What's missing from Siril?

[wise creek]

cd siril
git pull
ninja -C _build && ninja -C _build install```

just run the pacman -Suy every now and then (I run it every one or 2 months or so)

[merry perch]

ok thanks a lot

[wise creek]

and the meson reconfigure stuff seems a bit too much

[merry perch]

and for that you mean the whole block?

[wise creek]

sorry, didn't get that. Whcih whole block?

Ah the whole meson setup _build --buildtype release --reconfigure line?

[merry perch]

@wise creek I mean this

[wise creek]

Yes, if you want to be sure, You can run the whole block

but the pacman -Suy line may take long to execute, and it is not necessary to run it every single time

So you can skip it, and run it only occasionally

(it's an update of all the packages of the system, as if you were updating all the versions of all your software on Windows)

[merry perch]

Ok now I understand thanks

I don't remember if I mentioned it but in case it helps its Seestar subs

probably unrelevant but idk

[wise creek]

I had figured from the number of frames 🙂

No, the problem is with the fact that platesolving computes distorsion and registration tries to remove it. And at some point along the process, the map from distorted space to rectified space is cleared before the last image is processed

[merry perch]

hence the "segmentation" error?

[wise creek]

yep

[merry perch]

I noticed it also happens when stacking normal frames (so no sequence platesolving) but still choosing maximized framing

[wise creek]

@merry perch Do you get an error in the log saying something like "error reading one of the image areas"?

[merry perch]

In the console?

im gonna check

In the first test it crashed during stacking but it got completed but it looks like this, however yes it says "partial read from FITS file has been requested outside image bounds or with invalid size
"

ok doing it now it doesnt crash (gives that error though )but here's the output, this is with drizzle

[stark ermine]

looks like the issue I got @wise creek

[wise creek]

Yep, so the issue with stacking is known

I'll go back to trying sorting it tonight maybe. The one during registration will take a bit longer

[merry perch]

Could this be related to yesterday's issue where it crashed on apply registration?

[wise creek]

No it's not

[mellow vine]

@stark ermine OSC preprocessing without darks script does not work in Siril version 1.3.2 alpha dev

My folders are named correctly

With nothing changed, the script works just fine in Siril version 1.2.5

[tame surge]

Why are u using it if u have darks?

[mellow vine]

When did I say that I had darks?

[tame surge]

From the photo.

[mellow vine]

Are you talking about the darks folder?

[tame surge]

Yes

[mellow vine]

Because that's an empty folder

You still need it for Siril to work

There are no dark frames in it

[tame surge]

Huh

I never use this empty folder and it works just fine

[mellow vine]

I was always taught need 4 folders each named biases, darks, flats, lights. This was the way I was taught to do it and it works fines for me all the time

The issue with 1.3.2 is when it gets to the flats

[mellow vine]

Ok. then maybe it's changed. I'm used to needed all four folders

[tall sable]

pretty sure they changed syntax for scripts

recently

so it no longer understands what preprocess means

[mellow vine]

Ah

well, rip

[mental geyser]

need to go into the script and edit it

so anything with "preprocess" has been changed to "calibrate"

thats it and should be fine after that

that script isnt made by siril. its made by someone else. the OSC_Preprocessing is the one made by the devs and is updated

[stark ermine]

These scripts are not official and are not supported by new version

[tame surge]

Thanks bro I would also need this in the future

[safe shale]

maybe star trail reduction could be a good feature..?

[chilly marten]

Isn’t that just deconvolution

[tame surge]

Deconvolution

[stark ermine]

It is

[tall sable]

i got good news for you

[dim minnow]

Can we make temp file of siril become lesser?

[stark ermine]

https://siril.org/faq/#could-you-add-commands-to-delete-files-or-sequences

[mental geyser]

Yes this is what blurx does

[spring oxide]

I just make sure to delete the process directory when stacking is done. You really just need the results file after that point

[safe shale]

oh

[safe shale]

no

[mental geyser]

That is what deconvolution is

It changed thr psf of your stars

[safe shale]

deconvolution make stars smaller but don’t change the trails

[mental geyser]

It does

I've used blurx on my untracked images and makes an insane difference

[safe shale]

nice

I don’t think blurx is only using deconvolution

tho

[mental geyser]

Blurx is only deconvolution

[mental geyser]

are there plans to implement Mutliscale Gradient Removal?

[tall sable]

sirils deconv isn’t as good at removing trails as blurx is however

which im guessing is the point pirosow is making

[safe shale]

yea

[dim minnow]

Ive like 70 gig of raw data

Want to stack at siril

But... yeahh...

[spring oxide]

you don't have to stack at once. you can stack in batches if you're storage can't handle it

[dim minnow]

I dont have storage for it

[dim minnow]

But the result would be different right?

[spring oxide]

no, as long as you stack the stacks don't see why it would be

[dim minnow]

I see

Is it any technique

To do stack of stacks?

Median stacking

[spring oxide]

I think if they are all stacked with the same method it should work no matter the algorithm

[dim minnow]

Hmm i see

Thanks

[stark ermine]

It would be

Think about normalization, and pixel rejection

[dim minnow]

What does that mean then

It needs math right

[stark ermine]

This means that it's not equivalent to stack all the images at once, or to separate them into several stacks.

because you need to take into account the normalisation and the pixel rejection, that need to work on the whole stack

[dim minnow]

Hmm i see

Ive 70 raw gif of data, then how do i stack all of them without buying more storage again?

[stark ermine]

you can use 16bits format, or use compression in Siril preferences

[dim minnow]

Can you show the settings?

Does that make it 'lesser'?

[stark ermine]

You can watch the documentation:)

[wise creek]

the stacking problem is now solved, you can pull master and rebuild

[old timber]

are there plans to introduce masking and layers in siril any time soon?

[mental geyser]

coming in version 1.6

so a while away

[stark ermine]

Layers no. No interest for that.

[split smelt]

is there a plan to make it possible to use curves adjustment from the command line?

[stark ermine]

@ancient talon didn't do it because it made little sense. Moving control points on a curve is very visual.

[split smelt]

you could argue the same about stretching

[stark ermine]

not exactluy

stretching has 3 points

curve can have .... many more 🙂

[split smelt]

yeh, I was looking at doing something similar to a 3rd party pix script. They define the curves as x,y coords .e.g P.HS = [ // x, y [0.00000, 0.00000], [0.04910, 0.00909], [0.07235, 0.00909], [0.10594, 0.15455], [0.19380, 0.00909], [0.37726, 0.00000], [0.52972, 0.00909], [0.60465, 0.13636], [0.68475, -0.00909], [0.84496, 0.00000], // row 10 [1.00000, 0.00000] ];

would be nice to have similar to make it easier to port such scripts

[merry perch]

I tried stacking normal frames with maximum framing and it works fine now. The mosaic frame stacking (so with platesolving) still crashes but I suppose thats a separate issue to solve

Also I was wondering, since I was trying to platesolve , that some images still fails but they are using Siril's internal solver, I saw reccomendations to use local catalogues but I didnt quite understand how

Is it like astrometry.net

[wise creek]

yep, the platesolving issue is smthg else

[wise creek]

https://siril.readthedocs.io/en/latest/astrometry/platesolving.html#using-local-catalogues

[merry perch]

Would this platesolve more frames than Siril's internal solver?

[raven lodge]

My experience is no, though I don’t know why. Last time I tried it (several months ago) a local astrometry.net solver failed on the same images Siril’s internal solver did.

[ancient talon]

I thought it wouldn't make sense but I can add it sometime soon. Seems fairly easy to implement.

[tame surge]

hello, is there an HDR script for siril?

[stark ermine]

https://gitlab.com/free-astro/siril-scripts/-/blob/main/processing/WAF-HDR_brightness_weighted.ssf?ref_type=heads

[tame surge]

thankss

[tame surge]

What does this mean May I ask?

[tall sable]

i got that error when i accidentally applied registration when trying to drizzle

obviously if the stars are aligned with eachother drizzle wont work

but not sure how that error applies here

[tame surge]

Oof

Thanks anyways

Gosh what is this

[tall sable]

ik what the issue is

it only registered 1 image and the other ones failed

its probably failing cuz there is only 1 image in the sequence

[tame surge]

Why is that bad? Do I need multiple images?

And also, is the fact that it’s starless an issue?

[tall sable]

well what does the script actually do

is it for stacking

[wise creek]

This is siril internal solver. It will just be faster if you are going to process mosaics regularly. It would also remove the need to tick "fetch catalogues for each image" (and the time spent fetching those catalogues). You could also enable near search which may help finding solutions for more images (warning with this one, it can be long...)

But first, can you check if the Seestar writes diffrent RA dec for each image of a mosaic session?

You can use the command seqheader to do that (on a sequence of only original images, not already solved by Siril obviously)

[raven lodge]

Yes it does.

22:30:30: Running command: seqheader
22:30:30: Image 1, ra = 11.083335, dec = 41.316111,
22:30:30: Image 2, ra = 11.083335, dec = 41.316111,
22:30:30: Image 3, ra = 11.083335, dec = 41.316111,
22:30:30: Image 4, ra = 11.083335, dec = 41.316111,
22:30:30: Image 5, ra = 11.037495, dec = 41.315278,
22:30:30: Image 6, ra = 11.004165, dec = 41.315,
22:30:30: Image 7, ra = 10.975005, dec = 41.315,
22:30:30: Image 8, ra = 10.95834, dec = 41.316111,
22:30:30: Image 9, ra = 10.94583, dec = 41.316944,
22:30:30: Image 10, ra = 10.933335, dec = 41.318333,
22:30:30: Image 11, ra = 10.92084, dec = 41.319444,
22:30:30: Image 12, ra = 10.92084, dec = 41.319444,
22:30:30: Image 13, ra = 10.92084, dec = 41.319444,
22:30:30: Image 14, ra = 10.92084, dec = 41.319444,
22:30:30: Image 15, ra = 10.929165, dec = 41.175,
...```

Some of the coords appear to be the same, because it does up to 5 frames in the same position before dithering or moving on to another part of the mosaic.

[wise creek]

ok, then if you can send me one of the images it fails to solve, I will have a look when I get back to making the last touch-ups to astrometry before we release

[raven lodge]

This one fails when trying to plate solve as part of the whole sequence.

However, when solving this frame individually it resolves fine with this output:

22:41:30: Findstar: processing for channel 0...
22:41:30: Using 72 detected stars from image.
22:41:30: Using already downloaded catalogue NOMAD
22:41:31: Fetched 486 stars from NOMAD catalogue
22:41:31: Up is +169.60 deg CounterclockWise wrt. N (flipped)
22:41:31: Resolution:      2.377 arcsec/px
22:41:31: Focal length:  251.62 mm
22:41:31: Pixel size:      2.90 µm
22:41:31: Field of view:    42' 47.41" x 01d 16m 4.28s
22:41:31: Image center: alpha: 00 41 35.372, delta: +41 16 35.883
22:41:31: Was 16.34 arcmin from initial value
22:41:31: Saved focal length 251.62 and pixel size 2.90 as default values```

Not sure if this is relevant, but using Windows with symlinks enabled. When plate solving the whole sequence, this file's .fit (as with all the others that failed) was still a symlink, whereas all the ones that solved correctly became proper files. Even after solving this frame manually, the file did not become "real"

[wise creek]

yeah, we've done to protect original files. As writing the astrometric solution modifies the header, we re-write them as a separate fits file if the astrometry has suceeded during the seqplatesolve

Even after solving this frame manually, the file did not become "real"
It should become "real" if you save it

or maybe not... I can't remember this one. I know I spent a lot of time on handling sequences

not so much individual files

[wise creek]

Have you tried either with ticking "fetch stars for each image" or with local catalogues installed? for the sequence platesolving I mean

[raven lodge]

I just deleted everything, copied over the originals and tried again with "fetch stars for each image" checked. A few more images were solved this time (including the one I posted), but still many that did not solve, including this one

[wise creek]

But does it solve if you try individually?

[raven lodge]

Hmm. That one, no. It needs the downsampled checkbox check for it to resolve.

However, this one and a few more can be plate solved individually without that checkbox checked.

[wise creek]

Can you send me a log of the sequence platesolving?

[raven lodge]

The last thing I did there was delete the contents of process and lights, copy over the originals into lights, add them to a "lights" sequence without symlinks and try platesolving.

[wise creek]

Can't see anything special in the log. Will have a look when I get back to astrometry

thanks for sharing the files

[raven lodge]

No problem. Let me know if there's anything else you need.

BTW here's all the original frames straight from seestar. There will be a few frames that won't solve due to clouds or obstructions of some kind. https://1drv.ms/u/s!ArEUZ19_cutJmdM-oGaqWQyW29-JlA?e=yc6vsb

[wise creek]

Just tested with your sequence. With the local catalogues installed, I solve 980 images out of 1011. If I also enable nearsearch (this is only available with local catalogues installed), it solves 1010 images.

[merry perch]

in my case it says that all images were platesolved in the end even if there are "image X didn't solve" in the console

ok this should be the right number after platesolving

are these settings optimal?

[wise creek]

I did not test your sequence, I tested @raven lodge 's

[wise creek]

No, does not seem you are using local catalogues

And in that case, you can't use nearsearch

[merry perch]

ok, also i dont remember but im pretty sure the images were run through ASTAP earlier, does that give any problems in the sense that it modifies the header?

[wise creek]

No, it's not a problem

[raven lodge]

Ah thanks, I didn’t spot that near search is only possible with local catalogues. I’m still not solving 5 of the images, but 0.5% of subs failing to register is a lot better than 5% or more.

[raven lodge]

In addition to the local catalogues, try setting down sampling on. It seems to catch a few more. Also check the preferences to see if there’s a timeout for platesolving, I set it to 0 because some would simply time out.

[wise creek]

Actually, you should not downsample on images which have not been debayered. I will fix that. If you want to improve detection, try lowering the threshold to say 0.5 in the Dynamic PSF window

[merry perch]

So I detect stars in dynamic psf before doing the platesolving?

[wise creek]

When images are not debayered, we fetch only the green pixels and interpolate the missing pixels. Downsampling does not do that so whether or not it improves detection is pure luck

[merry perch]

Like I do it every time for registration why didnt I think about it

[wise creek]

Just set 0.5 for the threshold and that should be fine

[merry perch]

Do I hit the "detect stars"? As with usual registration?

[wise creek]

I'm not sure the dynamic PSF does the green pixels extraction thing, I will need to check consistency accross the different processes

[wise creek]

just setting 0.5 should be enough

pressing detect stars won't hurt

Just to be clear, I still have this list to look into before we release, but I'm busy with the registration right now
https://gitlab.com/free-astro/siril/-/issues/1406

[merry perch]

@wise creek I also downloaded the catalogue from the magnet link, I pu the files where they were requested but now how do I verify its properly installed and working?

Gave it a test run and only failed about 30 images this time

also enabled near search

so it technically works way better

That should be enough to tell its installed?

And also yes I put 0.5 in the detection treshold

[wise creek]

in the platesolver dialog, you should see smthg like:

(see the "local catalogue" mention on the right)

anyway, if you had a control about enabling/disabling nearsearch, it means it has detected the local catalogues

[trim steppe]

would it be possible to get a feature similar to multiscale median transform on pixinsight?

[mental geyser]

A way to do MSGR would also be great

[trim steppe]

that would be for msgr, siril has features that cover the rest

[mental geyser]

Oh yea lol

[merry perch]

Ok so now platesolving works fine, maximized framing stacking (with no platesolving) also works fine. The only thing thats preventing me from stacking the mosaic frames for now is that segmentation fault crash when trying to apply the registration on the platesolved sequence

[stark ermine]

could you provide backtrace with gdb?

[merry perch]

@stark ermine already did as I was guided by @wise creek and got this (look at answered message), so I'm pretty sure he already knows the issue

[stark ermine]

oh ok

oh it is releated to distortion

[merry perch]

Is there any setting I could try on my end to see if it persists or is it just a general issue

[stark ermine]

no. You just have to wait that Cécile fixes the bug.

It's her code, so it's best to leave her to it.

[wise creek]

Wondering if it's not linked to the drizzle bug in fact. If one reg fails (because of drizzle) it may clear the distorsion map before all the threads have completed

[stark ermine]

the bug fixed in the Adrian MR?

[mental geyser]

https://youtu.be/23rZ92g4Fzs?si=SbsVACfXtudaHukl

[stark ermine]

yep, this is me

[stark ermine]

Could you try this artifact: https://gitlab.com/free-astro/siril/-/jobs/8524846323/artifacts/download

[merry perch]

I'll try it today if I get the time

[stark ermine]

or, if you prefer to compile, try the branch drizfix

git checkout drizfix

and the compile as usual

[merry perch]

So like normal installation but instead of -release i just do "drizfix"?

[stark ermine]

no

git checkout drizfix

To change the branch

then after that, you can compile as usual

to come back on the normal branch, after your tests

git checkout master

(before git checkout drizfix you can run git pull)

[merry perch]

Ok, so after the git chdckout I run meson setup _build --buildtype debug --reconfigure and then ninja -C _build && ninja -C _build install

Not debug, but release right?

So as if Im normally updating it

[stark ermine]

we don't care about debug or release yet

just try to see if it xrashes

but do it:

git checkout drizfix```

ninja -C _build && ninja -C _build install```

[merry perch]

Ok thanks for the indication. I'll try as soon as I can

[merry perch]

Ok now it doesn't crash upon registration. When stacking though i constantly get this "normalization failed, check image X" error. And when I check the image it is completely black. If I uncheck it will continue but will find another image with the same issue

the images in question look like this

[floral plaza]

Is it possible to set multiple points (other than the symmetry point) on the curve adjustment in siril? Kinda like what you can do in photoshop and pixinsight

maybe im not looking in the right place

[ancient talon]

it is?

btw why are curve points not visible in your Siril screenshot? Is this something that happens often?

oh wait no this is GHT lol.
There's a Curves Transformation tool in Siril alpha, very similar to the one in Photoshop

[floral plaza]

I cant seem to find it :(

you mentioned its in Siril alpha, maybe a suggestion could be to add it to newer versions of siril? I am on the latest, 1.2.5

[chilly lodge]

Lots of discussion in this thread. To come back to the original poster; I’m just starting out with Astrophotography and siril is a lot. It would be very useful if there would be a “noob” interface. There are a few steps everyone takes to process planets apparently (found multiple instructors and they all do the same), so that could be a very simple flow where all the other things are clutter. Maybe for dso this is possible as well.

Something more workflow-oriented would help.

As a bug report; when I split colours and then recombine them, this fails 3/4 times. Missing information for recombining.

[stark ermine]

Curve transformation is on the dev branch

[stark ermine]

https://siril.org/tutorials/rgb_composition/

[floral plaza]

Not entirely sure what that means, could you elaborate please?

[stark ermine]

dev branch = development branch. Not released yet

[floral plaza]

Ah alr so it's coming out sometime in the future

Ty!

[stark ermine]

https://www.youtube.com/watch?v=tZPl78Cop1A&t=37s

[chilly lodge]

I can’t use this information. I recently downloaded siril. If I first do rgb split in siril into 3 separate images and then combine those back together, then sometimes there is a registration in the images and sometimes there isn’t. Sounds like a bug to me. I’m just following the exact same steps.

[stark ermine]

I always split and recombine. never saw a bug like this

[merry perch]

@stark ermine as stated here with the drizfix branch it doesn't crash anymore but does what mentioned in the previous message on stacking. Do I need to provide more info with gdb or is that enough?

[stark ermine]

Be sure to try with the last version of the branch

Because we have a branch dedicated to stitching now

[merry perch]

How is it called?

Same procedure as drizfix to install right?

[stark ermine]

stack_blending

Yes same procedure

[merry perch]

with the new version it gives the same normalization fail error

but its the same images that fail

i could go on trying to uncheck them all

they just keep failing

these are my stacking settings

used drizzle for integration

I'm going to try interpolation

[stark ermine]

with interpolation you will have no issue

[merry perch]

When registering with interpolation siril crashes as before and this happens when running with gdb

[wise creek]

yeah, ok, so this is not triggered by the bug in drizzle, I'll look into it when I'm done with what I'm currently doing

thanks for the tests

[merry perch]

Which is a bit weird cause someone got it working on commit f0a1355bf with Seestar frames https://youtu.be/np9Ld4oA1S0?si=hQt0xIHLOwxkh3R1

While I tested it on the latest stack_blending version

[wise creek]

I got it working as well. It's probably a memory leak somewhere so very much dependant of the machine which is doing the processing

[merry perch]

Ah interesting, could it be an Intel or AMD dependant? I have a Ryzen

[wise creek]

and it's not linked to the work done in the stack_blending branch

[wise creek]

no, I have a Ryzen as well. Just somethg to do with how we compute the required memory

[merry perch]

Should I test with the version in the video specifically or would the same thing happen?

[wise creek]

No, the same will happen

I'll tell when I get back to working on this one

[merry perch]

Ok, thanks for your hard work

[wise creek]

If you want to test and avoid the crash occurring when applying undistortion you can:

• debayer the sequence
• solve it setting the solver to Linear instead of Cubic (and untick Skip already solved images just to be sure)
• register and stack
  I would advise you built from the branch of this MR instead of the one for stitching, because blending is useless for the Seestar mosaics. Whereas this one optimizes the rotation angle of the max framing (will be faster to stack as there are smaller black zones).

[merry perch]

Thanks I'll try this

[merry perch]

I'm currently stacking, so it did not crash, and should complete just fine, its just taking quite a bit of time to do rejection stacking but it could be normal since this is a debug build maybe

[merry perch]

ok stacking is complete, it took over an hour for 1300 frames but I'll try again and see

really good

[stark ermine]

GESDT is better as rejection algorithm with 1300 images

[merry perch]

ok thanks for the advice

also when I installed the mosaic_max_framing branch I got a few errors before installing which i didn't read properly, and I ended up on commit e6c027ffe, which I can't remember if its from the previous stack_blending branch i was on or its the right version of the new mosaic branch

[stark ermine]

you should show us what errors

[merry perch]

ok just in a second

[merry perch]

here

so here's why I had the doubt I ended up on the right version

[stark ermine]

no because it did not compile well here

[merry perch]

I guess I'll try going back to master then back to stacking to see what happens

Ok tried again gives the same error

[wise creek]

rm _build/src/libsiril.a

it's a bug msys2/mingw64

[merry perch]

thanks this worked

stacking already went well in that version but I should probably try again here

[stark ermine]

ok sorry. I'm not a windows user 🙂

[stark ermine]

If volunteers wish to participate in the translation of Siril (documentation and/or software).
You can now go to the following address: https://weblate.siril.org/projects/siril/

Your help is appreciated.

[broken peak]

could be fun to do in off times, I'll save the page

[mental geyser]

i broke siril

or siril broke itself

[split smelt]

or perhaps the Gaia archive is down?

[merry perch]

Got this too today

[stark ermine]

I've no issue

[stark ermine]

ok: https://gitlab.com/free-astro/siril/-/issues/1448

@split smelt : we should really think about continuing the release of an offline version 🙂

[split smelt]

I was thinking that too. I'll dig out my previous work and see if I can integrate it somehow

I was working on a NINA plugin, but I guess that can wait 🙂

[stark ermine]

yeah 😄

you should focus on including the xp_sampled data for now: it's a bit more data than the xp_continuous data, but the conversion of xp_continuous still looks very tricky.

[stark ermine]

@split smelt : message of Adrian, working on SPCC:

OK. It looks like along with the change from COMBINED to RAW they have changed the structure of the FITS file returned. Previously there was a HDU for each source; now there is a single BINTABLE HDU with a row per source and the flux is contained in 343 adjacent columns. It means get_xpsampled() needs changing, but it's not a horrendous change. Just looking for the same data in a different place in the file. But it's more than just a URL change so definitely adds weight to the importance of an offline version during the next development cycle.

So that explains why SPCC do not work anymore

[split smelt]

@stark ermine when I did this before I limited to mag 15 or less. Is this still good?

[stark ermine]

For a first test yes

[split smelt]

we could also consider a mechanism where we break up the offline files into much smaller chunks, e.g. areas of the sky and/or magnitude and get Siril to auto-download them the first time they're needed. It would mean that everyone wouldn't need to download gigs of data in advance

[stark ermine]

I think we should proceed like PI. Providing two different sizes of database

For wide and narrow field

[mental geyser]

100%.
On my old wifi, spcc would take about 5 minutes to get done. Now it gets done within a minute since I've gotten new internet

[stark ermine]

ok good news, SPCC will be back in a few minutes

Of course you will need to update the sources

[wise creek]

I think I've found it and solved it in this branch: https://gitlab.com/free-astro/siril/-/tree/fix_mem_reg?ref_type=heads

If you can test

[merry perch]

I'll try when I get the chance

[wise creek]

thanks

[merry perch]

I tried on the new version without the steps you told me to avoid the crash to see if it still happens and it did, I built the debug version and the console says specifically to "report this bug"

Now I'll try with the steps you told me, that avoided the crash and stacked succesfully in the previous branch

but it doesn't say segmenation error anymore

[merry perch]

wait maybe it would have been more useful if I ran the gdb

Ok wait, now I reopened Siril, reloaded the platesolved sequence and registration was applied succesfully

I tried again creating a new sequence, used the default settings, not the ones to avoid the crash, and now it worked again

So that crash only happened once

[wise creek]

ok, thanks for all the testing. Quite sure that first crash is unrelated to the initial problem

I'll merge the branch then

[mental geyser]

This is on the weekly release that was posted today for 1.3.5

also for windows, is it possible if we could get a fully black UI along with keeping a dark grey UI?

[stark ermine]

No. We use adwaita theme. There is only two possibililty. Dark and light

[mental geyser]

no black/midnight option?

[stark ermine]

nope

[mental geyser]

thats a shame

[stark ermine]

we know. This is the message we wrote in our new notification system.

[plain birch]

trying to use SPCC, but this message keeps coming up in console, any input on what this may mean?

ik this isnt the page for it, but talking to an actual dev would help i think (hopefully)

and also if there are/arent servers accessible, is there a way to download a cache of some sort of backup from the last time we used SPCC?

[trim steppe]

download a newer dev version, think something got changed with gaia stuff

[plain birch]

where can you get newer dev version?

[trim steppe]

https://gitlab.com/free-astro/siril/-/pipelines?page=3&scope=all

get the one where all circle thingies are green checkmarks

[plain birch]

?

[trim steppe]

yep

then win64 native installer

[plain birch]

i forgot do you have to uninstall?

[trim steppe]

wrong one, not the installer

[trim steppe]

no, it will do it automatically

[plain birch]

6th one right?

[trim steppe]

yes

[plain birch]

is there a way to keep up to date when a new dev release comes out?

[trim steppe]

idk, just check the pipelines channel on gitlab and see if theres a new dev build

[plain birch]

do you know how to chose what specific startype for SPCC rather than just choosing the generic average spiral galaxy?

tried lookin on their documents, but they just say what kind the sun is

[trim steppe]

pretty sure nearly everyone uses the generic spiral galaxy one, doesnt really need changing

[plain birch]

well then i guess i have a suggestion for siril, SPCC star type identification tools, or some sort of document/recommendation that can help you figure out what you should ideally be using as a reference point

[mental geyser]

@dusk grotto scroll down that list and find this

[dusk grotto]

kk

found it

[mental geyser]

@dusk grotto this one

then youll be able to download siril from that easily

[dusk grotto]

if i could access it

[mental geyser]

hm interesting

[trim steppe]

builds over a week old cant be downloaded

[dusk grotto]

bruh

[trim steppe]

on the link i sent before theres one that was 6 days old, you might be able to get that

[dusk grotto]

it worked daddy

[trim steppe]

🎉

[dusk grotto]

it is soooooooo cool

[mental geyser]

mosaics arent yet in but will be soon apparently

[dusk grotto]

surely vc for once

[mental geyser]

[dusk grotto]

come on

#1223447961036394626 message

[stark ermine]

They are now

[mental geyser]

Oh sick

[trim steppe]

little MSGR script for siril, not ideal but should work, the way its written is hella borked and i could prob make it way simpler but i cba rn

instructions are in the file, just gotta open with notepad

[trim steppe]

(fixed some typos)

[merry perch]

Since I updated to the among the latest dev builds I noticed stacking (both mosaic and normal) became significantly slower, like it used to take a few minutes while now at least over 20. Is this expected for the dev build?

Normalization and registration takes as usual

[stark ermine]

that depends which option you use

If it is with amx framing and seamless option, and normalisation in overlaps, so yes

[merry perch]

Overlaps is disabled. But max framing makes it slower with the new mosaic stacking method?

[wise creek]

feathering option will make it slower yes, because some of the memory is consumed to compute the masks

[trim steppe]

fixed more stuff, still quite borked

[dusk grotto]

Honestly all the new beta needs is blurx

[mental geyser]

grAxpert

[dusk grotto]

?

[mental geyser]

Graxpert decon

[stark ermine]

blurX is not something we are intended to integrate 🙂

[dusk grotto]

https://tenor.com/view/the-wood-movie-omar-epps-gif-25850084

[mental geyser]

Graxpert deconvolution

[dusk grotto]

Yeah but that requires downloading new one

lol ok I’m just lazy lol and forgot it exists

[stark ermine]

https://siril.org/2024/12/we-need-you/

[trim steppe]

cant say i have the characteristics of my cameras' sensor

[stark ermine]

The web has

[limber parrot]

will deepsnr be ever added?

[trim steppe]

not finding em for my canon eos m5

[stark ermine]

No. We are not adding 3rd party program anymore. Starnet and GraXpert are the only one

[trim steppe]

what about multiscale median transform 👉 👈

[trim steppe]

or, actually, would it be possible to do the median filter at a higher size?

[stark ermine]

Gaussian filter is not good enough?

[trim steppe]

idk, i tried it for the MSGR script im working on and it was quite borked in the end

but i did have it mixed with median filter, guess ill try just gaussian

yeah no gaussian just isnt gonna work

it just blurs real wide, which is what its for ig but the way the median filter does it it essentially yeets outs stars and small structures, not sure how to explain it

[merry perch]

I'm not using feathering, is that still normal timing?

[merry perch]

here's the settings and how much it took

[stark ermine]

@wise creek did some benchmark. Siril 1.4 is a little slower, but just a little, nothing very significant.

[merry perch]

then what I'm experiencing is because its a mosaic?

[stark ermine]

I don't know. I'm even not sure to understand what you are experiencing in fact 🙂

[stark ermine]

well, you can use wavelets

[merry perch]

pretty long stacking times of over 1 hour, when it used to take a few minutes on stable. But it could be the mosaic. I don't remember if it takes the same for normal stacks though

[stark ermine]

mosaic should be applied to stack result to go fast

[merry perch]

I'm stacking seestar mosaic subs. what does that mean?

its over 2k subs

It works I just wanted to know if its normal that it takes that long

[stark ermine]

do you use FITSEQ?

[merry perch]

no, normal FITS

less than 2040

[stark ermine]

yes ok. In fact it does not work with FITSEQ, I forgot

Well, I can't tell you, I don't know

[merry perch]

is there any plans to make it work with it?

[stark ermine]

Not really, no. It requires files with different frame sizes and makes management rather difficult.

[merry perch]

ok that makes sense

Today I updated to the latest commit in the master branch and upon trying to platesolve the second sequence of my stack it crashes for some reason. On the same version it worked before but with these frames it doesn't somehow. I should try to change from Cubic to Linear though, as that worked to avoid the crash in previous versions

[stark ermine]

There have been no recent changes in astrometry

[merry perch]

When switching the solver from Cubic to Linear it works though, but weird because for other frames it worked just fine

[trim steppe]

cant say those gave the results needed, so far the best results have been with median filter :/

[dusk grotto]

wait so you have to send filter data?

[stark ermine]

?

[dusk grotto]

sorry that had no context

Just misread that link you sent

[split smelt]

yeh you have to send filter data. It's compulsory or you won't be allowed to download Siril 😛

[stark ermine]

@split smelt cheated a bit by sending Gaia data. But that's ok

[split smelt]

I also sent data for the IMX585 🙂

[stark ermine]

indeed 😄

I hope we will have contributors

[split smelt]

or, alternatively, everyone can donate their kit to me and I'll do it for them 🙂

[stark ermine]

that could be great 😉

[trim steppe]

id like to keep my gear for the off chance that its clear

[split smelt]

yeh, not this year

[mental geyser]

tthere is no command line version yet

[dusk grotto]

wait so can it be any data or does it have to be data with filters on, i am illiterate

[wise creek]

can you share the image?

[merry perch]

the subs?

[wise creek]

regarding speed for stacking, max framing will take more time than you are used to, simply because the final image may be 2-3 times bigger than your regular images if you are stacking mosaics

also, make sure you are not in debug when timing the performance

[merry perch]

ah thats why

[wise creek]

yep, just the one image that seems to crash

[merry perch]

Ok give me some time and I'll send it

[wise creek]

no rush, about to go out. Just tag me when you send the link. Will try to look into it later. Seems to be the same error as this ticket: https://gitlab.com/free-astro/siril/-/issues/1449

[merry perch]

So idk if its the 00140 image, as in the log its the last one that says "reading" before the error occurs, or its the 01010, that its the last one displayed in Siril as "plate solving. processing image: " when it freezes

here's also that portion of the log

[wise creek]

I can't reproduce with any of the images... weird. Can you run again using only one core so that we identify exactly which image fails?

[merry perch]

You mean I should reduce the count next to disk space and memory stats from 16 to 1?

I'll do it when I can

[wise creek]

yep, this:

[merry perch]

Ok I will try later

[dusk grotto]

wait why is that

[wise creek]

the setting to change the number of cores used by Siril

[merry perch]

Ok I'm trying it right now Im waiting for it to crash

So once it freezes how do I know what image caused it

[wise creek]

If you can do a info locals in gdb

And probably, you would see it in the log because that would be the last opened jmage

[merry perch]

yea it appears to be 950

[merry perch]

should I still do this?

[wise creek]

Can't hurt

[merry perch]

after the crash?

[wise creek]

Seems more like pltest_00951.fit

[wise creek]

Yep

[merry perch]

ah yes thats the one in the log

[merry perch]

says this

[wise creek]

Ok

So if you can share image 951

I'll try to see if i can reproduce

[merry perch]

ok

[wise creek]

Thanks, I'll have a look later tonight!

[wise creek]

@merry perch , I think I have a fix

You can try the branch targetted by this MR

https://gitlab.com/free-astro/siril/-/merge_requests/785

no need to restrain the number of cores anymore

[merry perch]

so now it shouldn't crash?

[wise creek]

yep

[merry perch]

ok now platesolving completed succesfully and image 00951 was solved

[wise creek]

thanks for the testing!

[mental geyser]

whats up with spcc?

ive not touched my settings and its not liking it

[stark ermine]

do you use an updated version of Siril ?

[mental geyser]

this is the most recent dev version distribution copy weekly build

[stark ermine]

Here, Siril cannot compute the fit

[stark ermine]

weekly is old

you should use daily

[mental geyser]

alr

[stark ermine]

by compiling sources

[mental geyser]

when would the weekly come out?

[limber parrot]

how do i download siril from here https://gitlab.com/free-astro/siril

[dusk grotto]

is there a mask feature?

[broken peak]

afaik it's in the works

[dusk grotto]

sweeet

[mental geyser]

Coming v1.6

[dusk grotto]

Yippee

[stark ermine]

Not really. Scheduled for 1.6, not 1.4

[broken peak]

I didn't mention in which version it was planned for, I only heard that it was being worked on

[stark ermine]

but you can already work like masks exist in Siril

For example, by removing stars and using the recomposing tool

[broken peak]

I see what you mean, I think

[stark ermine]

example on how I've processed my last image

The result is here: https://telescopius.com/pictures/view/211117

[trim steppe]

bigger median filters

[plain birch]

hi i have a suggestion/feature request for siril, i know its a big ask but, if there isnt one already (and i could easily just not be aware of one) could we have some sort of anti haloing command similar to that of something like Seti Astro Suite?

[mental geyser]

I believe his deconvolution thing is there

[trim steppe]

siril seems to just exit for me when i try anything other than 1x drizzle

[trim steppe]

nvm, it was running out of ram

[worn pendant]

https://tenor.com/view/borat-borat-very-nice-verynice-thumbs-up-gif-25080066

[stark ermine]

Not our fault then 🙂

[sage glade]

Hey uhm @stark ermine is it normal for the Siril DEV VER to have 80 GB worth of data in my temp file under the users with this name on them?

Or should I detele them?

[stark ermine]

This is swap files. Used for the undo. In normal situation they are removed at normal exit

[sage glade]

I guess that makes sense to bad the dev version doesn't have 😦

[stark ermine]

Why the dev version wouldn't have it?

[sage glade]

Because that these files are from Xmas when I downloaded the new files.Now it doesn't matter anyway since I've deleted them so if it'll happen again imma ping in here

[rigid breach]

how do i download the dev version of siril, i googled a video and the guy has a download button on the github that doesnt show up for me. i cant find anything else

[trim steppe]

https://gitlab.com/free-astro/siril/-/pipelines?page=3&scope=all

[sage glade]

Thanks man you're real

[broken peak]

Usable builds have all green checks? That's what I still haven't fully understood

[rigid breach]

thanks so much, this should be pinned lmao

[stark ermine]

This is not what I recommand to download

here You are downloading a weekly build, and especially an installer

I would recommend to not download the installer, but the portable binary, in order to be able to have both versions (stable / dev)

@trim steppe

[trim steppe]

isnt the installer still easier

[stark ermine]

not really. The problem is you will only have the dev version isntalled. ANd this is a bad idea

keep in mind this is still a unstable version

[rigid breach]

oh damn well too late and im lazy

[stark ermine]

Mhh, for me this is an issue. Because will you update often? With the installer, I'm not sure

Sometimes, dev version can corrupt data... THis is why I warn you

dev version is for test, debugging purpose, not for production

this is why this is better to use the portable version

[raven lodge]

I presume that the r and b channels are "not ok" because the ha image is smaller than the others? Ha had the most field rotation and needed to be cropped for decent background extraction.

[trim steppe]

where do you get the portable version

[stark ermine]

you need to load bigger file first

[trim steppe]

how do you use it

[stark ermine]

you find the binay, you click on it. It works

[raven lodge]

So if the largest one is Sii, Sii must be in red?

[trim steppe]

it really does, will recommend that to ppl who want to try dev version then

[stark ermine]

no. You load it where you want, but in first

[raven lodge]

I just tried to redo the crop and background extraction again but ended up with one image the largest in X by a few pixels and another the largest in Y so it won’t work at all

[trim steppe]

cant you just make them into a sequence and then register on the smallest frame?

[stark ermine]

ideally, dont crop before the composition

[rigid breach]

what does this mean, i downloaded it and i can only find the regular release of siril in the folder

[stark ermine]

what do you mean?

you need to download the right branch

if the installer was the good version, the portable binary is necessary the right one too

[rigid breach]

idk what this means :(

[chilly lodge]

Where are you looking? (URL you used for downloading, what folder on your pc…)

[mental geyser]

[trim steppe]

shit makes no sense

[broken peak]

the jokes write themselves

[stark ermine]

https://discuss.pixls.us/t/updated-stacking-script-for-seestar-mimic-pi-wbpp/47717/5

[rigid breach]

just using the link and directions that @stark ermine showed in the screenshot

these are the folders

the bin folder has the siril executable but its the official release version not the dev

[stark ermine]

You haven't downloaded the right pipeline then

[rigid breach]

which is it

[chilly lodge]

Missing from Siril, or maybe I just don’t know enough yet… I have used itelescope for imaging today and there are weird artifacts in it. I have absolutely no idea how to remove it, but it definitely looks “not star or nebula”-like. Could that be removed in some sort of option?

I’ll add a photo of the weird artifacts.

I think the sensor of the telescope is borked…

[stark ermine]

please, don't use smarphone as screenshot

[chilly lodge]

Made a screenshot. Wasn’t on discord with my laptop before.

[stark ermine]

This is hot pixel you did not removed

You don't use dark frames?

[rigid breach]

ya looks like walking noise, whats ur cam

[chilly lodge]

I couldn’t download those yesterday. Server simply gave up.

Setup I used; https://support.itelescope.net/support/solutions/articles/231920-telescope-33

I’ll try to download and process those today.

[stark ermine]

Well' so everything is normal

[stark ermine]

Exciting news for the astro and Python development community!

I have an important update to share, and I encourage you to spread the word throughout the community.

The development version of Siril has just been upgraded to 1.3.6, marking a significant step towards the upcoming 1.4 release. Why is this a big deal? We've finally integrated a Python scripting engine into Siril!

So, to all the developers eager to create Python scripts for Siril: now's your chance! The API documentation is available here: Python API Documentation.

And if you're looking to understand how Python scripting works in Siril, check this out: Python Scripting Guide.

We need your help to test this new feature and, most importantly, to contribute to making Siril even better!

Please note: version 1.3.6 is a test version and does not yet have official downloadable packages.

I'd also like to thank @split smelt for its enormous contribution to the creation of the GAIA local catalog, which will be usable for astrometry, but also for photometry! I don't have enough words to thank you for this enormous contribution.

[mental geyser]

[trim steppe]

[junior hound]

Wow that's great !
I'm actually working on a python program for measurements using PSDs and stuff, I'll definitively look into the siril python scripting

[mellow vine]

How do I view my integration time from Siril?

I've got my fits stack and want to see how much data is compiled

And more preferably, I want to see how much data it used

Does siril have exposure tolerance in stacking? My subs have mixed exposure times by a margin of like 40 seconds

[merry perch]

Tools - FITS headers and find the "stacked frames" section

[mellow vine]

the option is greyed out

[merry perch]

Uh thats weird

If its a FITS it should have that

[mellow vine]

It's just in the file name actually

Just now noticed it (which is the same thing)

So 18.6 hours

[stark ermine]

Please send the screenshot of the whole application when it's greyed out

[mental geyser]

If you're using a newer script, it should just put the integration time in the file name

[mellow vine]

It does. I didn't notice it until now. Thanks

[stark ermine]

Yes but I am inrrested by what I've asked you

[mellow vine]

I will send a screenshot later today

[stark ermine]

yes please 🙂

[mellow vine]

The fits header option is included now

Not sure what as going on before

[stark ermine]

Here the behavior is normal

[merry perch]

Quick question, the symbolic link option allows to save storage by not having to copy the fits files numerous times?

Or am I understanding it wrong?

[split smelt]

Yes

[merry perch]

So it basically creates the sequence by copying the files once and then thats it? What if I drizzle?

[stark ermine]

No. This is just for the first step. For the conversion step.

[mellow vine]

@stark ermine hey why is it that in siril the image is displayed in 8-bit? it's very compressed. In pix usually I fix this by hitting the 24 bit stf option but unsure how to fix this in siril

It's not an issue with the data as it appears normal in other applications

This is my full screen

[stark ermine]

Use the HD autostretch

[mellow vine]

That did not change it

[stark ermine]

So this is your image

[mellow vine]

The data is not like this

[stark ermine]

Are you sure ? Then in preference enhance the bitdepth of the HD

[mellow vine]

Because in any other software there is no bit depth compression

Same data

Ok, I rebooted siril and the issue is gone

Not sure why but randomly certain tool windows just wont show anything

[stark ermine]

what version of Siril do you use?

[merry perch]

Then if I drizzle for example its going to naturally create more files?

[stark ermine]

Symbolic links are only for conversion. Not any other process

[tall sable]

is there supposed to be a deconvolution tab? im on siril 1.3.6

and i do have the latest graxpert.exe, the one including stars and object deconv

[trim steppe]

uh oh

[tall sable]

why uh oh

[mental geyser]

You don't have the Decon in siril

[trim steppe]

i very much had the tab in 1.3.5

[mental geyser]

Ya thas odd

[tall sable]

i had it in a previous dev version and it's now gone after i updated

[stark ermine]

We have removed the tab

[trim steppe]

why

[stark ermine]

because this feature is not going to be integrated in GraXpert stable soon. It will stay in the beta version. So we cannot support a feature that is not officialy released

[mental geyser]

Oh interesting

[tame surge]

when will we get multiple nights stacking built in

[stark ermine]

You already can do it. since the begining

[tame surge]

Don’t u have to use Sirilic? Am I missing something?

[stark ermine]

SIrilic is just calling Siril

If Sirilic do it, then Siril can do it

https://telescopius.com/pictures/view/205383/deep_sky/ngc-206/m-31-with-the-oiii-arc/by-lock042?gallery_user_id=16193&gallery_order=is_featured&gallery_order_asc=false&gallery_page=1 : my M31 is more than 10 night session

And I do it very easily, manually, with Siril

[tame surge]

Oh really nice image wow

Alright thanks

[trim steppe]

i'd imagine it's also possible to automate with scripts

[stark ermine]

of course, this is what sirilic is doing

And with python, this will be even truer

[trim steppe]

hmmm

[tame surge]

But how exactly do u do this?

[trim steppe]

i'd do stuff with python but i have very little experience with it

could prob just make a script with the default siril commands and have it shuffle between organized folders to get it done

[tame surge]

How did the msgr script go btw

Been wanting to ask I’m curious

[trim steppe]

id want bigger median filters still

[tame surge]

Yea it was very noisy

[trim steppe]

this is where i left it, havent even tested it

[tame surge]

Isn’t Gaussian blur good?

[tame surge]

Will try if I can, thx.

[trim steppe]

doesnt work for it

[tame surge]

Would be nice to get some Orion data from my b9 in broadband

[tame surge]

Oof

Sad

[trim steppe]

(please give bigger median filters lock 🙏)

[stark ermine]

not really needed now. IT can be down through the python script

https://gitlab.com/free-astro/siril-scripts/-/merge_requests/21/diffs

[trim steppe]

is it not limited to what siril currently has?

[stark ermine]

no

this is done by opencv in the script

[trim steppe]

i cannot even figure out how to download it

nvm

[stark ermine]

download what?

[trim steppe]

that mmt script you sent

is there any way to call a python script through the siril script commands or is it better to just have a full python thing at that point

[stark ermine]

this is just python script. If you use Siril 1.3.6, just copy and paste in the editor. But this script was just to give you an example

[tall sable]

makes sense 👍

[stark ermine]

If you, too, would like to help Siril grow up. And if you speak a language other than English, then it's possible to take part in the translation.
Greek, for example, is only 34% complete.

You can do it by registering to this website: https://weblate.siril.org/projects/siril/

And this is the same for the documentation

[mental geyser]

i'd love to help, if i knew another language

[stark ermine]

@trim steppe is not from another country? 🙂

[trim steppe]

perchance

[junior hound]

Why does Graxpert only has bkg extraction and denoising in the latest siril dev?

[stark ermine]

Because stable version of GraXpert only has bkg extraction and denoising

[junior hound]

Ah okay, so which siril version had all the graxpert stuff?

I'm kinda lost in the siril dev versions tbh

[stark ermine]

None. We have removed it as it will not be implemented in a stable version

[merry perch]

In the sense that it wont be implemented soon or that it was scrapped by the developers?

[stark ermine]

GraXpert developers don't want to integrate it at the moment.

[merry perch]

Is there any way to estimate how much storage applying registration with drizzle is going to take before starting?

[stark ermine]

If drizzle 2x. Your images will be 4 times larger. So you can estimate

[junior hound]

I'm trying to make a script to automate my process but when cnverting the lights it just doesn't work because "no files were detected"

cd lights
convert light -ser -out=../process
cd ../process
calibrate light -bias=master_bias -dark=master_dark

register pp_light -2pass -transf=affine
seqapplyreg pp_light -drizzle -pixfrac=0.4 -kernel=gaussian

stack r_pp_light rej a 3 4 -fastnorm -out=test_1```

This is my script, I have a ser file with my images in the lights folder, which itself is within the working direcctory

(And this is the error) 16:46:34: Setting CWD (Current Working Directory) to 'C:\Users\raphb\Desktop\SharpCap Captures\2025-02-03\M99 (NGC 4254)\Script\lights' 16:46:34: Running command: convert 16:46:34: No files were found for conversion 16:46:34: Error in line 4 ('convert'): generic error. 16:46:34: Exiting batch processing.

[stark ermine]

SER files dont need to be converted

[junior hound]

Okay I didn't know that. But now I juste have to apply the commands with the name of the ser file at the sequence name argument?

[stark ermine]

Yes

[junior hound]

Okay I got it to work but after applying registration it crashes immediately without any error in the logs. The reigstred sequence is properly saved but the stacking doen't start. Am I doing smth wrong with the script or is it a bug?

############################################
STACKING
############################################

calibrate data -bias=master_bias -dark=master_dark

register pp_data -2pass -transf=affine
seqapplyreg pp_data -drizzle -pixfrac=0.4 -kernel=gaussian

stack r_pp_data rej none -rejmaps -nonorm -out=result

load result
save 

close

[junior hound]

@stark ermine registration no longer works in branch #1661334033 , siril crashes after any type of registration . The issue is fixed with the previous version #1659588282

[stark ermine]

I don't know what you mean by 1661334033

pipeline number?

the last is 1659588282

And you are in a branch that is not in master

[junior hound]

Yes

I'm just sharing this in case it might be helpful

[stark ermine]

branch that are not in master are branch that can be totally unstable

[junior hound]

Also a feature that I think could be important to implement later is rgb alignement during registration. Most of the time for deep sky the atmospheric dispersion needs sub pixel alignement so if it was done during the reg-drizzling stage it could be a lot better

[stark ermine]

The biggest problem with the atmospheric dispersion is that channels don't have the same focus point in general. Aligning layers together won't help

[junior hound]

I'm not sure to follow you. At least for reflectors and excellent refractors, the only source of "chromatic aberration" is a simple shift (assuming a narrow enough fov) in each channel which is easily correccted by alignement (like in my example)

[steady mountain]

This is kinda more of a pet peeve but

Being able to rescale a selection based from the center point;
Basically, having a keybind to allow equal scaling from all sides that will keep the center point the same on all occasions

[junior hound]

Any idea why those patterns appear on the image with 0.4 pixel scale gaussian drizzling? The other image is identical except I used 1.0 pixel scale and the patterns are gone

[stark ermine]

Probably artifacts due to sub-optimal choice of pixel fraction value / scaling in relation to the images, and also their number.

[junior hound]

WIth 1500 subs and scale of 1 I won't except that much artefacts

And with the point kernel the results are perfect too (although noisier)

[junior hound]

Btw it does the same thing on other sets of data, even with pixel fraction really close to the scaling value (like 0.8/1.0) @stark ermine

Not a normalisation or rejection issue either

[stark ermine]

ok

[stark ermine]

I don't know. Probably as I said, probably normal artifact. If it works better with a another kernel shape, take this one (for example the point)

[dim minnow]

ive got this problem

[merry perch]

Try switching to the FITSEQ format

[dim minnow]

how?

im currently searching for it

[junior hound]

When you import just change from fits images to fits sequence

It will avoid to create 3500 bloody files on your computer lol

[dim minnow]

so i have to do it manually then

?

[junior hound]

Well you have to change the setting manually yes

Oh do you mean you can't use scripts?

[dim minnow]

yeaa

i cant use scripts lmao

[stark ermine]

Yes you can... And the answer is in the screeshot you sent

[dim minnow]

How to add?

[stark ermine]

as said in the log. Just a parameter in the script

with notepad++, whatever

[floral plaza]

How is the final field rotation determined in Siril? Like, if you have frames with various field rotations, which rotation/cropping will be decided for the final stacked result? Does siril just take the crop of the first frame inputted?

[broken peak]

you can pick the reference frame

with the script I think it chooses the sub with the lower fwhm

[floral plaza]

Do you know maybe any way of changing that ? Perhaps within the script file itself ?

[broken peak]

no sorry, I process manually so I haven't checked if that's possible

[floral plaza]

I see

alright thanks!

i usually just let siril run overnight to stack

[broken peak]

damn overnight?

[floral plaza]

well it takes abt 4-5 hours to stack so why not do it when im asleep right?

[broken peak]

it makes sense yeah, I'm just used to stacking taking less than 5 minutes on my pc

[floral plaza]

goddamn that must be a powerful PC then lol

unless u take longer subs and therefore have less images to stack ? im limited to max 30-40 sec subs

[broken peak]

nah it was an old-ish r5 2600x

[broken peak]

even with 60s subs it doesn't reach 10 minutes, we'll see when I stack my 12h of orion at 60s

but yeah in general I take 180s subs so it's more manageable

[floral plaza]

whatttt 😭 what am i doing wrong lmaoo

i usually stack on my laptop but even on my PC it takes a similar amount of time

[broken peak]

do you know what cpu you pc and laptop have?

[floral plaza]

laptop has an Intel i5-9300HF
PC has a AMD Ryzen 9 5950X

i have no idea if these are good or not

[trim steppe]

5950x is pretty baller iirc

are you stacking on hdd or ssd?

[floral plaza]

ohh that might be it!

i have an external portable drive where i store all my raws and astro stuff

occasionally it would make a weird noise like a paper clip dropping

[trim steppe]

[floral plaza]

are HDDs really that significantly slower?

[trim steppe]

yeah, quite

[floral plaza]

damn

[trim steppe]

try on ssd and see what happens

[floral plaza]

might do that

[broken peak]

dude...

both laptop and pc are better than my old 2600x, just use an ssd or nvme

[floral plaza]

:((

[trim steppe]

time to get another ssd

[trim steppe]

im even below you, 1600x

[broken peak]

I just upgraded to a i7 12700 so we'll see how well this does

[split smelt]

I'm still using an old Haswell Xeon. It's still quick enough with 10 cores

in fact, I've just ordered a new cpu from ebay, 14 core for £20 🙂

[broken peak]

Nice

[mental geyser]

seems like deepsnr CLI is in the works. will it be possible to add on siril?

[stark ermine]

no

the release is near. ANd we do not add any third party program

Now it will be by script

[hollow loom]

How can i install siril version 1.3+ so i can enable the dev version when stackin in sirilic?

for drizzeling

[sage glade]

@hollow loom

[hollow loom]

Yes

[sage glade]

@hollow loom

[stark ermine]

don't use installer

we should warn user

Don't use installer...

because you will replace the stable version

and the weekly build is generally to old

[hollow loom]

Then this one ?

[stark ermine]

these days we do a lot of fix, everyday

[stark ermine]

better yes

[hollow loom]

I want to use it with sirilic for stacking with 2x drizzle

[stark ermine]

I don't think sirilic is ready for this

sirilic is developped by another dev. And I'm not sure it follows the Siril dev

[hollow loom]

How can i do drizzle in sirilic then

[stark ermine]

you can in Siril

You can use siril for everything

[hollow loom]

Yeah but i need to stack multiple nights with different calibration frames

[stark ermine]

As I say. This picture is done on multiple night. More than 10 ... I only use Siril: https://telescopius.com/pictures/view/205383/deep_sky/ngc-206/m-31-with-the-oiii-arc/by-lock042?gallery_user_id=16193&gallery_order=is_featured&gallery_order_asc=false&gallery_page=1

[hollow loom]

WOW

[stark ermine]

Well, In manual you can do what you want. Sirilic is just a front end. A nice one, but just a front end

and as siril 1.4 is not yet released, I don't think Sirilic can handle all the new functions / commands

[hollow loom]

Can i install 1.3 somehwere

so is there a stable version

[stark ermine]

if you download the archive you showed me, this is a standalone application

this is why I do not recommend the installer

[hollow loom]

im confused

[stark ermine]

why

[hollow loom]

idk

Ok now i installed version 1.3.6 and lets see how sirilic is gona stack

[stark ermine]

if it uses drizzle... it will probably be the same as in 1.2. So not the proper drizzle

[hollow loom]

Oh :/

How can i do proper drizzle with sirilic

[stark ermine]

as I said

you can't

[hollow loom]

ok

[stark ermine]

Sirilic is not updated

https://gitlab.com/free-astro/sirilic/-/blob/master/sirilic/lib/changelog.py?ref_type=heads I see that the last update is sep. 2024

which is very old

[hollow loom]

I am now learning how to stack manually in siril

I am currently making the sequences

[hollow loom]

now calibrating

(i know im slow but its my irst time stacking manually)

[hollow loom]

@stark ermine Sorry to bother you but why canti use the use drizzle option in the registration tab?

Btw. its version 1.3.6

[stark ermine]

If color image, probably because you've already debayered them

[hollow loom]

Mhh... so when calibrating can i not debayer but then dabayer later?

[broken peak]

you have to do everything with bayered subs

including the calibration frames

[hollow loom]

Mhhh.. i think it works now

and the file size confirms it

[stark ermine]

everything is explaiend here: https://siril.readthedocs.io/en/latest/preprocessing/drizzle.html

[merry perch]

@stark ermine I hope its an appropriate question for this thread, but as you can expect, registering a lot of drizzled frames eats up a ton of storage. I have around 500gb free on my drive, which means that to 2x drizzle 20 hours of frames I have to split the stack into 2 batches, and then stack the remaining masters from each. However my question is if by doing this I end up losing signal in my final stack as opposed to being able to register and stack all the frames at once, and if you know some strategies or settings to use for each batch in order to minimize data loss if there's any at all

For reference I use 10s exposures here's why they are so many

[stark ermine]

spliting the whol set in several set is not a good idea. It is not optimal for the normalisation and rejection part in the stacking

[merry perch]

But I suppose its the only way to do it if I dont have enough storage?

The other option is to not use drizzle but that defeats my objective

[stark ermine]

Is it mono or color camera?

[merry perch]

Its an OSC

The S50

[stark ermine]

well. You should not have bigger files with Drizzle. Because you just need a CFA Drizzle, with a scaling factor of 1.0 and a droplot of 1.0

[merry perch]

I used 2x drizzle though , I've seen it works well with Seestar data

The extra resolution helps

[stark ermine]

Do you use which version of Siril ?

2x drizzle on OSC data is useless

Just lost of space disk

Even worst if you use Siril 1.2.x

[merry perch]

The dev version (1.3.5)

[stark ermine]

well, it is not needed to do 2x for OSC (and 1.3.5 is too old for dev version)

[merry perch]

I still have it because even when git pulling and updating I dont get 1.3.6 for some reason

im on the master branch

[stark ermine]

You don't get the right version then

[merry perch]

What should I do?

[stark ermine]

You need to find the latest commit on master, which is only a few hours old.

[hollow loom]

I have rgb data (dslr) and i want to drizzle but i already debayerd when calibrating the lights with the masterflat and masterdark. For drizzle should i not debayer when calibrating. But then how do i debayer after drizzle?

[raven lodge]

You don’t debayer, it becomes debayered during the drizzle process

[hollow loom]

Oh

I understand it now

[hollow loom]

for 2x drizzle is this the right setttings (i have 621 frames (10h) of un debayerd)

[trim steppe]

scaling should be 2 instead of fraction iirc

[hollow loom]

Oh ok but i heard then it does interpolation

Oh drizzle is still enabled

Are these good settings then?

[raven lodge]

I’ve been using 2x scaling and a pix fraction of around 0.7 to 0.9, Gaussian model, with my Seestar data. This seems to work better on some some sessions than others, I think it depends how long and also much field rotation there was.

Seestar is slightly under-sampled but extremely well drizzled.

[merry perch]

How does the gaussian model work better?

I've always been using the square model but I dont know the difference with the others

[raven lodge]

I find it helps improve contrast and smooths out some of the colour noise, but you need a lot of frames to notice the difference. The drop shrink parameter is the most important by far.

[merry perch]

with 2x drizzle I use 0.5 - the description says the reciprocal

and thats what I always did

[raven lodge]

The thing about OSC drizzle is that you’re also compensating for gaps in the data, because only 1/4 pixels are R/B and there are G pixels between them. So you are trying to compensate for those gaps. Making the drop shrink a bit larger can help fill those gaps a little, reducing some of the noise (though blurring the data if you go too far).

[merry perch]

Ah i see thats interesting. So you'd say 0.7 works for you most of the time?

[stark ermine]

never

[stark ermine]

For OSC don't do 2x drizzle, this is useless

[junior hound]

It can also mess up some denoising AI's work (deepsnr and graxpert mainly) so you have to be a tad careful

[stark ermine]

Hummm really not sure about that

we explain a lot here:
https://siril.readthedocs.io/en/latest/preprocessing/drizzle.html

[junior hound]

Definitevly, it make those weird false stars

[stark ermine]

Au contraire, Drizzle makes noise beter (in a gaussian profile) while debayer can change that

Noise with gaussian profile is generally easier to remove

[junior hound]

Ooopsie misunderstanding

Drizzle is a lot better yes but it's just using a big droplet (like 1.0 gaussian droplet at scale 1) that messes up the AIs

[broken peak]

So like 0.8 droplet like deepsnr asks for?

[junior hound]

Idk for deepsnr, I don't have pix

For Graxpert it seems that 0.5 gaussian is the max (but it introduces stacking artefacts so I just use the point model most of the time)

[raven lodge]

I’ve seen people say that but I don’t see why it should be true. If you have enough frames it should help you reach the theoretical seeing capabilities of your equipment.

[stark ermine]

Because you have RGGB on a grid of 4 pixels

so Drizzle x2 won't help

We explain on our doc page that CFA drizzle is not expected to enhance resolution. At least to get a better background

[raven lodge]

Yes, but I don’t think that’s an actual limitation of the drizzle algorithm. It certainly seems to enhance resolution with enough data, better than 1x drizzle upscaled 2x.

[stark ermine]

Not really in fact, because debayer algorithm can restore resolution

[junior hound]

yep but not perfectly, those bayer algorithms don't work perfectly like drizle does and for noisy data they can introduce a lot of artefacts I have found

[raven lodge]

Ok, but that says that there are gains to be had, which certainly matches what I see.

To be clear, I’m not expecting a 2x increase in resolution, just trying to get the most out of the data. In my case when working with the Seestar’s low resolution, a 2x drizzle followed by deconv and noise reduction produces better results than simply working at 1x scale, and is sharper than 2x upscaling.

[stark ermine]

Is it something you could quantify, or just a feeling?

[junior hound]

Theoritically there shouldn't be anything retaining you from an increase in resolution with 2x osc drizzle, but the limited signal could make it hard to actually see a benefit

[merry perch]

Yes I can confirm. It really makes a difference

[raven lodge]

Yeah, I’ve done a side by side comparison in the past. Having trouble finding it now though, I’ll have to try it again when I have more time.

[junior hound]

Hey I'm curious, where did you find all the reference spectras for SPCC?

Nevermind it's cited in the jsons

[split smelt]

yeh, available as csv in the GAIA archive. We used it to create a gaia database that you can download for offline SPCC. This was a massive undertaking as the available GAIA data is HUGE and it took ages to process & distil into something relatively small but still useful https://cdn.gea.esac.esa.int/Gaia/gdr3/_catalogue_sizes.txt

[split smelt]

Surely a 2x bayer drizzle for OSC will result in quite reduced SNR unless you have a massive amount of data?

[junior hound]

Yeah 1.3x osc was already quite noisy for my data i guess you just need a LOT of data for 2x

[split smelt]

yeh, because there'll be quite a lot of bayer matrix "holes" to fill. The data gets spread thinner

[raven lodge]

Yeah, but when you regularly have 1500+ frames…

The SNR of each of those frames isn’t great, being only 10s.

But overall…

[hollow loom]

well

not good?

I think that the Canon eos 2000d is rggb but in the fit header was grbg or gbrg

[split smelt]

It depends which artistic style you're going for? 😄

[hollow loom]

well i dont think thats good, it drizzled it wrong right?

Its a single frame

[split smelt]

did you try to override the bayer settings in the Siril preferences?

[hollow loom]

I am doing tests now

i think 2x bayer drizzle is too much

with scaling 1,50 this is red

green

blue

and green is good

rggb

idk

[stark ermine]

This is what I'm saying above

[hollow loom]

But now i didnt debayer in calibration

[hollow loom]

is this good?

red

green

blue

rgb

this was at scaling 1 and pixel fraction 1

[raven lodge]

That’s not necessarily bad for a single frame, as stacking a lot of data should fill the holes, but artifacts can creep in. That’s why you should increase the drop size, and why you should use Gaussian instead of square.

[junior hound]

As he said, totally normal for a single frame

[split smelt]

why would he use that for a single frame?

the whole point of it is that it uses multiple dithered subs to achieve its effect

[hollow loom]

I dont i just tested it so i dont have to wait hours for the registration process, but currently im am registrating the 10h

[split smelt]

it shouldn't take hours if you're running on an SSD or NVME

[merry perch]

So what are the conclusions on this? Do you guys think 2x is too much for the Seestar?

Honestly it always seemed to work out decently

For me at least

this is 2x drizzle, using square droplet model, pixel fraction 0.5 - 14 hours of 10s frames. There is a lot of field rotation artifacting on the bottom part of the image but it doesnt matter as Im only interested in the galaxy

[split smelt]

do a 1x and a 2x and see what your SNR is like on both

[junior hound]

the snr looks a bit low but the stars are small, perhaps something intermediate like 1.5?

[merry perch]

Could try to see. Anyway im aiming for 20 hours, possibly 25

By the way whats the best empirical way to test SNR on Siril? I tried the noise estimation option but I dont know how to read the values, i dont know whats the baseline for a good SNR

[merry perch]

@stark ermine specifically the reason why i can't get the latest version is because this happens whenever running the meson setup and ninja commands

[split smelt]

https://siril.readthedocs.io/en/latest/installation/windows.html <-- mentions gtksourceview4 in the dependency installation

[merry perch]

ok seems like I might have to rerun the command

Ok its solved, now like the guide said I went on to run siril.exe from the windows enviroment

but it fails cause it cant find some dlls

[stark ermine]

Well. The documentation explains how to install all dependances because one is missing here.

(I'm on my phone)

[merry perch]

I went and reinstalled them all

now siril builds correctly

the only issue now is running the exe like the guide fails

probably best if I uninstall everything and reinstall

[merry perch]

Ok, reinstalled everything and it works

[old timber]

I just downloaded the master file for siril beta, how do I install it?

[dim minnow]

can i compress the size with siril?

i mean the temp files

[stark ermine]

https://siril.readthedocs.io/en/stable/file-formats/FITS.html#compression

[dim minnow]

how to do the rice compression, sorry im not a programmer

[stark ermine]

https://siril.readthedocs.io/en/stable/preferences/preferences_gui.html#fits-options

Here

[dim minnow]

thank youu

[broken peak]

did the way to drizzle get changes? I'm using one of the betas and the one two pass isn't there anymore

and global star alignment does have drizzle with the scaling and pixel fraction options

[junior hound]

have you seen the 2 pass tick box in global registration tab or is it another issue?

[broken peak]

I probably missed that

[ashen bough]

Wheres the download link to the dev version?

[stark ermine]

link changes every day, several times a day

go to the gitlab repository