What's missing from Siril?

[sage glade]

Yeah that works

bruh how I didn't thought of that

[split smelt]

Almost like it was designed for that 😛

[sage glade]

🤣

[mental geyser]

@stark ermine is this the dev version or another 1.2 release?

nvm i answered my own question

still getting this error on windows

[waxen ember]

is the directory there?

in file explorer

[mental geyser]

no bc siril isnt finding it

[trim steppe]

i get that error as well, not sure what it actually means, dont think it really breaks anything though

[raven lodge]

You don’t need cygwin unless you’re actually building

[waxen ember]

dev version quirks

[stark ermine]

Cygwin is needed for local astrometry.net

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

[raven lodge]

I tried the local solver for a while, it’s not really worth the effort, at least on windows.

[stark ermine]

Indeed. Astrometry is quite good in siril especially in the dev version

[tall sable]

I love how the platesolving is pretty much instant

[junior hound]

Yes as long as you got the right pixel size and focal lenght it's really fast and works almost everytime

I might have some time this week to write the "things to improove" list btw

[merry perch]

I dont know if anyone is experiencing this but in version 1.3.2 alpha dev it crashes right after sequence processing is succeded when applying existing registration

[stark ermine]

Would be nice to produce logs

[tall sable]

I had this issue once but I forgot how i fixed it 😭

[merry perch]

The only anomaly in the logs is the "cygwin/bin was not found at C:\Users\vesse\AppData\Local\cygwin_ansvr\bin - ignoring" - the rest of the console outputs normally as usual, and I noticed the crash only occurs when drizzle is selected. If you want I can send you a transcript of the console right before the crash occurs

[stark ermine]

we need more detailed logs: https://siril.readthedocs.io/en/stable/Issues.html#send-us-useful-information

[merry perch]

after starting Siril from the cmd I just replicate the crash normally and the log will save in the mentioned folder?

[stark ermine]

yes

[merry perch]

here we go, I hope its useful

the first lines are in italian even if its set to english for some reason but the rest is all english

[stark ermine]

ok. It should not crash for sure, but you have an inconsistency

log: Image #61: number of layers (1) is not consistent with sequence (3), aborting

[merry perch]

and usually what does this mean?

one of the images is corrupted or isn't a valid file?

[stark ermine]

You have a mix of images

Image with 3 channels and images with one channel

[merry perch]

technically all the subs I selected are color images

[stark ermine]

Here, your sequence was based upon a 3channel images. While image #61 has only one channel

so probably Siril tried to detect stars on the green channel but it does not exist

so siril crashes

[merry perch]

ah okay now I understand

tried on other datasets and it also crashes

[stark ermine]

probably same issue

[merry perch]

but only when drizzle is enabled

[stark ermine]

Then check your images, especially the 61

This is why the log gives the image number

[merry perch]

Thanks for the help, im going to check and see what happens

[stark ermine]

now. If you can reproduce the issue each time. This is maybe a bug somewhere in the CFA Drizzle. So detail what you are doing.

However, this is the first time someone report it. And many users play with the CFA Drizzle

[merry perch]

My steps are: convert images into fit sequence - no calibration frames - two pass global star alignment - apply existing registration and select "use drizzle" without changing any of the settings

This time im trying again by removing image 61 from the sequence

now same thing happens but with image 55, leads me to think that maybe drizzle has some issue with Seestar images or I am doing something wrong

[stark ermine]

are your images RAW images ? Seestar images are not different from others

There are no reasons the Drizzle would behave differently

[merry perch]

Yes they are fit files

[merry perch]

Yeah also because it works with other users

[stark ermine]

Yes but RAW ? I mean debayerd or not ?

[merry perch]

Technically not as they show as B/W files and they need to be debayered during conversion in fit sequence

[stark ermine]

ok but in your case you MUST not debayer !!

https://siril.readthedocs.io/en/latest/preprocessing/drizzle.html#cfa-workflow

[merry perch]

Ohhhhh I get it Im so sorry

[merry perch]

Now it runs

[stark ermine]

Good

[visual roost]

https://youtu.be/e7vcA4JERBo?si=DBAd1Hr-HT_OxXg8

Could we see something like at 6:01 in the future?

(low frequency and high frequency separation)

[merry perch]

Is it normal for the same image to look slightly different (mainly the color and contrast) when processed through Siril's GraXpert integration vs GraXpert's standalone? All settings and models used are the same

left is GraX on Siril, right is GraX standalone

[sage glade]

Hey @stark ermine what do you make of this at something that was posted at #💬-general-chat https://www.youtube.com/watch?v=Cs4aybq-h54

[waxen ember]

the stars look goofy

[sage glade]

The thumbnail look groovy tho I gotta admit

[waxen ember]

new spiritual rebirth dimension type of thumbnail

[sage glade]

Bruh we can judge the ai we use it all the time

[raven lodge]

Normal, yes. Intended? Not sure.

[raven lodge]

^

[stark ermine]

Windows only...

[sage glade]

Yeah and I can't even run it because it doesn't support Intel graphics card

😢

[stark ermine]

be patient

[sage glade]

Yup I already contacted them though email and they said they'll fix this asap

[waxen ember]

integrated or the new evos?

[sage glade]

A770

[waxen ember]

oh yeah, i meant arcs

[raven lodge]

I wonder how that compares to astrosharp? I never got on with astrosharp unfortunately

[sage glade]

Idk man

[merry perch]

It almost looks like as if its getting rid of green noise automatically and standalone isn't

[raven lodge]

Could be. But it could also be because of colour profiles maybe?

[stark ermine]

take a screenshort of the whole screen please, for each case

[merry perch]

solved, now it looks the same on both, don't know why that happened

[stark ermine]

strange

[tall sable]

how can I use these built in scripts without having calibration frames?

[mental geyser]

for Ha, just stack normally and extract the red channel

for Oiii i think you just combine them together??????

[tall sable]

does that do the exact same thing as the script

or does the script do some kind of magic

[mental geyser]

sshould be the same?

[stark ermine]

Small question. Why don't you use no calibration frames?

[stark ermine]

This is not the same no...

[stark ermine]

Again

If you do it normally, you will have a debayer. The debayer interpolation will mix Ha and OIII together. This is not wanted

[raven lodge]

You have to copy the scripts and delete the lines that stack and apply the calibration frames

[mental geyser]

What's the difference?

[sage glade]

Btw I would've waited for Christmas or January to get the bugs ironed out

But I use it to sharpen the starmask stretched even he said not to. And for that works fine. When it comes to the nebula there's a bug where if you have a bright line in the nebula the ai will be confused and assume that they're actually stars which is false. I'll post some screen prints to show you

[stark ermine]

Because of the debayer

[ancient talon]

So extract Ha/Oiii script splits the CFA channels and then stacks that?

[stark ermine]

yes

[waxen ember]

[stark ermine]

?

[sage glade]

🐟

[waxen ember]

siril is super cool

[sage glade]

@raven lodge you there?

Any way when you see this here's my expirience from time I have been using Cosmic clarity:

Although the guy says that you sould be using it unstreched in messes with my image so I can't make it work

In regards to the streched image it's working but it lowers the contrast in the image so I have ever so slightly to move the black point more to the right

Here are the starless

Before and after

And the starmasks

Before and after

Which neatly brings me to you @stark ermine because after I run the images from the sharpening tool the are really bad and I want to know if there's anything I can other than sharpening the stretched one

[tall sable]

I use a seestar s50

it’s just not possible

I guess I use darks but I don’t get access to the individual darks

[raven lodge]

I'll have to take a look on some of the images I've taken. I'm not sure I like what it's done to the neb there, it's sharpened the outline but not added any extra detail inside.

[tall sable]

I tried that and it doesn’t work

[raven lodge]

Are you using 1.3? If so I'll send you my modified Siril script that does drizzle

[tall sable]

I tried both the dev version and recent release

[sage glade]

This is the dataabse from @astral forge and I have no idea why the program does that 😦

[raven lodge]

There is a dedicated seestar script you can download for 1.2 but it doesn't drizzle https://siril.org/tutorials/seestar/

[tall sable]

I know but stacking isn’t my problem

[raven lodge]

Ohhhh

[tall sable]

I need to extract just the Ha from the dual band filter and get the mono file

[sage glade]

Just don't forget it lowers the contrast as well

Before and after

It's a bad tool if I learn how to use it I'll make it a hell of banger but that is one of it's shortcomings tho

[ancient talon]

You can stack normally and just extract the color channels

[sage glade]

Pointlism

[sage glade]

Hey _Krzy have you tried this Cosmic Clarity sharpening tool?

[tall sable]

that isn’t the same result as using the script

[ancient talon]

His repo threw an error when I tried to use his pix script

[tall sable]

the script uses pixel math to extract ha

I also tried that already and it sort of worked

[ancient talon]

What pixel math? Does it subtract Oiii leaks?

[sage glade]

Btw it's only windows.But that's strange.Any idea what happened?

[tall sable]

not sure, but that’s what deepspaceastro told me when I asked in a comment in his video

[ancient talon]

Idk. I do use windows. But from his comparisons on his website it doesn't look like it does a lot of actual sharpening and instead weirdly stretches the image

For the stars it makes a difference i guess

[sage glade]

That much I figured which is actually for the best. You need stuff where stuff doesn't exist.The error in the terminal. Did you managed to take a photo of it?

[waxen ember]

what am i reading, please rephrase

[sage glade]

No rephrase only image

gimme a sec

[waxen ember]

[sage glade]

I can't believe you have phish it looks so cool

[waxen ember]

you know you can take screenshots?

[sage glade]

That error @ancient talon was due to me having an Arc A770 graphics card but he fixed it within an hour

[waxen ember]

doubt he has an intel gpu

[ancient talon]

No for the script his repo threw an error for me and the script didn't install

[sage glade]

The terminal was going black an it was .3 sec massage

[waxen ember]

aah ok

[sage glade]

Did you contact him?

[sage glade]

Rtx 4090

[waxen ember]

anything above 3060 has CUDA cores, which lack thereof caused the error

[sage glade]

Really?

A770 doesn't have CUDA?

[waxen ember]

no, CUDA is a technology made and patented by NVIDIA

[ancient talon]

Nah I cba

[sage glade]

Ahhhhh I see

[ancient talon]

I didn't see anyone else mention it so I just figured it's a sign that I should go back to working on my project and not be distracted

[sage glade]

Bruh

[ancient talon]

Maybe I'll check it out at some point but it doesn't look promising from what I've seen

[sage glade]

it never hurts to send a massage.Give it time no need to doom something from the 3rd day

[ancient talon]

I'll try it before I make any final judgements

[sage glade]

https://tenor.com/view/dancing-cat-oh-yeah-dancing-cat-gif-10961226009468952545

[ancient talon]

it's just that his own showcase made it seem like it doesn't do much lol

for nonstellar at least

[sage glade]

It's 5 GB file in which in his video said that he tried this ai at 7200 images

But hey even if it's slightly better than nothing it beats to having that than to buy 600 euro software

Right?

[ancient talon]

size of the training dataset doesn't really tell you much about the performance

and yeah I'm not expecting it to beat blurx. I'm just hoping it does more than I think it does

[sage glade]

Well they were saying that for noisex as well and although Graxpert didn't beat it it got exptremely close for free

[ancient talon]

deepsnr beat it and its free. I think graxpert beats noisex too but I'd need a proper comparison.

[sage glade]

Tbf I can't believe that these programs have the ability to change the market so much

[ancient talon]

graxpert is working on machine learning decon. Hopefully it will be a good one

[sage glade]

I'm actually curious

Could astrophotography be to the Artificial Intelligence what videos games are to Graphics Cards?

[ancient talon]

what? I'm 99% positive it doesn't. @stark ermine, can you confirm?

[stark ermine]

You also can do it without script. Siril is not only script 🙂

[stark ermine]

Please, write a summary of the question 😄

to many lines to read 😄

[ancient talon]

does extractHaOiii script subtract Oiii leaks from Ha data?

or does it just extract CFA channels and stack that?

[stark ermine]

no. Ha is taken from red pixel only

OIII from Blue and Green

[ancient talon]

right

Astrophotography youtubers trying not to spread misinformation challenge:

[stark ermine]

who spread that?

[ancient talon]

Deepspaceastro it seems

[stark ermine]

oh

[tall sable]

oh, ok that’s helpful

[sage glade]

Ay yoooo @stark ermine look it came to Mac os as well

[stark ermine]

Yes. But not on linux.

And tests are not very good from what I've seen

[sage glade]

Bruh. What do you mean the tests?

[waxen ember]

the results.

[sage glade]

Ehh I'll after a month of use for only starmask it helps

[stark ermine]

Tests I've done

[sage glade]

I only use it for a stretched star mask so I don't mind all this much

[mental geyser]

his processing method is interesting... in one of his videos he colour calibrated his image and then run SCNR on max straight after

[split smelt]

why would you do that?

[ancient talon]

Because "remove green noise" sounds like a good thing if you don't know what it's doing

[mental geyser]

They also don't know its very data destructive

[ancient talon]

Which, don't get me wrong, can be a good thing. But most of the time people misuse it

[sage glade]

Hey @stark ermine I've stumbled in a bug where I am in autostrech doing RGB composition and it presents the channels white even before I put the all in

Thumbnail attached

[stark ermine]

You only load a channel

so this is normal

[sage glade]

As you can see this is only the red but it shows white and with all the channels

[stark ermine]

hum, if you close and reopen the tool?

THis too has been completely refactored in 1.3

send images to test

[sage glade]

Doesn't work. I'm posting it here because it is the second time that it has happened and it solved the last time with restarting the pc

[stark ermine]

send images to test

[sage glade]

https://discord.com/channels/794642864218439681/1288625466457985064
These are the images from this particular expample

[mental geyser]

he wants the images you are trying to compose together

[sage glade]

Yeah I know the data from this guy I tried to compose and this happened

[stark ermine]

Is this FITS files?

[mental geyser]

never seen an FTS file before

[stark ermine]

??

Oh

FTS = FITS = FIT

[mental geyser]

siril can open the image

[sage glade]

Ahhh ok then we have found the issue after I denoise them an extract them from from Graxpert as fits the RGB comp works as it should so from now on I'll be looking for this weird formats

[stark ermine]

My guess is that you should convert these FITS into Siril FITS.

[sage glade]

Yeah I'm already prossesing them at Graxpert and extracting them from there 🙂

Thanks @stark ermine I'll send a massage to this person as well have a great Saturday

[stark ermine]

thx, u2

[split smelt]

@stark ermine remember a while ago, I reported an issue where narrowband 03 stack was producing strange results? It's still doing it

If I go to stretch it, Siril offers to rescale pixel values. After that, it looks fine

[stark ermine]

Ok. There's no issues for me. Images with low signal can lose the signal while calibrating. This is why we allow negative pixels

[stark ermine]

you can use the command limit too.

[raven lodge]

@stark ermine if you check the “show sensor tilt” option it does not get updated if you change to a different frame in the sequence. To update the tilt display for a new frame you have to toggle it off and on again.

[split smelt]

Yeh, this is HDR mode on my Risingcam. I think I'll avoid using that for narrowband in future. I'm reasonably happy with the final image though, but this is begging for a mosaic to capture more of the region

[stark ermine]

Indeed. I know that

[stark ermine]

Congratulations to @split smelt whose patch have been implemented into Siril

[sage glade]

Hey @stark ermine I've downloaded the dev version and so far I'm having a blast.2 questions tho
Can I turn of the window that pops up and asks me if am I sure that I need to close this image?
And is it possible for the star recomposition tool the settings to be changed and the new setting become the default?

[stark ermine]

1 - which dialog please?

2 - not yet

[sage glade]

Sorry this was going for the first

2:There's some weird colour in the stars in the human weighted luminance and I'm trying to desaturate them by using the default independent channel values\

[stark ermine]

Oh this one cannot be swithced off

[sage glade]

Bruh

[stark ermine]

🙂

this is because when doing this you couls lose your data. Imagine that this dialog can be open by clicking on its entry in the menu. Maybe when lauch from starnet we should remove the popup

you have the same when oading RGB comositing tool if you already have an image

😉

dev version is very different from the last stable

the dev version is 1.3

yes

Idk

[sage dove]

@stark ermine where can I download the siril dev version?

[ancient talon]

https://gitlab.com/free-astro/siril/-/artifacts

[sage dove]

thank you

[trim steppe]

can get them from both afaik, not sure if theres a difference

[ancient talon]

pipelines is probably easier but you can get the same files from both iirc

[waxen ember]

at 100% it probably does a lossless compression

that 5% to 95 can enable so much optimizations, that's why the jump

greedy bastards 😔

[raven lodge]

100 is lossless, 99 isn’t

TBH Siril should default to 100 for jpegxl

[ancient talon]

if you want to save in lossless compression png is probably a better choice

it has better compatibility and I think it weights less too

[raven lodge]

No it’s worse, png is larger and only 8-bit

[ancient talon]

it supports 16bit

you don't need more than 16 bits when saving a processed image anyway

because the JPEG was 32bit probably

[raven lodge]

You should not be storing images in a format that has lossy compression at all. PNG is not a good store but it’s best for sharing.

If you want to reduce file size at 100 quality you can increase the effort. This trades file size for time to save.

[sage glade]

But why would I want the loosless since it's bigger size than the tiff. If it's really good at compression then I don't think it loses such critical info that it's deemed bad

[sage glade]

I know but 70 MB out 470 it's not really important at that sizes

[ancient talon]

lossless JPEGXL is 65MB, XISF with ZStandard is 76MB and original is 100MB

for one of my images at least

[sage glade]

I'll go to my pc in couple of hour and I post screenshots here from my files and the Siril terminal to show what and how

[raven lodge]

I’ve only been using jxl for the last few goes, but they always seem smaller than png and compressed tif.

[sage glade]

I'm tired now imma do it in the morning

[sage glade]

Hey uhm @stark ermine what does this massage means?

[ancient talon]

you can safely ignore it

cygwin is needed for using astrometry.net locally iirc

[waxen ember]

this is a nuclear warning

your computer will undergo fussion

[sage glade]

Hey uhm @stark ermine I have a problem

Thumbnails attached:

[stark ermine]

Without information, difficult to say something 🙂

[sage glade]

I was editing the starless and when I opened it in the star recomposition tool it became red.But after I opened the edited starless it was appearing as I left it. I closed the siril and opened it again and then when I fused the starless with the starmask and it popped a massages about rescaling the image and removed some details in the black areas. So I've deleted the 2 files and rerunned starnet and I didn't had any issues

Also is it possible to know why the photometric colour calibration tool is greyed out?

1:Color calibration
2:Photometric color calibration
3:Spectrophotometric color calibration

[stark ermine]

Because you need to platesolve first

a lot of things have changed

for good reasons however

[sage glade]

Ah ok thanks.

[worn pendant]

idk if this has already been mentioned, but I love that you can just open endless images in pix and have them side by side

[stark ermine]

You can open two siril

The philosophy of Siril is different 🙂

[mellow vine]

@stark ermine can siril tell me the integration time of my stack? if so, where

[ancient talon]

its in the name im pretty sure

it's result xxxx s

[mellow vine]

mine just comes up result.fit

I dont get that

[mental geyser]

You need to add something at the end of the script

I'll get it

[mellow vine]

Please do

im using osc preprocessing without darks script

[mental geyser]

Oh I'll just send you the script then

Gimme a bit

@mellow vine do you use the dev version?

so basically need to go to the bottom and add this

[stark ermine]

It is in the FITS keywords

[mellow vine]

where's that?

[stark ermine]

In the header. You can read it with Siril

[sage glade]

Hey uhm @stark ermine Siril is closing everytime I do negative transformation in a starmask and I press remove green noise.I don't have thumbnails because it simply closes and I reopen it. The starmask is from a narrowband image.So I combined the HSO

But before I do the negative tranformation it shows this massage

And then I press proceed and it does the negative transformation

[stark ermine]

which button do you click here?

[sage glade]

Proceed

And I made it work by deselecting the "maintain the brightness" from the remove green noise popup window

If I have the "maintain brightness" choice in the remove the green noise window it closes automaticaly

[stark ermine]

ok, let's me try. Probably becaus of negative values.

[sage glade]

Hey @stark ermine do you want me to give the data in which this has happened?

[stark ermine]

no, don't worry

[mental geyser]

whats the password

cant extract the file without it

[mental geyser]

@stark ermine

[split smelt]

not sure why it's passworded but it's not hard to work out. It's "sirilic"

[mental geyser]

oh

[stark ermine]

yes sorry. I'm not the dev of sirilic so I don't know 😄

I never use sirilic

[mental geyser]

oh thought you would lol

[stark ermine]

oh no. Siril in manual is enough for me 🙂

and I do multisession ....

I'm currently processing 31h of OIII filter

[sage glade]

What's the black hole there ?

[stark ermine]

an horrible halo around nu And... StarNet fails to remove it completely

[sage glade]

Damn is that 31 hours?

It's so faint

[stark ermine]

Yes but you are looking continuum subtraction

[sage glade]

Ahhh because you want to enhance it right?

[stark ermine]

The stack output:

[sage glade]

It's even less visible

[stark ermine]

You don't see the arc

[sage glade]

So how much did you managed to pull out with the subtraction?

[stark ermine]

What do you mean?

[sage glade]

Well you said that the continuoum subtraction is for faint targets or for low integration time which helps to enhance the signal

[stark ermine]

not low integration tome

only for very faint target

In fact, when shoot in a narrowband filter. Your image will be polluted by broadband signal

If you remove this broadband signal, you will only see the NB signal

[sage glade]

Ahhhh so this is what you subtract

[stark ermine]

THis is why I also shot 4h of Green

[sage glade]

And how much does it change the image?Is it the image you've shown with the gradient form the star which is the black hole?

[sage glade]

In order to say to Siril this broadband

[stark ermine]

This image yes, shows only (except the black hole) the signal from OIII

[sage glade]

You got quite a bit of siglan 😄

[stark ermine]

indeed. You see nebula in OIII

[sage glade]

But also you got this nice blob as well

What is it?

[stark ermine]

a dust

I have some images, not a lot, where flats does not fix the dust. Probably dust has moved during the night

[sage glade]

OOF

That's a problem

[stark ermine]

I will ignore the dust and the black hole and it should be nice

[sage glade]

How do you know this is dust and these are nebula?

[stark ermine]

Shape

ANd I'm looking image in negative view

Nebula are black

[sage glade]

Ohhh that's smart and it makes sense when you thinking about it

[stark ermine]

the positive view

[sage glade]

Damn that star will be a big problem

😦

[mental geyser]

Call it a new discovery

[sage glade]

Hey uhm @stark ermine get you tell me where I can get a guide doing that?#💬-general-chat message

[stark ermine]

There is no special guide

[sage glade]

I'm going to try and make one 😄

[stark ermine]

Good idea

[mental geyser]

yay version 1.3.3 is out

[trim steppe]

whats it got

[mental geyser]

no idea lol

[sage glade]

https://tenor.com/view/snow-snowball-throw-street-sign-gif-6718291626345698175

@trim steppe

[sage glade]

Hey uhm @stark ermine I have a question.I have done contiunioum subtraction but can you help me because my stars are like that afterwards

These are the settings I used in the pixelmath window

[stark ermine]

continuum subtraction is done on starless

[sage glade]

Ok nobody told me that

[split smelt]

I played with this today as well. Do it on starless. I also created a combined greyscale image to use instead of just using one colour channel (R + G + B, then save it as C and then load it into pixelpath)

then, I stretched the continuum image & the RGB image, then split the channels again, then combined like this:-

it seemed to work but I have no idea if I did it properly 😄

[stark ermine]

looks fine

My last integration in G:

Using same formula

[sage glade]

Old Ha

New Ha

[stark ermine]

you should fix the gradient before

your images must be cleaned

[sage glade]

REALLY?Why nooone tell's me these things

[split smelt]

we're telling you now 😛

[sage glade]

https://tenor.com/view/black-dude-crying-cry-emotional-gif-14046052

Bruh where's a youtube video?

[split smelt]

You could create one?

[stark ermine]

Working with cleaned image looks obvious IMO 🙂

[sage glade]

No yet

[sage glade]

Yeah but you said they should a little prosessed as possible

Also should I denoise too?

Too late I've denoised at .8 at graxpert both Red and Ha

[stark ermine]

denoising is a good thing for me at this step

[sage glade]

Old Ha

New Ha

[stark ermine]

It"s better

[split smelt]

@stark ermine have you considered expanding scripts to give them the ability define inputs which would pop up a box where you could enter values when run interactive? Perhaps when run non-interactive they could take inputs from command line?

[stark ermine]

We are working on plugin mechanism

[ancient talon]

I find that removing stars before continuum subtraction on galaxy images usually eats some of the H II regions

The stars residuals will be covered up when you add RGB stars anyway

[ancient talon]

it's way undersubtracted. M110 should be pretty much entirely gone after CS

[sage glade]

Eh it is what it is right now I'm not bothered to do it

Next time. It's not like there isn't data to practice on

😉

[trim steppe]

it really does enjoy yanking them out

[split smelt]

@stark ermine that Siril Python Interface MR looks fantastic. Perhaps I could have used that for the Unpurple filter. I might try to write a version based on that MR to see if it would have been possible

[stark ermine]

Don't waste your time I think. If you want to develop python script, please do it with GraXpert for example. The goal will be to use this for external software 🙂

and of course to create new tools

but as you already coded your filter 🙂

But in fact, I'm nobody to tell you want to do. So feel free to have fun 🙂

[split smelt]

will a python interpreter be built into Siril or will it be something to install separately?

[stark ermine]

Siril will be able to read python script. We are working on it. But this is very new

[stark ermine]

@split smelt : https://siril--96.org.readthedocs.build/en/96/Python-API.html

[split smelt]

Great, I shall take a read

It looks straightforward to be honest

[stark ermine]

Sure. But this is a first draft

Things could change

[split smelt]

I like that you're keeping the interface simple though

[split smelt]

@stark ermine I played with the Python plugin feature. I was trying to do the following:- get Siril to save current image to a temp file (using siril.cmd("save", tmpfilename)), load it with Astropy, do something to the image, save it with Astropy, then issue another siril.cmd to load the file back into Siril. Seems Astropy uses signal handlers which don't work in embedded Python interpreter. It throws an error like so "log: Error in Python script: signal only works in main thread of the main interpreter". This is a shame but perhaps I was trying to do something beyond what it was intended for.

[stark ermine]

OK, Adrian tell me that appears fixable. He think if we start the python interpreter in its own thread and save the GThread reference, then as long as the script is run in the same thread then it should work. We will try to get it working now.

[split smelt]

Great 🙂

[stark ermine]

We need your script however 🙂

[split smelt]

``import siril
import tempfile
from astropy.io import fits
import numpy as np

def hello_world_siril_astropy():
siril.log("Starting the process...")

with tempfile.NamedTemporaryFile(suffix=".fits", delete=False) as temp_file:
    temp_filename = temp_file.name
    siril.log(f"Temporary file created: {temp_filename}")

# Save current file
siril.cmd("save", temp_filename)
siril.log(f"FITS file saved: {temp_filename}")

# Use Astropy to open and perform a simple operation on the FITS file
with fits.open(temp_filename, mode='update') as image:
    siril.log("Opened FITS file with Astropy")

    # Modify the FITS file data
    if isinstance(image[0].data, np.ndarray):
        image[0].data *= 2
        siril.log("Modified FITS file data using Astropy (multiplied pixel values by 2)")

    image[0].header['COMMENT'] = "Modified by Astropy"
    siril.log("Added COMMENT to FITS file header")

    image.flush()
    siril.log(f"Changes saved to FITS file: {temp_filename}")

# Load back into Siril
siril.cmd("load", temp_filename)
siril.log(f"FITS file loaded: {temp_filename}")

siril.log("Process completed successfully")

Run the function

if name == "main":
hello_world_siril_astropy()``

it's a stupid, do nothing good, hello world type of script

[stark ermine]

please quote between code ```

thx

[split smelt]

basically, just to see if I could save the image, do something in Astropy, and load it again. If this works, then more useful things can be done also

[stark ermine]

sure

[stark ermine]

@split smelt : this is fixed

[split smelt]

great 🙂

[stark ermine]

Just for fun. Ha continuum subtracted... Far easier than OIII. And I only have 10h of Ha... Need more.

[mental geyser]

dust mote doesnt look too big an issue

[stark ermine]

No dust on these layers 🙂

Just Ha and R here

[trim steppe]

man i wish i could get that background ha

[stark ermine]

It will be better with more exposure. It is not enough yet. But weather is terrible

[sage glade]

Yooooo @stark ermine you even got the Ha tidal streams in the core

[sage glade]

Hey @stark ermine how many hours are you planning on getting on M31?

[stark ermine]

More than 160

[split smelt]

it's clear tonight here in UK, and I might even get 1st light on my new scope. Broadband only though. My new narrowband filter hasn't arrived yet, neither has the focuser 😦

[wise creek]

More than 3 months later, this finally got merged 😄!

Make sure you read the updated doc, it has changed quite significantly

[split smelt]

\o/

[split smelt]

Is this the latest doc? https://siril.readthedocs.io/en/latest/Commands.html

wondering why this throws and error now ?seqapplyreg pp_images -drizzle -scale=1.0 -pixfrac=1.0 -kernel=square

oh, -scale is removed

[wise creek]

I'm afraid I may not have thouroughly tested all commands. The parsing of this one is quite a challenge

can you tell m what error does it throw?

-scale is actually not specific to drizzle anymore

[split smelt]

now I get this... log: Filter Pattern: RGGB log: Cannot use drizzle on non-bayer sensors, aborting. log: Error in line 79 ('seqapplyreg'): invalid arguments.

[wise creek]

ah ok, this is intented.

You can't use drizzle on already debayered sequences

(agreed the message is not clear 🤔)

Ah no ok, I see, I'll have to check

would you know how to apply a patch?

[split smelt]

the images are not debayered. I removed the -debayer option from the script. This method was working ok earlier before I updated

yeh I can patch

[mental geyser]

are you a siril dev

[wise creek]

yep

[mental geyser]

oh sick

[split smelt]

yeh -scale has to be removed. It throws an error even if I reposition it before the -drizzle param log: Unknown parameter -scale, aborting. log: Error in line 79 ('seqapplyreg'): invalid arguments.

[wise creek]

what does your script line look like?

when you reposition?

[split smelt]

it did look like this seqapplyreg pp_images -drizzle -scale 1.0 -pixfrac=1.0 -kernel=square. I respositioned like this seqapplyreg pp_images -scale 1.0 -drizzle -pixfrac=1.0 -kernel=square but it only works without -scale

[wise creek]

try:
seqapplyreg pp_images -scale=1.0 -drizzle -pixfrac=1.0 -kernel=square

[mental geyser]

do you know which drizzle settings to use in order to use DeepSNR effectively?

[wise creek]

Nope, I didn't write the drizzle code, just refactored the whole registration process to better integrate drizzle within the rest of the codebase. Maybe @stark ermine knows?

[split smelt]

I already did, see above

I implemented the patch as a logical OR instead of an AND. Seems to work if (pattern < BAYER_FILTER_MIN || pattern > BAYER_FILTER_MAX)

[stark ermine]

I don't use DeepSNR... difficult to say 🙂

[wise creek]

oops 😄

but does the scale work?

because it works on my end

it has to be written as -scale=1. (with = sign, not space)

[mental geyser]

whats the closest thing to "varshape 1.5"?

kernel

its what pixinsight has

[stark ermine]

probably gaussian

varshape is like a moffat I believe

[mental geyser]

are there plans to introduce somethig like that?

a varshape fuction

[split smelt]

trying again...

[stark ermine]

no

I don't think it makes any real difference.

[mental geyser]

hmm alright, ill give gaussian a try

[split smelt]

Ok, this worked seqapplyreg pp_images -scale=1.0 -drizzle -pixfrac=1.0 -kernel=square

[wise creek]

great thanks!

[split smelt]

I'm liking this tilt and distortion analysis. Clearly showing I have an issue on my new scope 😮

[ancient talon]

kernel shape doesn't really matter for deepsnr

[mental geyser]

ive heard deepsnr doesnt like square function

[wise creek]

but now you can undistort during registration 🙂 !

mind you, your distortion is not that bad...

have a look at the test image in the doc: https://siril.readthedocs.io/en/latest/astrometry/platesolving.html#image-with-distortion-overlay

[split smelt]

-disto=image ?

[wise creek]

it would depend if the image is platesolved

[stark ermine]

Please Cecile, don't call my images test images 😢

😄

[wise creek]

didn't remember they were yours 🙂

[stark ermine]

You can even produce a master distortion

I use it

[split smelt]

I just tried seqplatesolve pp_images -disto=../MASTER/distortion.fit

log: Image #  228 - RA: 23.443 - DEC:+30.644 - Rotation: -89.4
log: Image #  229 - RA: 23.448 - DEC:+30.645 - Rotation: -89.4
log: Image #  230 - RA: 23.448 - DEC:+30.645 - Rotation: -89.4
log: Image #  231 - RA: 23.445 - DEC:+30.644 - Rotation: -89.4
log: Image #  232 - RA: 23.445 - DEC:+30.644 - Rotation: -89.4
log: Image #  233 - RA: 23.445 - DEC:+30.644 - Rotation: -89.4
log: Image #  234 - RA: 23.445 - DEC:+30.644 - Rotation: -89.4
log: Image #  235 - RA: 23.437 - DEC:+30.647 - Rotation: -89.4
log: Image #  236 - RA: 23.437 - DEC:+30.647 - Rotation: -89.4
log: Image #  237 - RA: 23.441 - DEC:+30.645 - Rotation: -89.4
log: Image #  238 - RA: 23.440 - DEC:+30.645 - Rotation: -89.4
log: Image #  239 - RA: 23.443 - DEC:+30.644 - Rotation: -89.4
log: Image #  240 - RA: 23.444 - DEC:+30.645 - Rotation: -89.4
log: No image selected after computing transformations, aborting(NULL)

the distortion.fit didn't get saved

then of course the registration failed ```log: FITS error: failed to find or open the following file: (ffopen)
log: FITS error: ../MASTER/distortion.fit
log: Could not load FITS image for distortion
log: Error in line 80 ('register'): invalid arguments.

[stark ermine]

the master distortion is created by a platesolve if the entry is filled

It is a wcs file

not a fit one.

[stark ermine]

https://siril.readthedocs.io/en/latest/astrometry/platesolving.html#distortions

[split smelt]

in my case, it's an issue with the code for "// find sequence cog and closest image to use as reference" in astrometric.c

int refindex = seq->reference_image; returns -1 which perhaps means there is no reference image?

because of this, none of the images get selected in the subsequent loop

[ancient talon]

I have used square for pretty much all of my images and I cannot tell the difference between varshape and square. Drop shrink factor is way more important

I can run some tests in a minute and see if deepsnr "prefers" one over the other

[wise creek]

I'll have a look tonight

[raven lodge]

It’s cool, but it doesn’t seem to accurately match the distortion you can see in the image exactly (stars bottom right corner)

Might be tough to get accurate using fwhm estimation. Tilted stars themselves might have a weird fwhm

I have horrible tilt right now so I’m going to take a look after work

[mental geyser]

That would be preferable. I don't have any data on my laptop since I've moved them to my hdd

So I can't experiment it

[split smelt]

This is working now with your latest commits

[split smelt]

@stark ermine if I turn on "Show Tilt", "Show Distortion" and then save a snapshot, the tilt shows on the snapshot but not the distortions. Is that intended?

[stark ermine]

Hello. No, probably I forgot something 🙂

[sage glade]

Hey uhm @stark ermine can you help on how I can stack comet images. This is my first time and I don't know what to do. I pressed the OSC preprocessing but the nucleus was weird so I don't know how to fix it 😦

[stark ermine]

fixed

[stark ermine]

We have a tutorial on our website

[sage glade]

Oh ok I will check it tomorrow morning thanks @stark ermine 😁

[mental geyser]

nico (nebula photos) has a tutorial on comet processing in siril. highly reccomend watching

much less tedious than dss

[stark ermine]

We are delighted to announce that we have just published an article in the Journal of Open Source Software (JOSS): https://joss.theoj.org/papers/10.21105/joss.07242
If, as a scientist, you use Siril, please cite it 🙂

[junior hound]

I wonder how it's going

[sage dove]

Does Siril have comet alignment?

[sage dove]

Good to know

[stark ermine]

I tried it. For now deconvolution only works on non stellar object

[junior hound]

I works nicely too, I already tried it with the little comet 1 or 2 summer back and it handled it pretty well even with my untracked setup

[junior hound]

So it sharpens nebulas and galaxies but don't correct stars /create artefacts?

Also can you share how you tried it?

[trim steppe]

is there a way to do batch starnet in siril? asking for future reference for comets

[broken peak]

there's apply to sequence in the starnet tab so yeah

[trim steppe]

oh damn

[sage dove]

Ok I may have to switch to Siril

[mental geyser]

[mental geyser]

so this is interesting. when i select to process only the green channel, the whole image goes green

i havent done anything to the image

and when i move the slider, it doesnt really do anything

this is version 1.3.3

ok so it seems to do that on pngs but when i use a fit file its fine

[mental geyser]

so theres an issue with 1.3.4

ive downloaded the new copy and its crashing when opened

quick screenshot

@stark ermine any idea?

[stark ermine]

I'm off this weekend. Will have a look later

[split smelt]

if it helps, I just did a git pull & recompile on my windows build and it loads fine

[stark ermine]

But it crashes at startup?

[mental geyser]

yep

[stark ermine]

Ok

[mental geyser]

this is in the windows command prompt thingy that normally opens up before the application does

[split smelt]

did you build it yourself using Msys2?

[mental geyser]

no

this is a fresh download

[stark ermine]

did you run siril-wrapper.cmd ?

[wise creek]

a fresh download of which artifact?

[mental geyser]

1.3.4

[mental geyser]

No...what's that?

[wise creek]

I mean the cross-build or the native build?

[mental geyser]

Native build

[stark ermine]

Which one? Because we do not often run the CI of the native build

[mental geyser]

https://gitlab.com/free-astro/siril/-/commit/9a39cda57b920a61356d20daca33fe7b99ef543b

i downloaded the "win64-native-installer:archive"

hope its useful enough

[mental geyser]

was this stuff helpful?

[wise creek]

ok, thanks

[wise creek]

it was the sunday morning automatic run

[stark ermine]

yep ok. Looks like a windows bug. I have no issue on my side

[mental geyser]

interesting

yea just reopened it and same thing happens

[stark ermine]

do you already compile Siril? Would be nice to try with it

[mental geyser]

what do you mean by that

[stark ermine]

Compiling sources is generally better to reproduce bugs and fix them 🙂

[mental geyser]

oh no i havent

[stark ermine]

The problem is that I don't use Windows. On my computer it works. The only dev in Siril using Windows is Cecile. But she works on a lot of features!! 🙂

@split smelt : do you use windows? Could you reproduce the bug and give us a gdb backtrace? 😄

I could fix it tonight with a backtrace

[split smelt]

I built from source using msys2 and it works fine for me on windows 10

[stark ermine]

even on last master?

[split smelt]

I tried it last night. I shall try again this morning and let you know

[split smelt]

This is the one I have running fine

[stark ermine]

Ok. Looks like either a binary issue or a user issue 😄

[mental geyser]

no idea what ive done wrong lol

ill turn off my laptop for the night and then ill retry opening siril in the morning

prolly windows being a donkey

[split smelt]

I used to have all sorts of issues running the snapshot builds. It's no big deal to build your own and they work great unless the source code has a bug in it

[mental geyser]

i wouldnt mind building it but the problem is....i know jackshit about coding

[split smelt]

all you have to do is follow this guide to the letter. No coding skills needed https://siril.readthedocs.io/en/latest/installation/windows.html#building-on-windows-with-msys2

literally copy/paste

[mental geyser]

oh

[split smelt]

but, make sure you actually read it & do exactly as it says

[wise creek]

It does work on my PC as well and I obviously do build

I'll need to check in the CI log if smthg went bad with the weekly build

[stark ermine]

did you tried the artifact too?

[wise creek]

yep, just now

and it crashes unless I invoke siril.exe from a terminal, in which case it succeeds....

weird

[stark ermine]

wow... yes, indeed

for me this is linked with the graXpert init

It's my intuition

[wise creek]

do you know if graxpert init tries to cd somewhere?

[stark ermine]

I don't know no

[wise creek]

because from this: #1220109001682059275 message
it seems like it's trying to load pixbuf dll from the "usual" location that is C:\Program Files\Siril etc etc

whereas when starting the portable version, this folder does not necessarily exists or it doesn't contain the correct version of the dll

that we linked against when building

[stark ermine]

yes I see. The path is not good

What kind of change has been done here...

We could create a branch removing GraXpert init, just to be sure

[wise creek]

I'll have a look in the code first

The CI will take at lest an hour to run

[stark ermine]

yes.... 🙂

[wise creek]

I think there has been a breaking change in this package:

https://packages.msys2.org/packages/mingw-w64-x86_64-gdk-pixbuf2

I can't find libpixbufloader-svg.dll in the file list

[stark ermine]

@wise creek : an issue just opened on this subject

[wise creek]

where?

[stark ermine]

#1401

[wise creek]

which site? msys2 git?

[stark ermine]

Oh no. Our gitlab. Sorry

I mean the crash is now reported

Someone just did it

[wise creek]

ah ok, we need to add an optional dependency: mingw-w64-x86_64-librsvg

I'll do it in a branch and start the native CI

[stark ermine]

Good

[dim minnow]

i want to ask something i got 600 + image of nebula and i wanted to stack them at siril

but i got no space left

so can i stack them 200 images first and convert them to 3 fit files and then stack them together again?

so 1 fit files 200 images

is it different than stacking the raw files directly? i mean the aspect of the result

[stark ermine]

Yes it is

THis is not a good idea

[dim minnow]

why

[wise creek]

if you run into the same issue after updating msys2/mingw64
run:
/mingw64/bin/gdk-pixbuf-query-loaders.exe --update-cache
in mingw64 console.
the loader dll handling svg has been recently updated to a new name in mingw-w64-x86_64-librsvg package

[stark ermine]

Because statistics cannot be split like that. You need to run the whole sequence to get consistent statistics. For normalisation, pixel rejection,...

[mental geyser]

its worth the lik $60 to get an ssd for stacking

[dim minnow]

But siril is smart right...?

[stark ermine]

Well, you can't change math

[ancient talon]

It will give you an alright result. It will be slightly worse than stacking all your subs in one go though

[mental geyser]

redownloaded 1.3.3 distribtion version and booted up fine

[wise creek]

Should be fixed tomorrow.The time for the CIs to build the new packages

[wise creek]

This artifact should work: https://gitlab.com/free-astro/siril/-/jobs/8158257695

[mental geyser]

will give it a shot

[chilly marten]

Can the seti Astro sharpening tool be build in to siril like the graxpert?

[trim steppe]

i think it had a cli soo probably?

[stark ermine]

We will not add every 3rd party astro in Siril. Because it would be too much work at every release of another software

[junior hound]

Additionally to lock's answer, graxpert is soon realeasing a deconvolution ai so it will be much easier to implement

[chilly marten]

Is there a big difference between these ai deconvolution software and the deconvolution option already in siril?

[junior hound]

Yes quite a big one actually!
First off, part of the ai is to decide where to apply the deconvolution (masking basically) to keep the noise level low. Siril's decon will often add significant noise to the bkg for example
Ai will also decide the most fitting psf for deconvolution which can be quite hard with siril, especially for non trivial psf (with aberrations, atmospheric blur...)
And finally with those better psf's and ai magic it can correct aberrated stars much more efficiently than with standard decon (especially when aberrations are not constant across the frame)

[trim steppe]

thats already a thing?

[stark ermine]

working on it

[trim steppe]

im guessing they havent released any actual models for it yet

[stark ermine]

not yet, but I have it

[trim steppe]

any results? 👉 👈

[stark ermine]

this is a first step. Working only on object, not stars. Stars is for the next iteration

[junior hound]

Oh niiice

Can't wait to try it out

[stark ermine]

And for your information @junior hound , Siril deconv was used in the training 😉

[junior hound]

Oh nice

Now I realise it sounded like I was critizing siril's decon in my earlier message but it just "naive" deconvolution vs ai decon, siril's decon is actually nice!

[stark ermine]

I don't take it as a criticism 😉

[ancient talon]

What's "Blur PSF size"? Does it tell the NN how "blurry" the image is?

[mental geyser]

downloaded the latest distribution copy. 1.3.4 boots up fine on my laptop

[split smelt]

tbh I've never had much luck with Siril deconvolution either. It's probably very capable when used properly, but I've never managed to work out how

[candid kiln]

What about adding Cosmic Clarity into Siril?

[waxen ember]

what about stable diffusion

[sage glade]

@candid kiln

[stark ermine]

#1220109001682059275 message

[junior hound]

Btw do you have any info on when the graxpert decon model will come out @stark ermine ?

[stark ermine]

Soon

[mental geyser]

👀

[mental geyser]

is the decon version released or not yet?

[stark ermine]

nope. GraXpert team has not released yet. Anticipation is our watchword 🙂

[spring oxide]

which branch should be built to get the GraxPert integration. Is it siril-dev?

[dense heron]

i need the dev version

i try compiling and it keeps saying im missing some header

unfortunately i have no knowledge on compiling programs. i just copy pasted stuff from the website

tries to go to something, that thing doesnt exist and then just terminates

[spring oxide]

could you post the error? I'm getting an error too and I'm curious if its the same

[dense heron]

i have -100% knowledge in this stuff. it seems it tries to look for something and cant find it

people have similar problems but i dont understand any of the code. they use Cmake or something. but in the website it says to use meson setup _build --buildtype release

[spring oxide]

nevermind, I was getting somehthing else

[dense heron]

guh

[mental geyser]

omg its andy and andy

[dense heron]

strange seeing another andy. idk if i met another andy irl

my name is different on other serves lmao

[dense heron]

i got the dev version

https://gitlab.com/free-astro/siril/-/pipelines?page=2&scope=all

i downloaded it here instead of compiling

look for one with all 6 check marks. it says something like weekly builds for devs

on the right it it had a downlaod thingy

download the natvive win 64 installer archive

[mental geyser]

make sure you downlaod the dev version instead of stable otherwise youll go back to version 1.2.4

[split smelt]

did you run "Installing Dependencies" on this page? https://siril.readthedocs.io/en/latest/installation/windows.html

[stark ermine]

make sure that your opencv package is updated

[dense heron]

its strange, i have the package, its on my pc, i can even find the directory to it but still nothing. some people do something like #include or something but i have no knowledge in these things but i got the dev working anyway

[dense heron]

i think i figured out manual stacking with siril

wondering if there are other tutorials that can help ? theres one by deep space astro but i dont trust him lowkey

nvm the siril website has one

[split smelt]

I always use Sirilic for stacking. I find it easier for multi-night stacking than native Siril and not as restrictive as using pre-prepared scripts

[stark ermine]

Why don't you have a look to the tutorial from website?

[spring oxide]

I love the tutorials the website provides. Used them the first few times I tried processing with siril and it was easier for me to follow than video tutorials

[dense heron]

idk if i did this right. the colors look good

is there an option for nebulae? i was told to leave it as the defautl option of average spiral galaxy

[stark ermine]

If you're ok with some read, I wrote some stuffs about this feature 🙂

https://siril.readthedocs.io/en/latest/processing/color-calibration/spcc.html#spectrophotometric-color-calibration

[dense heron]

this is amazing

more stuff to learn

[merry perch]

A good read indeed. When it talks about choosing the white point reference being set to average spiral galaxy by default, is this option also suitable for nebulas or just galaxies?

[stark ermine]

for everything

[tardy lily]

Hello, where would i be able to download the de version of siril

[mental geyser]

sail the 7 seas

[hard cypress]

Could you put a feature to let users open each individual channel within its own mini-window in the program, instead of having to open multiple instances of the program please? 🙂

[stark ermine]

no. It is not scheduled

And each channel of a RGB image can be seen in one tab

[hard cypress]

Fair, thank you for the response.

I know that each channel of an RGB can be seen in its own tab; I was hoping for an "all individual channels at a glance" solution.

Again, thanks. 👍

[mental geyser]

Graxpert deconvolution seems to be out now

[stark ermine]

yes

And siril can use it

[trim steppe]

[broken peak]

siril seems to get stuck on this, it doesn't launch

nvm it launched after like 3m

[stark ermine]

network issue?

[split smelt]

ooo this is nice "Improve pixel math compatibility"

[split smelt]

will "show distortions" show backfocus issues? If so, will it differentiate between too close and too far?

[stark ermine]

It only shows distortions in the image 🙂

It is used during stacking, we can now undistord the stacking result

[mental geyser]

im using the most recent distribution copy for the week and star recombination is making siril crash

when i go into the tab and load a stretched starless image, siril entirely crashes

[stark ermine]

then this is not the most recent

I've fixed it 5 days ago

[mental geyser]

hm interesting

ill try redownloading it

[stark ermine]

most recent has 10hrs

[trim steppe]

quick question @stark ermine, are subs that just got hit with drizzle (square, scale 1.0, dropshrink or whatever its called 0.8) supposed to look like this?

note that these were calibrated with bias and flats, files kept undebayered until drizzle step

[stark ermine]

Yes it is normal

[trim steppe]

stack of it came out quite goofy as well, are there any specific settings i should use when stacking the files that were drizzled?

[stark ermine]

What is your results.

How many subs

[trim steppe]

55

[stark ermine]

What scale. What pixel fraction

[trim steppe]

may or may not have deleted it just before, currently doing new stack though

scale 1.0 pixel fraction 0.8

harddrive making it omegaslow

[stark ermine]

Should be 1 and 1 for CFA drizzle

[trim steppe]

thats what it was with the last stack

still had that pattern as in the screenshot above, dbe just made it purple

[trim steppe]

well, this time it came out like this, had normalization disabled and rgb channel equalization enabled

[stark ermine]

Looks like the CFA pattern is not good

[trim steppe]

it chose RGGB which should be correct

[stark ermine]

You choose manually?

It is always better to let siril choose

[trim steppe]

nope, let it choose automatically

[stark ermine]

So, to be sure try another.

[trim steppe]

or not, had this checkbox not ticked, does that affect it?

[stark ermine]

Should be ticked

[trim steppe]

do i do the registration again?

took 11 seconds on ssd, holy shit

[stark ermine]

Haha

[trim steppe]

alright, had that option ticked and it came out like this now, looks more correct

oh yeah, thats the stuff

[stark ermine]

Yes

Always trust the software:)

[trim steppe]

yeaah, i dont remember why i turned that automatic thing off, ty for help though

[spring oxide]

In standalone GraxPert, when using background extraction, it saves the points that are placed. When done through Siril, the points are removed after BE is done. Would it be possible to keep the sample points as long as the GraxPert window is open? Would make it easier to fine tune

[broken peak]

With the standard background extraction if you don't click apply the points will remain, you can change them and press compute background to see if anything improved

[stark ermine]

Yes to fine tune, just don't apply.

[plain birch]

hey all i was told i could reach out to some siril devs here, im having issues with my starnet++ integration

[tame surge]

Give more context

[plain birch]

turns my red picture blue when removing the stars

one sec im taking ss

Before -> After

ive handed this exact stack fits file to someone else, and his did not turn blue, im uploading that in a second

[tame surge]

I’ll let the others handle this, I have never had this issue sorry.

[tame surge]

Hmm, maybe different settings?

[trim steppe]

(i used different software)

[plain birch]

you said you used starnet

[trim steppe]

starnet isnt exclusive to siril

[tame surge]

Starnet but not in siril prob which might make a difference

[plain birch]

well yeah but its suppose to be the same process no?

key word suppose

[trim steppe]

its most likely not a starnet issue

[plain birch]

what would it be then

[trim steppe]

something in siril? idk

[tame surge]

I would attempt the data to see if I could if I had my laptop

[plain birch]

oh you mean like the actual command and integration part?

in siril

[tame surge]

What settings are u using

[plain birch]

cause it isnt stretched

this is linear

[plain birch]

^

[tame surge]

Really have no idea then sorry

Wait for lock042 to say something

[stark ermine]

Siril just call starnet. The CLI version of starnet. There is no process in Siril. So that must be from StarNet

[mental geyser]

What were you guys expecting him to do

[plain birch]

Would it be better to run it in a stand-alone rather than the integrated starnet?

[mental geyser]

Its the same program

[stark ermine]

that change nothing.

As I said, Siril just call the program. It is then more conveniant because Your image stays in the same program

I don't remember, but probably, on Windows, you can see the command line used to call StarNet

in the black dialog opened with Siril

no. I'm wrong. We do not write it. I probably will add it

Yes, logs will be added in the next version. Just did it. Will help you to reprocude command line to the standalone programm

[mental geyser]

will the mosaic thing in siril be similar to photometric mosaic in pixinsight

[stark ermine]

I don't know the photometric mosaic in PI

[stark ermine]

I believe this script was dropped by the PI team

[mental geyser]

Yea its a downloadable script

It uses a repository link

[stark ermine]

but no, we don't do like this

[stark ermine]

https://youtu.be/mx-lU75Ui3g

[trim steppe]

wowie

[safe shale]

Hi

I don't know if it has already been done or if it's even possible but

A way of "increasing" sub exposure time for stacking untracked astrophotography:

It would divide the light data into batches;
Stack those batches by adding pixels (instead of averaging them);
Then stack those sub stacks into final stack by averaging pixels

There could be an UI for exposure multiply factor

For example:
Let's say you have 600 light frames

It would divide them into, let's say, 100 batches;
So there would be 100 batches of 6 light frames

It would then stack them with sum, resulting into 100 stacked images;

And finally stack them like usual

That could increase the exposure by 6x?

[mental geyser]

why would you want to add the pixels

theres a reason why we average pixels and its so that rejection algorithms can remove outlier pixels

the only way to increase your sub exposure time for untracked astro is buying a mount.

[safe shale]

Adding the pixels in sub stacks, not the whole image. Like I said, it could "increase the exposure" of the final result.

[spring oxide]

you're "increasing the exposure" but also the noise. might as well just stretch it further

[ancient talon]

Addition and averaging works out to the same final SNR

And you can use rejection in both. The outlier pixels get rejected before they're integrated

There is no point in summing the images in a single batch though, averaging will yield the same results but it won't clip data (unless you'd normalize the integrated batch to [0-1.0] after adding them together in which case you might as well just use averaging)

[safe shale]

Oh ok

[stark ermine]

https://siril.org/download/2024-11-22-siril-1.2.5/

[stark ermine]

There is an issue with the Windows package of the new version. We are rebuilding it. Stay tuned!!

[stark ermine]

It is ok, let's download now 🙂

[mental geyser]

is it just me or is the gaia thing acc down

ok it works nbm

the status is yellow

[stark ermine]

yes it works but there is a warning

[merry perch]

@stark ermine I heard that you are working on making mosaics easier for Siril 1.4, so I think I should ask to understand what that actually means. I suppose it is for traditional panel stitching? Cause as a Seestar user, the new mosaic mode was released but it doesn't shoot and organize in panels, but rather it rotates around the enstablished framing and saves all frames unorganized. Pix users can stack the subs by using the live stack as reference frame, but by using a script that enlarges it with the Gaia catalogue and thats how it works best, cause the outer frames would probably be rejected due to not enough overlapping stars. I found that in Siril these can be stacked (both with live stack as reference, both without) by using the maximum framing option, however many frames get rejected, so with like a 60% success rate (eg: over 4500 frames I got 3000 succesfully stacked. And usually those rejected frames are around the edges, so those parts get less signal.

I use 2 pass or global star alignment, 4 minimum pairs with homography and I even tried Similarity with less required star pairs, and I even lowered the star detection treshold to get more, but it barely makes any difference, again the same thing happens even if I put the mosaic live stack as reference. So is this just how it works or is there a batter way I'm not aware of?

[stark ermine]

The mosaic mode is not yet released. So this is normal you can't use it.

[merry perch]

Will 1.4 make this type of mosaic stacking easier or is it for normal panel stacking?

Yeah the way I'm using it right now isn't how Siril its supposed to be used for at least for now I know

[raven lodge]

He posted a video about it above #1220109001682059275 message

I tried in a build a few days ago and couldn’t get it to work, unfortunately. I don’t think it was integrated into the main dev branch at that point.

I got the same behaviour as you did, the frames that are too far away from the reference point get rejected by the batch astrometry process, even though I can solve those frames individually. So we have to wait a bit.

[merry perch]

This looks great. Would this only work with a few organized panels or should this also be suitable for many unorganized frames?

[stark ermine]

The mosaic feature is on the main branch dev. Not the seamless

You need to plate solve the sequence first and then register in maximum and stack

[raven lodge]

I'm using 1.3.4 commit a557140 from November 20th. If that has mosaic mode then I'm seeing about 10% failure rate platesolving a sequence of seestar images. I've tried lots of different settings but these are giving me the best success rate. I can't see any obvious reason why those images fail to platesolve, they're not trailed or have a large FWHM or background noise value. I can run a plate solve on the individual frame and it works, but if I do that then they don't seem to be added to the registration.

[stark ermine]

If it works for individual it should work for sequence

Share your files

[raven lodge]

Here you go. https://1drv.ms/u/s!ArEUZ19_cutJmcc3S8vAr-ypEm1dng?e=iLZq2O

[stark ermine]

did you debayer image or use drizzle?
In other word, do you register on CFA image or on color image

[raven lodge]

No, I’m simply adding them all to a sequence without debayering and immediately trying to plate solve. Ideally I would use cfa drizzle later in the process.

[merry perch]

@stark ermine I'm also trying Seestar mosaic subs with the astrometic alignment, not on the latest version, but 1.3.2, if thats the issue then I'm sorry. So my problem is that the sequence will stop platesolving at image 40. First I accidentally left debayer on, so i tried again with no debayer but same problem. Here's the log from the no debayer sequence (I think it froze so it might not contain the error message), and the debayered sequence (second file) that I'm sure contains the error message. Specifically this part: 10:08:58: FITS error: Failed to move to HDU number 41 (ffmahd).
10:08:58: Could not load frame 40 from FITS sequence testSEQW

[stark ermine]

1.3.2 is old. When using dev version you need to update everytime

We cannot analyze bug report from an old dev version

Keep in mind that dev version is in active development. Mosaic is not finished

There is a lot of things to do

[merry perch]

Yeah I love seeing so much effort on free software, I can't stress that enough, thanks for the hard work.

[mental geyser]

3.2 is like September/October old

[stark ermine]

I'd also like to add that your screenshot shows that you use online catalogs. This is a pretty bad idea for astrometry on the whole sequence. It takes too much time. It's advisable to install local catalogs.

[broken peak]

You can use astap for that?

[stark ermine]

I don't know. However, our new astrometry tool will be more powerful

[wise creek]

There's one case where it may be useful, is if the seestar writes the same ra/dec for all the images of the mosaic sequence. ASTAP near search architecture is more efficient then our own when the whole mosaic field starts to be large. Plus, Siril nearsearch can only be used if the user has the local catalogues installed (otherwise, the catalog fetch would be too long).
That said, yes, you can platesolve the whole sequence with ASTAP. It should update each fits header with the astrometric solution. Then:

• load the sequence in Siril
• open the astrometric solver (Ctrl+Shift+A)
• at the bottom of the window, if not already ticked by default, tick the Solve Whole Sequence box
• make sure the Use as registration information and Skip already solved images are ticked.
  As said, siril will skip platesolving the images (because ASTAP has already done that and updated the headers), but will load their astrometric solution to compute the astrometric registration for the sequence.
  And then, Registration->Apply exxisting registration. Use the maximum framing and there you go

That's a possible workaround. Anyway, the Seestar mosaic mode mode is fairly new, so we did not have this use case in mind when we refactored the whole astrometry/registration part of Siril. We will see when we release if we have to adapt

[raven lodge]

I tried all sorts of combinations of those flags as well as the downscaling one.

[raven lodge]

I believe the Seestar writes a different ra/dec for each frame. How accurate that is is a different matter 🙂

Hmm. It might be that those images I provided are not raw from the Seestar but have had their headers already updated by Siril. I will try it fresh tonight.

[merry perch]

I tried I can't really get it to work

I used astrometric alignment and then as stacking method: image stitching mode

But 99% its me not doing something right

[wise creek]

you can use command seqheader to export RA and DEC key values to a csv on a whole sequence

[merry perch]

Updated to 1.3.4 commit cb69eb2f4. Now platesolving. The subs were already ran through ASTAP a few times without succesfully stacking some times. Plate solving currently, set to skip platesolved frames but it freezes for long on some of them. Am I doing it right?

[wise creek]

Seems some were not platesolved by ASTAP right?

[merry perch]

Maybe

[wise creek]

well, in the log I see some with already platesolved, skipping and others with platesolved and updated

so on the skipped ones, Siril will go blazing fast (it just reads their headers), on the other ones, it will spend time trying to solve them

[merry perch]

So ASTAP just saves it onto the header every time it solves, correct? No need to Grant permissions or save files?

[wise creek]

Still, to perform this in Siril, you should try using the local catalogues

[wise creek]

Seems so, otherwise, you wouldn't have some already platesolved, skipping

Well, that's my guess at least

but maybe you should check what ASTAP does yes

I've not sequence-platesolved in ASTAP in years

[merry perch]

Well its solving thats what matters tbh

[wise creek]

I would have thought it would be able to solve them all 🤔

ASTAP I mean

[merry perch]

Some images are failing but most are getting solved

Probably some trashy frames

This is actually my first time building the source code via MSYS2, I had the previous 1.3.2 as an installer and every time I wanted to switch between stable and dev I uninstalled the other. But by building it manually could I keep both stable and dev with no issue?

[wise creek]

You should use the installer only for stable

and then build dev from msys2

then stable is in C:\Program Files\Siril\bin

and dev is in C:\msys2\mingw64\bin

[merry perch]

Ok thats good

after plate solving this is all the registration types I have

[wise creek]

registration from astrometry has been computed when platesolving the sequence

Just pick apply existing registration and framing=max

[merry perch]

oops

[wise creek]

Ah ok, that's why it was so slow...

[merry perch]

So it means I must stack the frames in batches?

[wise creek]

how many frames have you got?

[merry perch]

2890

[wise creek]

yep, two batches then

[merry perch]

So I gotta redo the sequence again but using normal FITS

Ok good

Should be faster though?

[wise creek]

yes, it's the constant rewriting of the fitseq which was taking long

we will disable that because it's a pain

[merry perch]

Is it better if I use interpolation or drizzle?

[wise creek]

with that many images, you can probably use drizzle

anyway, at the registration step, you won't have much choice

if the sequence has been debayered, registration will only allow for interpolation

If it's still undebayered, it will only allow for drizzle

[merry perch]

Oh ok

[merry perch]

Ok, applied registration, tried both interpolation and drizzle, but it ends up crashing and I have the Segmentation fault error in the terminal.
The logs give no error but I still tried to print them out from the terminal

With interpolation though it crashes almost at the start of the registration applying process, while with drizzle it goes on for a bit more time but always ends up crashing

[wise creek]

when you've built in msys2, did you build in release or debug mode?

(just to see if I can get you to print out a back trace when it segfaults 🙂 )

[merry perch]

I followed the instructions here (https://siril.readthedocs.io/en/latest/installation/windows.html#building-on-windows-with-msys2) so its release

[wise creek]

ok, can I tell you how to build debug and run gdb then? I will be slower but I should get a lot more info

[merry perch]

Should I just follow the same process but write debug instead of release at that precise line? Do I have to delete the files or do I just redo the installation again

[wise creek]

meson setup _build --buildtype debug --reconfigure

you need to cd to siril first

sorry it's debug yes

in the first meson command

[merry perch]

in my case it created one siril directory and then anothe siril inside it. the first one?

[wise creek]

uhhh

I would say the one inside the first one

when you are right one and type ls

you should get stuff like this

AUTHORS            ChangeLog        README.md    build         devdoc   meson.build        paper          po                 siril.1         subprojects
CITATION.cff       LICENSE.md       _build       configure.ac  dlltest  meson_options.txt  pixmaps        python_module      siril.doxyfile

[merry perch]

didnt read and Im doing it in the first, it seems to be working though

[merry perch]

yes exactly

[wise creek]

ok great then

[stark ermine]

Sometime the reconfigure is not enough for the debug symbol and I need to remove _build

[wise creek]

yeah, that's because you are on Linux 😄

[merry perch]

ok now I start it again?

[wise creek]

now yu type

gdb siril

and when gdb has started, you type r

[merry perch]

ok its booted up

[wise creek]

it will launch siril

take the debayered sequence and register it with interpolation

(the code is simpler than drizzle and I know it better)

Instead of crashing, it should say smthg in the msys2 console

[merry perch]

ok it froze

[wise creek]

print disto->xmap```

I suspect it will say NULL

[merry perch]

says this $1 = (float *) 0x0

[wise creek]

yep ok

thx, I've had this one in the past but I thought it was gone

I'll have to keep searching

to exit gdb, type q

you can now go back to release

meson setup _build --buildtype release --reconfigure

and the ninja line

[merry perch]

perfect thanks

[wise creek]

thanks very much for your testing and following along all my mumbo jumbo 🙂

really helpful

[merry perch]

Sorry if the question sounds stupid, but in order to install the latest dev build I use this? Always in the Siril directory?

in the future when it comes out now I'll reinstall release

[wise creek]

no it's just the first time

then every time you want to update

launch msys2/mingw64