Seems like some stars in the center are well aligned and the “duplicate” stars appear when we move off the center…I’m afraid this is caused by an important distorsion of your optics, a bit too much for what Siril can handle. Could you share 2 subs?

guille2306 · October 24, 2021, 8:51pm

I’ve seen this on non-tracked very wide field series (14mm in a full frame). Your problem is that for such great angles the field of view is usually not flat, so shift + rotation is not enough to align all the stars at the same time. You also need to correct for the warping (distortion) of the image at the edges, something Siril can’t do.

Hugin (or lensfun) can do it, but it won’t export raw data to feed back to Siril. I don’t know which is the best way to proceed on these cases. Calibrate the images, export debayered tif, correct distortion and align in Hugin, then import back and stack on Siril?

RomainD · October 25, 2021, 6:05pm

Ok ! Thank you both it’s more clear now on my side! I did not though that field distortion (and there is a lot!) would have such an effect on star alignment.

I would try to see with the soft you mentionned if there is a way to decrease this effect. What surpise me the most is that I stacked it with DeepSkyStacker and there were no issue at all (though the result is off course blurry at the egdes). This is the final stacked image I got from DSS.

Also, I really want to focus on M31 and will crop it anyway so if for you the center of the image is working well then I might just go with that

@cissou8 I’ve uploaded the files here - I put the few images I was trying to stack with so you get the same conditions. Many thanks for giving it a shot !

Again thank you both - it’s very fun to try to figure this thing out.

cissou8 · October 25, 2021, 10:51pm

Hi
downloaded your files, but I don’t understand at which step you got them. They are r_pp_brute files, which means that you have already preprocessed and registered them. Still, they are undebayered which is surely not good for registration…

What kind of preprocessing have you applied? And most importantly, could you share the raw files and the masters you’ve used to preprocess if any?

C.

RomainD · October 26, 2021, 5:18am

My bad I’ve sent the wrong files ! Here is the new download link

I tried again with some different raw files (same than the one I sent) and got some slighty improved results in the center of the field.

I have also ran again the alignment phase on the full file with the global method and got half of the frames rejected (305/602).

kityr · October 26, 2021, 6:12am

Maybe try the manual alignment function? It’s not that hard once you get a hang of it. Cool idea with the smartphone and binoculars btw lmao

cissou8 · October 26, 2021, 6:29am

Ok, makes more sense now. Don’t know what you did with your masterbias as it should not be sustracted from the lights (except if you also substract it from individual darks before stacking your masterdark, and I would not recommend you chose that path unless you’d want to “optimize” you darks. This is probably useless with 2 s subs).
So I went ahead and calibrates your lights with the masterdark only and applied debayering. I then chose one of the calibrated lights with M31 roughly centered as the reference frame and ran registration which was succesful for all of them. And then stacked. I’m getting decently round stars in the center, as shown below:

I can see on the r_pp_lights that some serious warping is applied, caused by the distorsion of the optics but the stars in the center should be ok.
I guess the problem with the registration comes from the fact that the calibration is wrong if you have substracted the masterbias as well. Makes star detection trickier.
One last think, there is some important light leakage on your masterdark. You may want to reshoot a series.
C.

PS: what kind of message do you get in the console when the frames are rejected from registration? Could you show the log?

RomainD · October 26, 2021, 2:41pm

Hello Cécile,
I cannot thank you enough for trying it yourself and walk me through it! Awesome

I’ll try it tonight and let you know what I got ; indeed I think I messed-up with the dark and biais - will recheck everything tonigh also. Will capture the log as well at that moment.

Romain.

RomainD · October 27, 2021, 12:14pm

Hello ! Tried last night, the results are better but far of course from what I expected, I guess this is coming from the number of rejected frames (300 / 600). I’ve also cropped the image to reduce this effect (which led to more rejected frames I guess).

For the log it said (in French):
14:08:10: Lecture du fichier FITS : cropped_pp_brutes_00149.fit, 3 canal(aux), 2504x2338 pixels
14:08:10: Findstar : en cours…
14:08:10: Fichier FITS enregistré : fichier r2_cropped_pp_brutes_00001.fit, 3 canal(aux), 2504x2338 pixels
14:08:10: Fichier FITS enregistré : fichier r2_cropped_pp_brutes_00343.fit, 3 canal(aux), 2504x2338 pixels
14:08:10: 80 étoiles trouvées dans l’image 149, canal #1
14:08:10: Impossible d’effectuer un alignement sur les étoiles : essai #3. Image 149 écartée
14:08:17: Lecture du fichier FITS : cropped_pp_brutes_00150.fit, 3 canal(aux), 2504x2338 pixels
14:08:17: Findstar : en cours…
14:08:17: 37 étoiles trouvées dans l’image 150, canal #1
14:08:17: Correspondance des étoiles dans l’image 150 : finie
14:08:17: 13 paires correspondantes.
14:08:17: Pts OK : 0.923
14:08:17: échelleX : 1.059
14:08:17: échelleY : 1.033
14:08:17: échelle : 1.046
14:08:17: rotation : -0.462 deg
14:08:17: dx : +304.43 px
14:08:17: dy : +58.00 px
14:08:17: FWHMx : 8.53 px
14:08:17: FWHMy : 5.99 px
14:08:18: Fichier FITS enregistré : fichier r2_cropped_pp_brutes_00150.fit, 3 canal(aux), 2504x2338 pixels

Which give (more or less)
14:08:10:Reading file FITS : cropped_pp_brutes_00149.fit, 3 canal(aux), 2504x2338 pixels
14:08:10: Findstar : in progress…
14:08:10: Fichier FITS saved : file r2_cropped_pp_brutes_00001.fit, 3 canal(aux), 2504x2338 pixels
14:08:10: Fichier FITS saved : fichier r2_cropped_pp_brutes_00343.fit, 3 canal(aux), 2504x2338 pixels
14:08:10: 80 stars found in image 149, canal #1
14:08:10: Impossible to perform stars alignment : try #3. Image 149 kicked
14:08:17: Reading file FITS : cropped_pp_brutes_00150.fit, 3 canal(aux), 2504x2338 pixels
14:08:17: Findstar : in progress…
14:08:17: 37 stars found in image 150, canal #1
14:08:17: Mathing stars in image 150 : finie
14:08:17: 13 match.
14:08:17: Pts OK : 0.923
14:08:17: échelleX : 1.059
14:08:17: échelleY : 1.033
14:08:17: échelle : 1.046
14:08:17: rotation : -0.462 deg
14:08:17: dx : +304.43 px
14:08:17: dy : +58.00 px
14:08:17: FWHMx : 8.53 px
14:08:17: FWHMy : 5.99 px
14:08:18: Fichier FITS registered : fichier r2_cropped_pp_brutes_00150.fit, 3 canal(aux), 2504x2338 pixels

Will try another night with some better focus / cleaned lens and proper darks

What would you suggest makes the best frames for this bino/smartphones thing ? Let the object drive through the lens for 200 frames or keep it with the 50s frames then recenter? I am wondering what is the best ratio for lost imagesdue to rotation / optics abberations due to edges.

Even if not working as expected, this is a lot of fun

RomainD · October 27, 2021, 6:26pm

Tried it with M45 this time, more easy to find the star and alignment worked pretty well ! Only 20 rejected

cissou8 · October 28, 2021, 9:15pm

No problem reading logs in French
I think cropping beforehand is not going to do any good as you may discard stars which are meaningful for the alignement process. One thing you can try though, with the non-cropped subs, is to increase the searchbox radius in the Dynamic PSF window so that stars away from the center also get caught by the star detection algorithm. They tend to be too large with the default setting.

I would suggest keeping the target as close as possible to the center to avoid heavy distorsion in the area of interest.
Have fun !

C.