What's missing from Siril?

[ancient talon]

Download the latest one from here

https://gitlab.com/free-astro/siril/-/pipelines?page=1&scope=all&ref=master&status=success

[subtle night]

Have you ever considered adding a Savitzky–Golay filter?

[stark ermine]

1.3.6 has an edge preserving filter

[split smelt]

SciPy has a Savgol filter. I'll create a Python filter later to do this

[split smelt]

https://gitlab.com/IanCass/siril-scripts/-/blob/savgol/processing/savgol.py?ref_type=heads <-- savgol filter as a python plugin. I spent about 30mins on this. Shows how easy Siril is to extend now that we have this Python interface

It does work in so far as it does something. However, I think this MMT filter I made is better https://gitlab.com/IanCass/siril-scripts/-/blob/mmt/processing/multiscale-median-transform.py

and, the built-in Edge Preserving filter is also better

[ashen bough]

Why does my siril keep crashing when I try to drizzle?

[stark ermine]

Why don't you give more information?

[merry perch]

are you debayering by any chance?

[ashen bough]

No I didn’t debayre anything. And I did 2 pass registration then apply existing registration

Fenice helped me out with all the settings

[trim steppe]

could it be running out of ram?

[ashen bough]

I have 32gb ram

[trim steppe]

so do i and it still topped it out, check on task manager while running it

[ashen bough]

Hmm ok I’ll have a look at that the next time

[sage glade]

~((~starless)*(~starmask))

Putting this equation in here for future reference

[stark ermine]

??

[mental geyser]

You're meant to ignore it.

[sage glade]

https://tenor.com/view/naked-gun-please-disperse-nothing-to-see-gif-25791374

[mental geyser]

is there a way to get around this

im cropping after running graxpert bge

[stark ermine]

Well close the GraXpert dialogue and crop the image

[mental geyser]

The dialogue isn't open.

[stark ermine]

@mental geyser : which version do you use?

because I have no issue

[mental geyser]

1.3.5

[stark ermine]

...

🙂

[stark ermine]

So yes. Updating

When using dev version, you need to update everytime

[mental geyser]

Am aware

I've just not updated it bc I want graxpert Decon in siril

[stark ermine]

Well. You should update, because your version contain many bugs and you will be stuck in an old version

We removed deconv for good reasons

[mental geyser]

🥺

[trim steppe]

would be nice to have an optional toggle in preferences or sth to still be able to use it

[stark ermine]

Nope. This tool will be stuck in a beta state

It will never be released

Ricardo, working on it, left the GraXpert team

[mental geyser]

Oh

How come?

[stark ermine]

They disagreed on whether to release the stable version as is or not.

[merry perch]

ok so basically we are never going to get it

[mental geyser]

wonder if theyre gonna make a star removal program

very hard to design but curious to see how it performs

[split smelt]

what like starnet++ that's already in Siril?

[mental geyser]

Referring to graxpert devs

[merry perch]

It would be cool as starnet is the only free star removal program really and it has some issues

[split smelt]

That's not entirely true. There's also "nox". However, I don't think it's quiet as good as Starnet++. I did try to train new star removal model using it without much luck in improving the quality

[merry perch]

I mean, Starnet is quite good for being free software

As I think in most cases StarX isnt really worth over it

[sage glade]

Hey uhm Lock042 do you know any other equations that can be used in the pixel math window? Such as :~((~starless)*(~starmask)) is better than starless+starmask due to the equation takes into account the pixels that they're on top of each other so it doens't clip the whites

[merry perch]

Pixel fraction is the same exact as Drop shrink on Pix right?

[safe shale]

Masks, range selection, etc

curves?

[broken peak]

in the beta

[split smelt]

range selection?

[broken peak]

it's for masking, it's both on ps and pix

[split smelt]

ok. I do that in Gimp

[worn pendant]

is it possible to cfa drizzle in siril just like in pix so I can use deepsnr?

pix won't stack my nb data but sirill does do it

[split smelt]

yes, with latest dev version, or wait a short time for the new beta

[worn pendant]

alr I'll look at it when I get home

but idk how to install the dev version

I tried a bit and failed lol

[mental geyser]

Yes but in my experiance the siril stack makes everything green and SPCC doesn't correct it

Deepsnr works on it tho fine

[worn pendant]

hm

thats gonna be difficuilt though

[mental geyser]

Concordavirus made a script. I'm about to sleep now so my laptop is off so you can ask him

[split smelt]

Really?

[mental geyser]

That was his script

I've not done manually stacking bc I'm lazy

[stark ermine]

SPCC correct it

[mental geyser]

Perhaps I should retry

[trim steppe]

i remember it having a rgb equalize thingie near the end of it so shouldnt really be coming out too green

[junior hound]

Hey I'm trying to stack my first mono images but the files are detected as rggb when clearly, they're mono

and I don't see any way to correct that

[mental geyser]

if youre using scripts, make sure youre using the mono script instead of OSC

[stark ermine]

need more information for sure

[junior hound]

Sorry
I'm using manual pre processing, importing my ser raw files as ser sequence and it's detected as rggb

There's not a lot more to say I think

It might be coming from my capture software in which case it's not a bug but I think it's import for siril to have an option to manually change the Bayer pattern (and change from osc to mono) after a sequence has been loaded, I've had so much frustration with wrongly detected Bayer patterns which forced me to reprocess the entire sequence to find the correct one, and now this issue

I've tried using a fits sequence to change the Bayer pattern in the header but I can't get it to work

[stark ermine]

If your SER is detected as color sequence this is because it contains color information in header. Easily visible in the logs by loading SER file in Siril

Just don't debayer it

[junior hound]

I don't have any control over that when drizzling, that's what I made the suggestion to be able to manually change it

[stark ermine]

So far we cannot force the file to be mono

However, this is just an integer to update 🙂

you can do it with hexadecimal software

https://free-astro.org/images/5/51/SER_Doc_V3b.pdf

Just change the colorID field to 0

but save your file before 😉

Only 4 Byte to change

[junior hound]

Nice, thanks for the fix!

[mental geyser]

whats this issue

ok it just works now

legit just closed and reopened siril

[stark ermine]

This error cannot be solved by restarting Siril

It says that you try to stack 16bits and 32bits images together

[mental geyser]

oh right

ok it turns out i had XISF masterflats and biases in the folder

[stark ermine]

ok. So yes, it was the issue

[merry perch]

I was reading the documentation to understand if I can automatically uncheck frames based on conditions (eg: FWHM < 3&& Stars > 30, like in Pix's subframe selector), and based on what I understood I can do something like -filter-fwhm=3 -filter-nbstars=30?

Would this be correct?

And that unchecks it from the sequence instead of going through it manually?

[hollow loom]

Im thinking about getting a dualnarrowband-filter and i dont know how to manually stack the data (in Siril). So like a night without filter (or even more) = one stacked picture (ok thats allright i know how to stack that). But like a few nights with filter = two stacked pictures (Ha, Oiii) (idk how to stack that). So i have no idea how it works, but i think that im just gonna calibrate and register (with drizzle) the data with filter, just like the data without filter. Then seperate the r_pp_alllights sequence, so that i have a sequence with Ha and a sequence with Oiii. (no idea if that is how it works). It would be really nice if somebody would explain how to stack dualnarrowband data manually in Siril so that i have a stacked Ha picture and a stacked Oiii picture. Thanks😄

[broken peak]

there's the extract Ha and OIII script, you can also run that command manually tho

siril manual stacking is fairly simple
https://siril.org/tutorials/tuto-manual/

[broken peak]

I think you have to do everything up to registration without debayer, then run the command

[hollow loom]

I already am stacking one shot color data manually in Siril. So i just calibrate with the correct calibration frames as usual (no debayering) and then i run the command on the sequence (pp_alllights)? And does this give me 2 sequences with Ha and Oiii (so seperated), so that i can register and stack those 2 sequences to get a stacked picture with Ha and one with Oii?

[broken peak]

the extraction command will extract and strack Ha and OIII if I'm remembering it correctly

if you open the txt file with the script it will tell you everything it does

[hollow loom]

But i dont want to use a script i want to stack manually

[hollow loom]

So i need to register before running the command?

[broken peak]

I know, but if you open the script file it tells you which commands to use and what they do, that's what I meant

[broken peak]

seems like it extracts before registering

[hollow loom]

Im stupid so i dont really understand what i have to do now

so i just run the command "seqextract_HaOIII pp_light" on the sequence with the calibrated lights and then i register the sequence with ha and oiii and stack these two sequences?

oh i tested if the command runs and it does (on data witout filter): NICE. So now im gonna have to get the correct dualnarrowband filter so i can use it 🙂

[dark sleet]

Can I also extract Ha and OIII as drizzled fit?

[merry perch]

Is there a way, after registration, to like type a command and unselect frames from the sequence automatically based on specific parameters (like FWHM < 5 in Pix's subframe selector)?

[stark ermine]

Use the frame selector, sort by FWHM, unselect by batch

[merry perch]

Oh thanks, so like select them and then unselecting one unselects them all?

[stark ermine]

Like you would select files on your computer

https://siril.readthedocs.io/en/stable/Sequences.html#frame-selector

[merry perch]

It was simpler than I though. Because I read the documentation in search for a solution and found this part but now I understand that its for scripts?

[stark ermine]

no. This is the filter during the stacking. But it does not unselect frames. It does not take it, but after the stacking they are not unslected

[merry perch]

So this is used for scripts or I can tell it to stack in the CLI and add these as arguments?

[stark ermine]

the command lines are for the script but you can use GUI too

[merry perch]

the first time I restacked this data it aligned everything just fine but now it rejects too many frames, star registration parameters are default in both cases

oops wait I probably found the cause

There was a weird image being used as reference for registration

[mental geyser]

what a weird message

[stark ermine]

Why weird?

[mental geyser]

Fixed it

[sage glade]

Bruh why it dissapeared?

[raven lodge]

Does anyone know how you're supposed to set up colour profiles on Windows? I'm thinking specifically about a Siril workflow, but I know nothing about colour profiles in general and I have problems where I think something looks fine and then I view it on an iPhone and it looks ass.

Also the same image can look quite different in Siril, Graxpert and Affinity on the same machine

[stark ermine]

first make sure the vizualisation sliders in Siril are on their min and max

[clear sparrow]

Hello 🙂
Does anyone know of new pre-processing scripts for Siril 1.4 already? Ideally incorporating the new drizzle and distortian correction

[mental geyser]

distortion correction i believe is a process and the drizzle script is in siril already when you download the beta. if not, just go to the siril gitlab page for scripts and you can get it from there

[stark ermine]

No. We do not download from this page anymore. This is now managed in the Siril preferences, script page

[clear sparrow]

This is a rather confusing situation... Guess I am going to stick with manual processing, while the scripts situation clears up (or work on my own scripts based on manual experience...)

[stark ermine]

while confusing?

You have a menu called "Get Scripts" in Siril

you click on it, and you have a choice of scripts

I think this is straight forward

[clear sparrow]

sure, but that is a rather limited selection, do you happen to know if Sirilic is being updated?

no complaining btw. Siril is AMAZING 😛

[stark ermine]

no idea about sirilic no.
The selection is limited, I'm not really sure about that. And script python choice will grow up

but of course, you can also write your own, download script from other places

you even has an editor inside Siril to run scripts

[mental geyser]

holy wow siril 1.4 opens so fast on my laptop

like 3 seconds and its open wooow

it was like 10 seconds before

[mental geyser]

wow so helpful

would need to scroll down quite a way

[wise creek]

09:51:06: FITS error: light_00042.fit
09:51:06: Could not load image 41 from sequence light_```

[stark ermine]

You are using old and not supported scripts

Now there is a new mechanism for using scripts

[mental geyser]

how do you fix this error

[stark ermine]

By installing local catalogues

Here the server are down

[mental geyser]

oh

when are the servers coming back online

[stark ermine]

It is not our servers, so I don't know

[junior sonnet]

@stark ermine 1.4 support xisf right? I can load images in with drag/drop, but I can't select xisf files for calibration files in the calibration tab

[mental geyser]

yes

oh

yea so you need to convert those to fits

xisf can only be read on siril iirc

[broken peak]

They're read only, you have to save them as fits to use them iirc

Nvm sword said it already

[junior sonnet]

ah ok

rip

[stark ermine]

This is normal

You need to convert in FITS

https://siril.readthedocs.io/en/stable/file-formats/XISF.html

[stark ermine]

We don't love this format that much 🙂

[junior sonnet]

is there a plan to love it a bit more maybe

[stark ermine]

oh no. FITS are the way to go

FITS are universal

FITS are used by everyone, even professional astronomers

[junior sonnet]

:(

[stark ermine]

just use FITS 🙂

[junior sonnet]

pixinsight doesnt so

i really cant

[stark ermine]

Pixinsight does. It just say it is deprecated, to force the use of XISF

but if all users complain about that...

[junior sonnet]

as if the devs will change it

if I could I would 100% change all of the temp files from wbpp to fits

[stark ermine]

They might. PixInsight is a commercial product. They have to satisfy their customer.

[junior sonnet]

their method of satisfying the customer is ignoring everything they say

[compact skiff]

improve denoising

currently i have to switch to graxpert to get any meaningful dennoising

siril's denoise feature may be insufficient

[junior sonnet]

isnt graxpert already built into siril

[compact skiff]

is it?

maybe i need to try that then

[broken peak]

not built in but there's the integration like starnet

[stark ermine]

yes it is

[mental geyser]

This is because it's for mono images, not OSC

Unless you're Bayer drizzling your images

[dark sleet]

I have a question. I have the developer version 1.4 installed and recently I drag and dropped ad xisf file into siril. With the interface port of Graxpert I tried to denoise it but I always got an error. I thought that maybe Graxpert is the problem but Graxpert never had issues with loading and process xisf files. When I converted the xisf file in graxpert to FIT, reopened it in Siril it worked just fine when I tried to process with Graxpert via Siril.

Is this a communication issue between Grax and Siril?

[stark ermine]

What is the error.
More information needed

[stark ermine]

Do you want to learn more about your setup? You can do it in one-click now 🙂

[subtle night]

AYO

[steady mountain]

What has changed with platesolving? It doesn’t work that well anymore in the newest version

[merry perch]

Did you try installing Gaia DR3 with the catalog installer?

[steady mountain]

Does it work for coma?

[steady mountain]

Mm no

[stark ermine]

It works better than before normally

but without more information..

[steady mountain]

Haven’t got it to work

[merry perch]

That should perform better then

[stark ermine]

Yes but more information is really needed. Like logs for example

99% of cases, if it does not work it is because the user did not fill the pixel size, focal or coordinates correctly.

[steady mountain]

I’ll send you a log once I’m home

[dark sleet]

where

Aberration inspector is a bit buggy at non debayered images

"show distortion" is greyed out :(

[dark sleet]

ah....platesolving..

[stark ermine]

I think this is Moiré issue

[safe shale]

When I do a stretch (any), the image becomes black and white... Have I done something wrong?

Autostretch view works fine

[stark ermine]

@safe shale : Share your image plz

[subtle night]

Have you tried using human or even weighted luminance?

[safe shale]

It happens with any image

[safe shale]

Yea, this is human weighted luminance. I've tried every other option, and the same happens with histogram stretch, etc.

[stark ermine]

I use it everytime, never seen something like that

[safe shale]

It was working fine with every other version

[stark ermine]

I use the last one

So this is very strange. First time I heard this

If I can't reproduce it will be difficult to fix it.

[stark ermine]

clearly the histogram is not monochrome, but the image is very desaturate. How do you obtain the image? OSC?

[safe shale]

Yes

I managed to process it without any issue using beta 1

[stark ermine]

I can bring color back easily however

[stark ermine]

beta2 has no change here

what is your workflow before the stretch? all the steps please

[safe shale]

SPCC (g2iv), bge (graxpert), starnet, cosmic clarity, denoising (graxpert)

[stark ermine]

bge must be done before SPCC

[safe shale]

Oh

It happens with the base stack though

[stark ermine]

I owuld say that It is written everywhere in the doc 🙂

how do you stack?

[safe shale]

I did it manually in siril

[stark ermine]

Sure, but what options do you use?

screenshot from the stacking tab?

[stark ermine]

you don't normalize your data??

you should normalize, especially when doing pixel rejection

[safe shale]

I was just testing the manual stacking.. But I get the same results with any image stacked with any other software

[stark ermine]

ok. So the root of your issue come from your data?

[safe shale]

No, it happens with images from other people too

[safe shale]

What did you do?

[stark ermine]

I played with saturation in the GHS tool

[safe shale]

I've restacked everything with normalization and did bge before spcc, but it did not change much

I used human weighted luminance though

[safe shale]

Ok I think it was working before because I was doing starnet prestretch; it looks like human weighted luminance on linear data washes out color

[merry perch]

any idea why I might be getting this error trying to open a python script on beta 2?

it worked in beta 1

Ok maybe I need to try updating to Python 3.10+

No it still doesnt work

[stark ermine]

no

Reset your python Venv with the dedicated button in the Siril preferences

[merry perch]

Thanks

[subtle night]

One thing I'd find super convenient is having options for the most common/popular SPCC references at the top of the menu. Perhaps sort it by recently used options? I'd like to not have to dig through the menu for G2V stars when I want to calibrate against sunlight

[merry perch]

When viewing sequences based on their weighted FWHM compared to normal FWHM is it normal to notice much greater values on the higher end?

[mellow vine]

@stark ermine hey I can't use the align button in Siril anymore for some reason, using the RGB compositing tool. I add my RGB files and hit align and Siril just instantly closes altogether. I have no idea how to fix this.

Pretty sure it's a bug

[stark ermine]

please, first share a screenshot before pressing Align

[stark ermine]

ok, could you share your files?

[mellow vine]

https://www.swisstransfer.com/d/53bbf2d9-2f65-45bd-a50f-b72d34d9567c

[subtle night]

I've actually noticed some problems with alignment when doing RGB compositing. I get swirly beach ball of death stars with one of my datasets and nothing I do seems to get them properly aligned.
Here's the data (stacked channels + composited) if you want to have a look. It's not very good quality (I had focus/tilt issues), but in case it's of interest to you @stark ermine.
https://1drv.ms/u/c/4f69734a5f0092dc/EQZFgnjukoNDh8fV00y3msQBwQY0_ynJmW0mizHPJe9s6Q?e=L02KeL

The main reason I'm asking is because someone mentioned getting better FWHMs when stacking in Pixinsight - obviously this is up to the settings/method used but I figured I'd ask at least

[mellow vine]

@stark ermine Hey for some reason I have all my light files loaded but my tabs still say B&W... I dont know why im having so many issues all the sudden. This just started today.

[mellow vine]

@stark ermine im having a glitch where siril is having a conflict with the fact I have an an image processing dialog box open while im trying to execute a script. the problem is this is the GUI for the script.

Im reverting back to an older version of Siril. Too many problems

[stark ermine]

This one is fixed in the dev version

[stark ermine]

will try later today, thx

[stark ermine]

what if you click on reload all?

[stark ermine]

Did you stack using distortion correction. If not, this is normal you have issues at the end.

[subtle night]

Yes I did

[stark ermine]

will have a look to your pictures later today too

[subtle night]

I can try to restack it to rule out any problems

Thanks for having a look!

[gusty spoke]

Where do I find the drizzle stacking script? It doesn’t show up in the script repository

[stark ermine]

probably crash is fixed for next beta I cannot reproduce

[stark ermine]

I'm sorry but that looks good to me

[subtle night]

Is this 1.4.0-beta2?

[stark ermine]

the dev version, but no changes in the way we align images

[subtle night]

Welp, I guess I just need to get better data then

[stark ermine]

Try to apply the Abberation remover script

Just applied, and it works well

[stark ermine]

you can do something really nice

It could be better because I haven't protected stars here

it was just for fun

[subtle night]

Are you using a histogram stretch or GHS?

[stark ermine]

GHS

always

far better

but here, you need to work on the starless image

I did not do that

this is why stars have artifacts

but as I said, it was just a test

[subtle night]

I wasn't sure if the color spread on the stars was an indication of poor registration or the data just being really difficult

long story short, I don't think I secured my imaging train properly and the meridian flip ruined everything

[subtle night]

You know anything about the repo used for the aberration remover? By choice I don't use any AI tools in my processing, but I was kind of surprised to see the repo listed in the script having nothing but a README.md and one commit

Oh, it's in the releases

I didn't know you could do that lol

[stark ermine]

🙂

It is really easy now

[subtle night]

one last note, that data (and the raw/calibrated lights if you need them) is CC0 so feel free to use it for whatever purpose, no attribution required

[stark ermine]

Too bad, just finished to write my book about Siril :).

[mellow vine]

Not sure why it's happening to me. Siril just closes every time I hit align on the beta 2 version. im back to beta 1 and no issues.

[stark ermine]

Try with this one please: https://gitlab.com/free-astro/siril/-/jobs/10253191624/artifacts/download

[mellow vine]

Okay I will later today when im done stacking

[stark ermine]

ok/ Let me know

[stark ermine]

did you try?

[mental geyser]

whats the difference between the new dev version and the beta 2 stable?

[stark ermine]

bug fixes

[merry perch]

Little question, if I happen not to find a script in the preferences, even if it is checked to have the online catalogue refreshed and clicking to manually update, what is the next logical thing to try? Always reset python venv?

[subtle night]

For denoising and other tools which allow you to alter the fraction by which the original image can be mixed with the result, it would be cool if we could preview the result rather than having to undo/redo at different values

[stark ermine]

If you are sure the script exist, there is a small bug in beta2 that will be fixed in beta3. But first check the logs at startup. Do you have some error loading scripts repo? Please show me.

[stark ermine]

Just use ROI

[merry perch]

I get this. Scripts repo works perfectly fine on my main windows machine. This is Ubuntu running on Termux, so on top of Android, so this isnt the enviroment where I'd expect things not to go wrong of course

[stark ermine]

this is the bug. You should remove the siril-scripts and siril-spcc-database

[merry perch]

The directiories right?

[stark ermine]

yep, all dorectories

they will be built again at startup

[merry perch]

Would you mind telling me the path of those directiories? I can find them on windows but on Ubuntu Im kind of lost

I cant seem to find them

[stark ermine]

Somewhere in usr/share I believe

[merry perch]

As of currently, I have the Gaia DR3 local Astrometry catalogue installed to use for plate solving, and with near search enabled, I often get 80-100 rejected images over 1k frames for example. Is there a way to improve this or are the frames just so bad they cant be solved?

[merry perch]

If I platesolve one individually it says not enough stars, because yeah their shape isnt very good, can I alter star detection tresholds to platesolve the sequence?

[worn pendant]

why do you want to plate solve each individual frame?

[raven lodge]

Mosaics

[zenith ridge]

Support for compressed fits !!! (.fits.gz). Right now at least in the file browser only FITs files are discovered.

[stark ermine]

https://siril.readthedocs.io/en/latest/Dynamic-PSF.html

[stark ermine]

that depends why they are rejected

[stark ermine]

Siril supports compressed fits for a long time

[merry perch]

well the ones that have been rejected seem to mostly be towards the higher FWHM

Can the dynamic PSF settings be used in sequence platesolving?

[stark ermine]

yes, this is the goal

but each time you restart Siril, it does not save the changes (this is on purpose)

[merry perch]

Ok

[merry perch]

So in order to make sure the sequence plate solver is using those star detection settings, do I keep the Dynamic PSF window opened in the background?

[stark ermine]

you can close the dialog

just don't close Siril

[merry perch]

Should I detect stars too or just set the values since its for the whole sequence?

[stark ermine]

you can test detection on an image. To check

Check the relax button at first

it should help

[zenith ridge]

really? I dont see any file named .fits.gz if I click Open (1.4.0-beta2)

[stark ermine]

I use compressed files all the time yes. Drag and drop the file in Siril ?

[zenith ridge]

ok, missed that. Probably they should be discoverable by the Open dialog...

[stark ermine]

They are normally

Will check tonight

[merry perch]

Ok, I tried but now the reason is that it can't align the image with the reference stars

Even by increasing near search to 30

does that have to do with the quality of the image itself?

[stark ermine]

Maybe. Change the reference image to see

[merry perch]

Changing the reference basically changes the center from where solving starts?

[stark ermine]

So I confirm, it works

[subtle night]

@stark ermine I have some stellar data that uses IR/R/G filters instead of RGB. If I can get spectral data for my IR pass filter, would I be able to use SPCC on that data?

[stark ermine]

I believe SPCC only contains visible and not IR data

[subtle night]

I figured I'd probably just have to calibrate that data manually

[merry perch]

Is it already a known issue for Siril (from dev up until beta 2) to crash whenever its finished applying registration to a sequence? When rebooting the sequence is correctly registered but if its needed I can provide logs

[stark ermine]

no. no crash here

looks like a memory issue

[zenith ridge]

nope, at least not in MAC, or at least not in the Conversion/Source tab where.

file name is M_20_Light_XXX.fits.gz ...

[stark ermine]

Oh.

[zenith ridge]

and it is not recognized. Which is silly because if you use cfitsio fits.gz should be supported

😛

[stark ermine]

Set your extension to fits

In the settings

[stark ermine]

But we use it. And as I said. This is supported

So I don't know why on your machine it is not

But it is on every machine I use, and I always use compressed files created by NINA

[zenith ridge]

still not

if I move the file to .fit or .fits (even if it is compressed) then it opens it

looks like you are capping the filename extension

not allowing .gz

[stark ermine]

Very strange

[zenith ridge]

so it is supported, as expected. But you expect the gzipped fits to be named .fits / .fit which is very non-standard

[stark ermine]

As you see here

We read the extension....

[zenith ridge]

.fz is not the extension I am using

can you check if .gz is allowed ?

[stark ermine]

Ooooh gz.... Ok I got it

What software create these files ?

[zenith ridge]

gzip ???

[stark ermine]

Sorry I read fz

No I mean what capture software

[zenith ridge]

none? . I capture in kstars and gzip it

[stark ermine]

Hum ok

[zenith ridge]

but this is SUPER standard

like extremelly standard practice in astrophysics

[stark ermine]

Well extension is normally fz for compressed fits files

[zenith ridge]

In my 8 years of postdoc in astrophics + 5 years of PhD student I have not seen a file named .fz 😛

.fits.gz or .fit.gz yes, .fz never

[stark ermine]

Well, I have a background in astrophysics too and sorry this is the standard for compressed fits files:)

[zenith ridge]

not sure if that is the same compression?

.fz is FZ compression algorithm, right ?

in any case, whether they are the same or not, cfitsio supports .fits.gz so probably we should allow it 😛

[stark ermine]

I'm on my phone, it's difficult to read cfitsio doc ^^

[zenith ridge]

no worries

https://heasarc.gsfc.nasa.gov/docs/software/fitsio/c/f_user/node64.html

[stark ermine]

Ok thx

Will work on it for beta3

[stark ermine]

I will work on a version and will give you for tests

[zenith ridge]

great !

one minor suggestion also.

It would be awesome if you could add some sort of white text on transparent background overlay on the image display (something you can turn on or off) displaying basic properties of the images: exposure, temp, etc from the header

also maybe on the 'conversion file-list'

I mean, I have a script to bulk rename the files and include those basic metadata as part of the filename, but still it would be great addon to have in siril

[stark ermine]

that could be a python script

anyway, this is not for 1.4 as we are in feature freeze

[stark ermine]

I sent you a PM

[subtle night]

@stark ermine I'm having SIRIL close/crash while attempting RGB composition on a globular cluster (M12). Any suggestions for how to get around this? If I convert the data to a sequence and register that way, there are no issues.

Let me know if you want the data

[stark ermine]

This is a known bug that is fixed in beta3

[subtle night]

Oh that just dropped, I didn't even check

[stark ermine]

beta3 is not released yet

[subtle night]

Just wanted to share that mosaic stitching with SIRIL has been an absolute blast
https://media.discordapp.net/attachments/1376159504961503304/1388624265011658792/m24.png?ex=6861a8a0&is=68605720&hm=88311ffc2fa4ba0ab7b214bc6509e8b3a53e7f241b9c04dbcfaee3dfce85f5dd&=&format=webp&quality=lossless&width=2692&height=984

[merry perch]

Working on a Seestar mosaic from M24 to M16 and can confirm this, basically mosaic of mosaics (using the mosaic mode, generating 1.7x mosaics, each one is a panel, and its going to be 9 of them in total.
Siril really makes it fast

[stark ermine]

Good to read 🙂

[merry perch]

@stark ermine is it normal for the weighted FWHM readings to be the same as FWHM when viewing sequences that were just platesolved instead of normally star aligned?

[stark ermine]

hum.. it could be the same when pictures are good. But not in every case

[merry perch]

And that would be weird cause my subs are almost never "good"

[junior hound]

Hey, quick request
I'd like to have the border feathering option available for all sequences (fits and ser, not just individual fits) . I know it's designed for mosaics, but I find it quite useful for normal stacking to reduce edge artefacts when the fov moves quite a lot around the target

[stark ermine]

to reduce edge artifact you can apply a background extraction of degree 1 to the whole sequence

[limber parrot]

why is it that when im clicking go register its not doing anything'

nvm

[subtle night]

Any tips for getting mosaics to register in busy starfields? After adding data to my M24 image I cannot get one of the panels to register in the IR channel

[subtle night]

Update: I needed to uncheck the box that skips images in a sequence that have an astrometric solution when I use the image plate solver

[stark ermine]

what the console is saying.

[junior hound]

Hm I'm not sure we're talking about the same artefacts because this is not nearly as effective as feathering

[dark sleet]

@stark ermine can I somehow apply the unlinked autostretch to my image? For like Sho the unlinked always looks just perfect. The linked is trash

[mental geyser]

its the chain icon next to the autostretch thing drop down menu

[dark sleet]

I know, but I want to apply it to my image

In Histrogram transformation you can apply the autostretch stretch to the linear image. But I want to have unlinked autostretch

[stark ermine]

No, but you can do better in fact

the autostretch should not be applied

You have color calibration tool that do a better job

The manual one is good for SHO

[stark ermine]

I've finished to write a book. It will be translated in English, German, Spanish and French.

In the book I will give a lot of example. I hope it will help you

[dark sleet]

Ill try

Oh damnn, you are right

[stark ermine]

🙂

[subtle night]

@stark ermine is there a way to easily extract the variance, standard deviation, or MAD associated with each pixel in a stack? I'm interested in what this looks like for my bias/dark stacks.

For straight averages, this is easy enough to do with pixel math, but for averages with rejection I'd need some way to exclude the rejected points

[stark ermine]

these are computed but not exposed

[subtle night]

No way to get to them even with commands?

(For reference, I'd like to see if I can use any of that information to better denoise my images; if needed I can just write that code myself)

[stark ermine]

You can only have global stats

[subtle night]

You'd probably know more about this than me (I'm entirely self taught with statistics) but I'd like to try seeing what happens when you weight the stacking of light frames by the variance of each pixel in a dark/bias frame. (Though I'd also guess I'm not the only person who's come up with this idea)

It could make a difference with dithered frames, assuming that the variance between each individual pixel isn't consistent

[merry perch]

I searched around the file system and I see that those directories are nowhere to be found. I even updated to beta 3 by now but still nothing. Perhaps I should create them?

Or maybe install the script manually

[stark ermine]

they are in %localAppData%

[merry perch]

On Ubuntu?

I did a catfish search all around and cannot see them

[stark ermine]

no on Ubuntu this in in share

[merry perch]

This is what I find

scripts only has the proprietary ones

Can I try creating them myself?

[stark ermine]

This is not the path I gave you

[mental geyser]

lol

[merry perch]

Oh I see I was wrong

Yeah it worked thanks

Im a potato using Linux its not my main machine lol

[stark ermine]

but you don't need to change something in these folders

So I don't understand why you want to go there

[merry perch]

I was mistaken between local and .local

[stark ermine]

yes, but why do you want to change something here?

[merry perch]

I just wanted to find the folders you told me to delete and fix the script repository issue

Or may I say directories

[stark ermine]

ok. But beta3 should be bug free now

[gusty spoke]

@stark ermine an issue I’ve been having is that the drizzle script is completely gone in my scripts, and the scripts menu. How can I fix his? On the beta 1.4

[stark ermine]

what are the script you have? You should have BayerDrizzle script in the built-in scripts

[subtle night]

@stark ermine how would you recommend working with bicolor data (like this hydrogen/oxygen) where the data channels are colored with non-pure colors?

This is HO(H+O) where I just combined the H and O images 1:1 for the blue channel.

Though I should add I stretched the blue channel independently

[stark ermine]

I don't understand your quetsion 🙂 What do you want from me 😄

[subtle night]

I guess, is there any way to work with a composited image with two data channels by working with the data channels and not the color channels?

In this case both hydrogen and oxygen contribute to the blue channel

I feel like it'll be easier to work with the data if I can stretch the hydrogen and oxygen data independently of color, then alter their mapping to particular colors

[stark ermine]

You can use GHS on each channel separately

[subtle night]

Right, what I mean is I'd like to be able to work with channels that correspond to the data rather than RGB color channels.

So in this image, hydrogen is mapped to #FF0080 and oxygen is mapped to #00FF80. That produces a 1:1 ratio of hydrogen and oxygen data in the blue channel, but that may not be a good proportion depending on what colors I'm aiming for or the relative amounts of data in each channel.

Instead I would like to be able to stretch the hydrogen and oxygen data independently of one another, and then remap their colors those channels correspond to in order to achieve the desired palette.

I guess this would not be a good way of working with data that's rigorously color calibrated (though it doesn't matter in this case), so there might be a better way to accomplish this.

[dim minnow]

i got this message and i want to install the script, how can i download it?

[stark ermine]

Looks like internet issue

[gusty spoke]

hey lock, forgot to respond, and im still having this issue. I dont have the built in drizzle script.

It used to be there, but then disappeared

[stark ermine]

Remove all paths in scripts tab of siril preferences. And apply.

[subtle night]

Hmmm...any way to interface Julia code with SIRIL?

[stark ermine]

If it's python yes

[subtle night]

Well it's in Julia, but there are ways to interfacing Julia with Python, so it might be possible

Though it would be a telephone game

[worn pendant]

so is this like a better way of star removal?

[subtle night]

Potentially. I need to test it out, or better yet, help others test it out to compare it to tools like Starnet++ and StarXterminator

[worn pendant]

alr, sounds interesting

[subtle night]

(I just don't do star removal in my own workflows)

[junior hound]

Afaik there's no way to align a sequence to an external reference (like aligning the color sequences to the stack result of the luminance). I think that would be a really useful feature since aligning after stacking can produce heavy artefacts in noisy regions and crop some of the corners out

[stark ermine]

This is something I want to implement for next version (1.6)

[junior hound]

Great ! for now sneaking the reference in the sequence frames works but yeah it's a bit hacky

[stark ermine]

Sure. I have to admit

[stark ermine]

As soon as final 1.4.0 is released I will work on it

[steady fern]

Not having to re-open some tools in order to apply them again would be nice (like Color Saturation and Asinh transformation). Toolbar of some sorts in an empty area with ability to pin most frequently used tools there

[junior hound]

1.4 will be such a big release, it's amazing all the work you have put into siril

[steady mountain]

Will the manual RGB composition accept XIFS next update?

this fella

[stark ermine]

No. I don't think so

But siril can convert xisf to FITS

[stark ermine]

XISF is really a format we don't want to play with.

[steady mountain]

Okeh I see I see

[subtle night]

Are there any glaring technical issues with XISF a developer should be aware of?

[stark ermine]

I wrote something about it here:
https://siril.readthedocs.io/fr/latest/file-formats/XISF.html

[subtle night]

Yikes...

[stark ermine]

So probably XISF is a nice thing with PI. But that's all.

Funny things. NINA is saving in XISF. But all the file keywords are stored in a FITS header that is written in the XISF header ... Not very efficient

[wise creek]

If the FOV moves a lot during the session, you can register using max framing and then feathering will be available at stacking. No plan to change it for other stuff than mosaics as it's quite an intensive process (and it also unlocks overlap norm which should really NOT be used for regular stacking)

[merry perch]

@stark ermine I tried moving a python script I made into the siril-scripts directory on Windows (the script runs fine from the script editor), but even after adding and rescanning the scripts folder it doesnt show up in "Scripts". Does this require me to reset the venv?

[stark ermine]

no. It should be ok. There must be another reason

[zenith ridge]

FITS is even used even to store response functions from telescopes, event lists (e.g. XMM Newton, Chandra, Fermi-LAT), spectra, 3d cubes, polarimetry (IXPE), catalogs... . We even tested storing in there raw data from cherenkov telescopes serialized with protobufs and compressed with custom algorithms. FITS allows everything ... I really don't see how one needs to reimplement the wheel with XISF for simple images like the ones we astrophotographers produce... . Unless of course you want to lock down your users into a closed ecosystem like what ZWO does for example. Having read-only support only is the correct call here.

[subtle night]

Yeah, I'm not about proliferating an effectively proprietary standard

[zenith ridge]

@stark ermine one question regarding the pre-processing/calibrating/stacking. As of now, this produces an incredible amount of intermediate data that is not really needed (for instance, when stacking flats you basically get all the files that are bias-subtracted, and then the stack per se). For lights you get the dark subtracted, the registered frames, etc.

For small systems, or even external flash drives (I am thinking on mini pcs running data storage and early reduction of the data out of an usb drive or sd card) this requires a lot of space, and lots of disk wearing.

Is there any possibility you may consider some strategies to simplify this? Maybe if there is enough RAM using ramdisks? Or look at how for example DSS or astap do it?

[stark ermine]

Well. We do not want to remove temporary files. At least from Siril (we have a FAQ entry for that). However, a new python script is being developed to take care of it.

[merry perch]

If instead of using astrometric alignment for mosaics I instead used regular global star alignment, using a larger frame (such as a pre stacked mosaic) to align the frames to it, then exclude it from the final stacking, would this be a viable alternative mosaic stacking method? This is mainly when Seestar frames go beyond 2048, I wonder if I can do this instead of having to split in batches since mosaics don't support FITSEQ, or if there are diminishing returns

[stark ermine]

Nope. Our mosaic alignment method only works with astrometry

[rigid breach]

@stark ermine im trying to run cosmic clarity sharpen script on siril but i keep getting this error

i cant find any reason online, also cosmic clarity denoise runs fine

[stark ermine]

Looks like an issue with wide char. Do you have some non utf char in the path?

[rigid breach]

i mean i havent even put in the path for the executable yet since you are supposed to do that on first run, but first run doesnt work

[rigid breach]

no

[hollow loom]

i always get this error when i try to run the continum substraction script

[hollow loom]

also tried the reset venv button but it dosent work

no difference

[hollow loom]

still

[stark ermine]

Need more details

[stark ermine]

Open a ticket on siril-scripts repo

I'm not the dev of this script

[prisma valley]

if siril implements a multi window interface i’ll switch right now

[stark ermine]

I use mono camera, and I use Siril with no issue

And generaly I do HOO-RGB

[prisma valley]

The way i process is probably completely different to yours then

Very very different

[stark ermine]

Surely yes. But I think we have to adapt to the tool and not the other way around. Right?

Maybe someday we will allow multiwindow, but we don't want something like PixInsight.

Something like GIMP maybe

[prisma valley]

Honestly after experiencing it, i could never go back. I just love it

[subtle night]

It's exactly the opposite in ergonomic design at least. Adapt the task to the human, not the human to the task. Obviously with software you're not directly considering ergonomics, and we can benefit from a radically different workflow (there's a reason why die-hard vim users exist), but I do think that there's a balance to be struck between the two.

I don't use Pixinsight but I would also find multi-window support game-changing for SIRIL.

[subtle night]

SIRIL somehow turned my colors crazy during deconvolution, but I'm not complaining.

(I was using a PSF I composited from individual PSFs for each channel in an SHO image and it broke)

[stark ermine]

WHich deconvolution?

[subtle night]

Richardson-Lucy

[steady mountain]

i scnred your lagoon

[subtle night]

You have committed a felony

Okay but fr the pink stars and grayscale background go hard

[subtle night]

I've uploaded the data, including the PSF I used for deconvolution, here: #1408507804117110786

[merry perch]

So basically with 2k+ subs mosaics I have no other choice but do it in batches and stack the masters, correct?

[stark ermine]

you can platesolve easily a sequence in Siril

this is how work the astrometric registration

[merry perch]

Yes but with more than 2048 subs i need to use the FITSEQ format in windows, that as far as I know doesnt support maximum framing.

So I've always been stacking batches and then stacking the masters

But I was wondering if there was a way to stack them all together in one batch

[stark ermine]

Oh yes if you have more than 2048 images .... We will work on it for next big release

for now ... Use macOS or Linux ? 😉

[merry perch]

Thats great to know

[merry perch]

Linux is tempting yes

[stark ermine]

Of course it is 😉

[split smelt]

if you're a windows user, you can install Siril under WSL2 and it will still show its GUI under Windows. I'm not sure if that fixes the 2048 file limit but I suspect it will

[sage glade]

~((~starless)*(~starmask)) update into more recent date

[stark ermine]

?

[sage glade]

Imma try and get more pixel math equations in due time

[trim steppe]

[stark ermine]

yep. Good recommendation in general 🙂

[trim steppe]

what "memory settings" do i change to fix this

[stark ermine]

in Siril preferences

[trim steppe]

changed max ram usage from ratio to fixed amount and now it instacrashes after pressing start stacking

[stark ermine]

you shoul decrease the ratio

[trim steppe]

.50 ratio and it instacrashes

ok whatever i try now it just crashes as soon as i press the button, gg

[stark ermine]

You still can have the logs

it could help

https://siril.readthedocs.io/en/stable/Issues.html#send-us-useful-information

[trim steppe]

this is all it wrote, it just ends there at the bottom

[stark ermine]

you should delete seq file

[trim steppe]

[stark ermine]

only *.seq files

[trim steppe]

how do i open the sequence if i delete the sequence file

[stark ermine]

Siril will create another one

I suspect the file is corrupted

you will need to register again

[trim steppe]

woohoo

[stark ermine]

??

[trim steppe]

well

when going with maximum framing it crashes

on "current" its fine

log stops on the same computing weights based on noise part

[stark ermine]

did you removed the seq files ? which version of Siril?

[trim steppe]

removed seq file for the registered subs yes, version 1.4.0 beta3

[subtle night]

Is there a way to orient images when registering/stacking so that north is exactly up?

[stark ermine]

You can do it at the end

Dealing with the result only is easier

[subtle night]

I have a dataset on IC 10 which causes SIRIL to crash whenever I attempt SPCC on it. Log file attached, and here is the stack I'm using.

I'm going to try a different datset and see if that has issues

Yeah it also fails on another dataset

The Gaia catalogues appear to be online

[stark ermine]

The server is down. Causing SPCC to crash

you can use local database

[compact turtle]

Narrowband normalisation

Saturation curves

Masking?

[stark ermine]

You can do GHS saturation

[compact turtle]

Really?

I didn't know that

What about hue curves

[mental geyser]

coming in V1.6

[stark ermine]

Very often people request features already existing:)

[merry perch]

So is it true masks are possibly in the works for 1.6?

[stark ermine]

Yes

But 1.6 is not even started.

As 1.4 is not rereleased yet

[subtle night]

I'll try the local database, but the servers are reported as up in SIRIL, their status page says they're online, and I'm still getting this issue

[stark ermine]

With local database everything works

No crash anymore

[merry perch]

Ok thats good. Another question that popped up in my mind is if I want to add rgb stars to my dual band Seestar mosaic, on how to do it here in Siril. Now my intuition is that knowing how seestar mosaics work, im going to do a larger mosaic than the starting one to make sure I have wiggle room and all the final area is covered, though my issue is that of course at the end the two masters, dual band and rgb wont match, so I guess they need to be star aligned, but also be the same exact size else star recombination wont work. Is there a way for me to turn a bigger mosaic into a smaller size image, like my final crop of the dual band mosaic, through star alignment, but without actually stacking the rgb stack and the dual band one together?

[stark ermine]

make sure to align RGB and dualband image together

it will be easy to compose after that

[merry perch]

So its what I said? I was thinking to take the final crop of the db image and then the larger rgb mosaic and use astrometric alignment, and apply registration but only in "current" with the reference frame being the dual band image, would that work?

Else what do you mean by together?

Because the dual band image is already processed and I'd like to just add to that one

[stark ermine]

I mean: make sure both images are registred. So yes, what you said should work

[subtle night]

@stark ermine my thanks to the SIRIL team for putting out updates like this

[mental geyser]

very nice feature

[mental geyser]

wow

just stacking failed

very useful error message

[floral plaza]

"please check the log to fix your issue" 😩 😩

[merry perch]

It says nothing else if you check all the log?

I dont think I ever got a blank log from Siril

[mental geyser]

Nope.

That is the log

[stark ermine]

no you can find more logs:

https://siril.readthedocs.io/en/latest/Issues.html#send-us-useful-information

[subtle night]

Left is the linear preview of my image - right is the image exported by SIRIL. What's causing this discrepancy?

[floral plaza]

Try changing the bit depth in the export settings?

[safe shale]

It might be because of the min and max sliders at the bottom

[subtle night]

forgot to respond but you were right, I was very stupid

[stark ermine]

For users who think Siril cannot be used for a monochrome camera. A script is currently being developed to allow multiple monochrome images to be used easily at the same time. Here is an example of R G B Ha OIII images opened at the same time and easily usable in Siril.

[broken peak]

I was wondering abut mono processing with siril, I haven't found a proper tutorial yet

[prisma valley]

this is more like it!

good stuff

[stark ermine]

Yes I think some of you will like it

[subtle night]

Uhhhhhh what is NL-Bayes doing

[subtle night]

@stark ermine I can give you the FITS for this one if you'd like to have a look. This is after deconvolution, but trimming to remove edge artifacts didn't change anything

Anscombe VST does work though

[stark ermine]

hmmm, never seen that before

[subtle night]

Also any idea why the Helix plate solves with massive distortions?

[stark ermine]

Because of optics

You can undistord during calibration

[subtle night]

It's not due to optics. The images register incorrectly as well when I attempt to align the color channels

This is the distortion map for my IC 10 data

[stark ermine]

Share your image please

[subtle night]

Calibrated individual frames are available here: https://1drv.ms/f/c/4f69734a5f0092dc/EtYv21zy0WVPn8Nz9RGlD3EBllOv3I1CYePhGkdCeRGiMw?e=gKvAgK

[subtle night]

Still trying to find/reproduce this image but I will include it too

I talked to some others about this issue and they have had problems with plate solving around the Helix in different software (apparently this occurs with the ASIAIR's plate solver). I wonder if it is due to something wrong in a star catalog

[stark ermine]

I've just downloaded one image (I'm in the train)

Looks good to me

Of course I redo platesolving with Siril

and to be sure the platesolve is good:

I did a conesearch with the gaia catalogue

[subtle night]

I also did, down to magnitude 15 with Gaia, and there's definitely something wrong

Both the solution and the conesearch were done down to magnitude 15

[stark ermine]

It is important to note that the distortion is greatly exaggerated. If I were you, I wouldn't worry about it.

if you want a better representation you could use a script

Distortion3D

a python script

the distortion could come from atmosphere refraction. I don't know where you leave bue this object is very low in general

[subtle night]

Madison, WI (43 degrees north). Here are some other images, and note that the last three were taken with an altitude of about 20 degrees, which is comparable to when I shot the Helix

IC 10, Crescent, Lagoon, Orion, M4

[stark ermine]

Note that the number of stars around Helix are weak too

The computation will be less good with few stars

[subtle night]

Yeah I think the move is to not use undistortion here. If I don't use an undistortion correction it registers without issue

[subtle night]

I'm finding that the cosmetic correction feature does not seem to do anything in 1.4.0-beta3 if I use a bad pixel .lst file. If I try to perform the correction during calibration, nothing happens, and if I try to perform it with seqcosme it just crashes SIRIL.

[stark ermine]

OK thx, I will have a look

about the first part, how do you know it dos not work? Did you compare with old version?

[subtle night]

I've been using 1.4.0-beta3 exclusively since it came out but I can install the stable version to compare.

I have been using master dark frames to calibrate my images, and that gets rid of the hot pixels with cosmetic correction checked. I decided to switch to using a bias frame with a hot pixel map (the darks have some weird banding and produced a noisier stack) and found that the results are identical after calibration with or without the cosmetic correction box checked.

[stark ermine]

Crash was fixed, however, the bad pixels maps works here

[subtle night]

In a new build I assume? I can give it a try to see if that also fixed the issue when calibrating

[stark ermine]

yes but nothing has been changed in that code, only the crash

[merry perch]

this is the result after stacking a seestar mosaic. I dont know whats causing those weird vertical bands, I think clouds maybe? Feathering did not work, set at 80pixels it created worse low signal borders

any solution to this?

any idea what could have caused this?

I am going to inspect the frames better for clouds potentially, but its weird for sure

There dont seem to be clouds left in the stack

[stark ermine]

Probably the seams because of the gradients. You could apply bge on all subs before.

[merry perch]

how do I manage the fact that the subs shift so I guess one BGE wont be good for all frames?

There was a way to do this to all subs at once if Im not mistaken right?

[stark ermine]

yes, siril can hendle this

[merry perch]

Ok yes I tried that. But how can I trust it to do a decent job on every frame (since the FOV also moves)?

Is there any setting you would recommend? Apart from polynomial as the dialogue warned me

[stark ermine]

I explain everything here: https://siril.org/tutorials/gradient/#removing-a-gradient-from-the-images-of-a-sequence

[merry perch]

so polynomial extraction will be way less destructive than RBF?

[river matrix]

Hey lock, how do I get siril to live stack faster? I am shooting and it is maybe stacking 2 frames every 30-50 subs

[stark ermine]

He takes a new sub when it sees it

you can't speedup it

I would recommend to use SSD if it is not the case

[river matrix]

yea so its detecting it but for example out of 50 subs it only stacked 2 of them

[stark ermine]

ok. Without logs I can't say something

Moreover this feature is not really supported

[river matrix]

Would you have a recommendation for live stacking?

[stark ermine]

Not really. I'm this kind of guy who prefer waiting for the result 🙂

[subtle night]

I actually need to try out live stacking since I might do it for outreach. I prefer doing visual but sometimes I may want to do EAA instead

[trim steppe]

the grax python script really doesnt want to work

[stark ermine]

what is your siril version

did you try the reset VENV operation?

[trim steppe]

reset didnt work

1.4.0 beta3

[stark ermine]

can you open the script editor, write :

import sirilpy
print(srilpy.__version__)

and show me the output

your installation has an issue

and please, telle me what's going on when you reset the VENV

[trim steppe]

it insta crashed the moment i pressed save after putting that in the editor

[stark ermine]

Don't click on save

just run it

[stark ermine]

too old...

the reset venv shoud fix that

what's going on when you try to reset the venv please

what's the output

[trim steppe]

ran it, did the version check thing but it stayed at the same version

[stark ermine]

are you sure you are running the beta3?

[trim steppe]

dont think that can mean anything else

[stark ermine]

I suspect it uses an older venv due to an old installation you probably did with a unstable version

so try to remove the folder I said

[trim steppe]

the venv folder?

[stark ermine]

If you are somehow stuck with an older version you could try manually deleting the whole of ~/.local/share/siril/venv and ~/.local/share/siril/.python_module (or the OS equivalents) and then reinstalling beta3

[trim steppe]

deleted the stuff, getting a new error now

[stark ermine]

could you type again
import sirilpy
print(srilpy.version)

[trim steppe]

well thats interesting

[stark ermine]

hum... did you reinstall Siril beta3 after the directory deletion?

a complete reinstallation

[trim steppe]

i got another installer from the siril website and let it uninstall it and then install it

[stark ermine]

what if now if you reset the venv?

[trim steppe]

did that, gonna try the version thingy again

same message as before

[stark ermine]

ok

so now, redo the

import sirilpy
print(srilpy.__version__)

in the script editor

if it does not work ... I don't know

[trim steppe]

that worked ig

[stark ermine]

GraXpert should work now

[trim steppe]

it indeed does not

[stark ermine]

Hum

Something is weird with your installation

[trim steppe]

oh well

[stark ermine]

Is this windows?

[trim steppe]

yep

[stark ermine]

could you uncheck the GPU acceleration?

and try again

[trim steppe]

does work that way

painfully slow though

[stark ermine]

it does not change anything for background extraction

[trim steppe]

for denoising its extremely slow

[stark ermine]

did you try if it GPU works for denoising?

Because we were speaking about BGE

[trim steppe]

before it was the same error for denoising and bge, didnt try decon but with no gpu accel enabled both work

[mental geyser]

what would be causing this?

been having this issue for a while, only pixinsight is able to stack this data

there very clearly are stars

been having this issue for a but but ive not changed any settings

[stark ermine]

with only this screenshot I can't say

Many things can cause it

Bad reference frame (try to change it)

Update star detection parameters, ...

[mental geyser]

Will send a log file in a few hrs

[stark ermine]

OK I think I've figured it out. I think ypu have an outdated version of C:\Users{username}\AppData\Local\siril\siril_onnx.conf.Delete that file and try again with the GPU enabled

[broken peak]

I may be stupid but is there a way to do narrowband normalization? I'm trying to make an sho palette but it obviously comes out green with just a basic combination

[stark ermine]

The manual color calibration works well

@trim steppe : is it ok for you ?

[trim steppe]

havent tried yet

yep its workin

[stark ermine]

Nice

[subtle night]

embrace the green
I usually just stretch the channels independently with GHS in several iterations to get the result I want.

[broken peak]

works for sho?

[subtle night]

Yeah and it gives you pretty fine control over where exactly contrast is distributed, which is really helpful

[broken peak]

I need to try that, thx

[merry perch]

@stark ermine I tried sequence BGE but it didnt work, the seams look same as before

maybe I should make it more aggressive? gooing from degree 1 to 3-4? or RBF alltogether?

[stark ermine]

it's also a good method

[stark ermine]

no. Maybe this is not the seams and something else, like noise

[merry perch]

weird, I never got anything like this with any mosaic stacked in Siri, it was alwasy seamless

now this needs side to side comparisons to other broadband data

[stark ermine]

this is probably why it comes from images here.

[merry perch]

lest say maybe its caused by my seestar being in a certain position where there may be some extra local light contributing to the LP, then i move it after midnight

I found this situation not to be an issue with other stacks though

[merry perch]

how can it be noise as you say?

but yeah I see there must be something in these frames

as another batch is not showing this

buts its thousands so even if I scanned many of them I may not be finding those responsible for it

Maybe I'll only find out with more subs

[mental geyser]

where do you change star detection parameters when using a script?

[trim steppe]

should have the stuff on the register line

[mental geyser]

thats all i get

no where else i can adjust star detection parameters

unless theres a way to adjust it on the scripts?

[trim steppe]

you need to open up the script file and adjust it in there if youre gonna be using a script

could also just load up the pp_lights sequence into siril and then play around with the register settings there

arguably easier since then you wont have to dick around with having the script go through every single step

[mental geyser]

ive always been on the assumption that the scipts run with star detection on max

[stark ermine]

because no sequences are loaded. This UI is for UI things

you can change parameters in the script

[merry perch]

What? This is very cool

[stark ermine]

yep 🙂

[merry perch]

Is this going to be documented publicly anywhere? Not that I plan on understanding how it works but I'm curious 🙂

[stark ermine]

the code is publicly available yes

and the bug report too

https://github.com/spacetelescope/drizzle/issues/196#issuecomment-3277705196

[subtle night]

Yo I didn't know there was SIRIL merch

[mental geyser]

damn $42 for a shirt

from 2021 as well

[subtle night]

That's like $3.50 in 'Murica Bucks

[trim steppe]

expensive stuff 😔

[merry perch]

There's no price for this kind of aura

Imagine putting on the Siril branded apron to let everyone know you are cooking some insane image

[stark ermine]

you can save your money for a book I will publish soon

[subtle night]

Hey if it funds SIRIL development I will get some merch

[trim steppe]

guess thats why it shits itself during drizzle at times

any idea how to fix that @stark ermine?

[mental geyser]

buy a new pc

[trim steppe]

seems like it doesnt happen when i reduce the amount of memory allocated by a significant amount

[stark ermine]

this is a memory issue yes

Drizzle take a large amount of memory

[trim steppe]

but if i use the ratio of available memory, why does it still fail and say its out of it? on task manager it does not reach 100% usage

[stark ermine]

because the ratio is taken at the beginning of the computation. It is possible that the OS allocate more stuffs during the process and it changes the values

Do you have a browser opened?

[trim steppe]

yeah

[stark ermine]

bad idea 🙂

[trim steppe]

during stacking it happily uses the available ram though

its just during drizzle that it shits itself

[stark ermine]

ok. We will have a look

now, the drizzle weights are 32b when computed, so they do use space... A lot more than in beta3

[merry perch]

@stark ermine this shows the problem more clearly as this is now in EQ, the seams are very precise. I have 4 batches of subs and 2 are affected by it, and as I said it has only happened on M31. As a first thing I can try feathering. Last time I set 80 pixels but it didnt work, is something like 300px okay? is there any degradation of the data by using it?

I even considered scrambling the subframes, but it does average out when stacking all the masters together, almost disappearing, but not completely

this one yet doesnt have BGE hence the gradient

[stark ermine]

You clearly have a gradient issue

[merry perch]

something of that realm it seems

it does persist even withn sequence extraction as I noted the last time so I dont think I need to test that again but I may

[subtle night]

Have you tried normalizing on overlaps?

[merry perch]

Ok so feathering did work this time. Set at 300px, does it degrade the final stack of the subs in any way?

[merry perch]

I unfortunately have thousands of subs it would take days lol

[subtle night]

This might sound stupid to someone more knowledgeable but I just bump the feathering up to like 1000 px and it works fine

Also what instrument did you use to get this data?

[stark ermine]

It should never be used in this case

[merry perch]

Usually Seestar stacks dont need feathering since the subs are so many. Infact it comes out pretty perfect every time, this is the exception

[stark ermine]

Ithe doc is clear about it.

[stark ermine]

you need to feathering yes

I thought you did

[merry perch]

I did 300 pixels now. I usually dont activate feathering because the stacks come out smooth anyway but this time it looks like it needs that

I was told Seestar stacks dont strictly need it but this looks like an instance where it does

I just want to know if it impacts the final snr in any way, just to be sure

[stark ermine]

put the max

1000 px is good

[merry perch]

It works so thats good, problem solved pretty much

[stark ermine]

HDR multiscale will be soon available in Siril

[mental geyser]

heres the log

[stark ermine]

other examples

[stark ermine]

I released the script

[subtle night]

Thanks for the new script!

[gusty spoke]

@stark ermine any plans on adding this new stacking method?

https://youtu.be/A1uSOnMaME8

[subtle night]

For reference, here's the paper on ImageMM

Seti Astro also has a white paper on its implementation

[gusty spoke]

I’d love to see this in Siril, SAS Pro is a great tool but it’s stacking just never works for me

[stark ermine]

We have so much priorities before.
First stabilize 1.4.0 and release it.

[stark ermine]

some of us in the Siril team (not me yet) have read the original paper and the white paper.

They have some doubts about the seti astro implementation

[trim steppe]

sounds about right

[wise creek]

when you got this message, was it using max framing?

[trim steppe]

not entirely sure but possible, i remember it crashing on current and center of gravity

[wise creek]

Any scale being applied?

[trim steppe]

1,5 iirc

[wise creek]

ok, thanks... looking at the memory consumption calcs right now, the one for the registration process is probably the worst we have in siril

[wise creek]

Sorry more questions:

• did it happen with beta3 already or is it new with beta4?
• do you build siril or should i send a test version if I need to get more info?

[trim steppe]

it was a thing on both, just didnt look into it much on beta3 (aka didnt check console to see what happened)

never tried building, would need a test version

[wise creek]

ah ok, so it's not related to the changes we've made with beta4. I will make a version which logs a bit more info about mem consumption than the usual and send you a download link to a test version. which OS?

[trim steppe]

windows

[wise creek]

In 20 min or so, there should be a download button to the right on this page: https://gitlab.com/free-astro/siril/-/jobs/11598278581
Download the zip, extract it to say C:\temp\astro. Open a cmd window and then type:
"C:\temp\astro\siril\bin\siril.exe" > siril.log 2>&1
When it opens, just do the apply registration step and let it crash. It should save a siril.log in %USERPROFILE%. If you could share this

[trim steppe]

what folder does the log file get generated in?

nevermind, i cannot read 😭

[trim steppe]

issue

the log file doesnt generate right

[trim steppe]

imma dip for now, gotta wake up real soon, i can try more stuff tomorrow night most likely

[merry perch]

@stark ermine is there a reason why Siril does this when I absolutely have enough space for this? I registered the same amount of frames if not more earlier and it only took 50gb

[stark ermine]

The operation asked 501Gb here. So this is not the same who took 50Gb

For sure

[merry perch]

This is very weird

Its 1745 subs being registered with maximum framing

Each being 3.9 mb

So registration with maximum framing will take more

But that much? I am using 1x drizzle

I am going to redo the sequence once more

[merry perch]

Maybe one or two subs make the mosaic so big it has to compute an insanely large image?

Because for just 500 more subs that wouldnt meet the demand

I think I should check for shift on the axis?

[stark ermine]

Maybe, difficult to say

[merry perch]

Thats not the issue apparently its doing it with another batch

Never did this before with a mosaic

And its with drizzle only

It didnt have any problem with normal interpolation, and its 1x drizzle too

Do you think it could have anything to do with the recent drizzle fixes? I am going to see what happens on beta 3

[trim steppe]

still cant get the log stuff to work

[merry perch]

Yes, works fine on beta 3

[stark ermine]

Beta4 added weights

These are files that takes a lot of space

[merry perch]

Maybe its that combined with that specific dataset?

Because the other batches of the data take way less

Definitely not 500gb

[trim steppe]

got it to work have fun @wise creek

[wise creek]

Can you show a log?

[wise creek]

Thanks, I'll try to have a look

[merry perch]

I'll send it once I am done stacking in beta 3, I'll reinstall beta 4 and then just export the log up until apply registration?

[wise creek]

I also need you running apply registration, that's when the output space is computed

You don't need to "install" both, you can just download beta4 portable and use it from there

https://siril.readthedocs.io/en/stable/installation/windows.html#installation-of-the-portable-binary

[merry perch]

A normal "export log" from Siril itself?

So I click apply and it'll provide the data you need even if it fails with the storage warning right?

[wise creek]

Yep

[merry perch]

here it is @wise creek

hope this is what you need

[wise creek]

In 20 min or so, there will be a download button here: https://gitlab.com/free-astro/siril/-/jobs/11623519179. Download, unzip the archive, and test again using siril.exe located at ./siril/bin/. I've changed the output storage calculation which was very conservative when using max framing for a long seestar mosaic. I think the output size should be much more accurate now. In the log, I've temporarily added the expected output size, just below the drizzle parameters, smthg like that:
22:45:40: Required storage space: 4889 MB

If you could check that this works and correctly reflects the storage used (counting both the r_testC_ images and the weights saved in drizztmp folder)

[wise creek]

OK, many thanks for the log. So, I've checked the computed requirements, they look correct. My best guess is, based on the log:

Memory required per thread: 1518 MB, per image: 621 MB, limiting to 8 threads```
Per thread, we have 2 big ticket items, which are the output image (621MB) and its weights (621MB) as well (plus the input image, plus the transformation maps but ok). As 8 threads are used, there needs to be 16x621MB blocks (each contiguous) in your RAM plus 8x207MB blocks (and the rest). Depending on how the memory is fragmented when we start the process, it is very likely that theses blocks do not exist, hence why one of the allocation fails and stops the process. Long story short, 8 threads is too much for this size of output images.
We've had another instance of this contiguous block allocation problem on Windows at another step (stacking a very large mosaic), and the solution will be long to implement, probably not in the 1.4 cycle. We basically need to change completely the way we allocate for Windows...
So in the meantime, my only "solution" is either you decrease the number of threads or you reduce the percentage of used memory (which will in turn limit the number of threads).

[subtle night]

So I'm noticing, whenever I use SIRIL and share my raw stacks with people, the stacks are always mirrored in images others generate from my data. I don't care about which way north is (though usually north is up in my stacks), but I do care if my data is unexpectedly mirrored, and I'd like to figure out where that's happening.

When I import my frames from NINA, they are mirrored top to bottom even though the preview windows shows them with the correct orientation without selecting the mirror option. I think most people who process my data are using Pixinsight. So I guess either NINA is not writing the ROWORDER keyword, or Pixinsight is not respecting it?

I guess SIRIL is displaying images in the bottom-up order regardless, so opening the FITS files from NINA and getting mirrored images at that stage is expected.

[stark ermine]

This is perfectly normal because siril reads data bottom up, while many other software do top-down.

The standard should be bottom-up. All professionnal software read FITS files bottom-up.

ROWORDER is not here to change image orientation but only to correctly debayer the image

Apply a platesolving and your image will be flipped in the right orientation.

https://siril.readthedocs.io/en/stable/file-formats/FITS.html#orientation-of-fits-images

And NINA is writing ROWORDER

[merry perch]

Thanks, I'll try it when I can and let you know

[trim steppe]

so thats why reducing the percentage made it work eventually, how many threads should i limit it to?

although ill prob be changing it constantly since stacking is fine with taking more for whatever reason

[wise creek]

No idea, the computation depends on the current fragmentation of memory at the time you start the calc. Hence why it will require some thourough development.

If you need to change it often, you can use the set command instead of opening the Preferences dialog

[trim steppe]

[merry perch]

Ok so I dont get the storage warning anymore and its proceeding to register. Once its done what do you need to know?

Ok once its done I'll count if both r frames and files in the weights folder are the same number?

Yes they are the same amount

Since now it works is it worth m restack my drizzle data all in beta 4? Do the weights cause a noticeable increase on final quality?

[wise creek]

The drizzle algo is now what it is theoretically supposed to be. Now whether or not it will bring significant enhancement to your result is hard to say

[stark ermine]

for your information

[mental geyser]

thought you guys were finding it questionable

[trim steppe]

setis implementation of it

[stark ermine]

This is why we want to try our implementation

[mental geyser]

@stark ermine siril website is down

(in case you didnt know)

@wise creek

[stark ermine]

ok thx

[stark ermine]

Fixed

[subtle night]

Do you need to select "use dark" to get cosmetic correction to work? I tried this again on 1.4.0-beta4 with a hot pixel map and found that it still did nothing, but I only selected a bias and flat frame.

(I find that my darks are actually introducing banding to my images, so I use a median stacked bias frame and it gets me far better results)

[stark ermine]

Yes, dark is always needed

[limber parrot]

im 1 month late but thats absolutely awesome

[stark ermine]

the script is now available

[limber parrot]

ooohh cool

will it ever be made into a single window? kind of like what pixinsight does

[stark ermine]

No probably not

[trim steppe]

is flathub not a thing anymore or

nvm

[subtle night]

Does this script allow for common registration of different monochrome data channels? That would be a really nice feature

[subtle night]

Oh also, for all you SIRIL enjoyers out there, I want to draw your attention to this issue: https://gitlab.com/free-astro/siril/-/issues/984

Damped Richardson-Lucy deconvolution would be a big improvement over what's currently available in SIRIL; it amplifies noise a lot less.

I may try to provide a Julia implementation of this algorithm for DeconvOptim.jl

The relevant paper is available here: https://www.adsabs.harvard.edu/full/1994rhis.conf..104W

[obsidian obsidian]

Clonestamp