Various questions on Siril & GIMP

[Robculm]

Hi all, Since my previous post getting started with a smartphone (through a 200PDS with HEQ5 pro), I've made several 'upgrades' & spent a lot of time 'trying' to improve processing skills etc, but have several questions on which would very much appreciate any advice / suggestions.

Firstly I've progressed to a DSLR (after much deliberation I went with a Canon EOS800d) & so far very happy :-). Have also just completed the Rowan belt upgrade as I was getting elongated stars on quite a lot of frames. Have yet to try it out, what with the weather as it's been!

I will attach the latest endevour (M101) (pre-belt mod, but having deleted the bad frames!).

First question is, I'd like to repeat this image with an OHC filter, to try & pick up some of the nebulosity in the galaxy. I've found how to add separate image in another layer (in GIMP), but how about alignement? Is there any way of doing this automatically, or have to manually align? Since I would always need to remove the camera in this scenario (to fit the filter), I assume I would need to fully stack / pre-process the two sets separately & then 'merge' them together in GIMP. But the alignment is what's worrying me...

2nd question is really about processing flow. So thanks to previous advice from HTH, I'm using Siril to stack (lights, darks, biases, flats), I then get a bit lost as to what aspects it's best to edit in Siril & if it matters which order you do them in before exporting to GIMP. I'm typically doing photometric colour calibration, green noise remove, autostretch & background remove (which sometimes works well & sometimes doesn't, depending on the particular image!). Then export (was doing in 32bit floating point, but having now just added the pyastro plugins, I find I can only use them if I export as a 16bit!?).

In Gimp, I haven't played much with pyastro yet, only just added, but typically I'm fighting with more background removal (gradients, fill with background colour etc etc (I find all the blur & despeckle techniques often mentioned don't work for me, always end up removing lots of data from the image), to be honest I've found that most attempts to get rid of the noise result in loss if image data!) & maybe some saturation & the basic contrast etc. Actually, this particular M101 image didn't need much processing at all in GIMP, so I'm thinking the key is to getting more imaging time (signal) & hopefully minimise noise to start with! I'm using ISO400 which seems good for the 800d & 3min exposures on this one. Previous targets were shorter exposure (various ISO's) & I don't think I was getting enough signal over the noise...

Anyway, so really just some advice please on the best process flow with these two software packages & any other suggestions!

Many thanks,

Rob

 

Edited March 29, 2021 by Robculm

[The Lazy Astronomer]

I don't use SIRIL, but I assume it has a similar functionality to DSS, and if so, it should be possible to stack lights from multiple nights together provided the framing and camera orientation are not too different.

In DSS, this is done through use of groups to keep sets of lights and calibration frames associated with each other during stacking - does SIRIL have any feature like this?

Edit: great image by the way 👍

Edited March 28, 2021 by The Lazy Astronomer

[alacant]

22 hours ago, Robculm said:

about processing flow

Siril: use images from two nights, even if you've moved the camera.

One of many ways:

Pre process both nights, say folders n1, n2 but do not register and stack.

Create a new folder for the combination and set it as the new working folder.

Use 'Convert' to populate from the pp_ prefixed files in n1 and n2.

Select the new sequence.

Register

Stack

That's it.

HTH

Edited March 29, 2021 by alacant

[rickwayne]

In general, you want to do as much processing as possible while your data are linear, i.e. unstretched. So do everything you can in SiriL, then stretch and export as the last two steps.  It sounds as if you're doing it pretty right (as your image demonstrates -- nice one!).

One  of the reasons I sprang for Astro Pixel Processor is its world-class gradient reduction feature. That's something you really, really, really do want to accomplish with linear data if  at all possible. It might take a couple of goes, mind.

Eliminating noise digitally without losing detail or introducing artifacts (e.g. the "plastic" look) is a real challenge and no mistake. I have struggled to find free software that really does the job; again, I resorted to the wallet, and  picked up Topaz DeNoise AI. It's not magic, but it's pretty good. The noise-reduction tech in PixInsight is supposedly quite good, but that's getting practically into nosebleed territory, both financial outlay and the effort required to master it.

To reiterate: If you're doing this well early in your imaging career, you should be proud and delighted.

[bottletopburly]

You can also stack the finished stacked image from both nights together, Siri’ls developer told me that and worked very well tbh , don’t asked me how I did it though it was a while ago and I always scratch my head using Siril but it is very good at stacking .

[alacant]

4 hours ago, bottletopburly said:

don’t asked me how I did it

Method here.

Edited March 29, 2021 by alacant

[Robculm]

Hi All, many thanks for your comments & feedback, much appreciated.

I do find Siril really nice to work with, but looks like I need to understand it in more depth. So far I've just been putting the lights, darks, flats, biases in to one folder & linking that as the current working directory, then selecting the osc_preprocessing command, which runs & outputs the 'result'. I haven't needed to understand the conversion, sequence, pre-processing, registration tabs!

HTH, when you say pre-process both nights (n1 & n2) what does pre-process actually mean? Is it just to sort the lights, darks, flats, biases from each night in to separate folders (n1 & n2)? Put those two folders in to a single folder & select that as 'working'? To be honest, I'm very much a 'hardware' person! On the 'software' side, anything other than 'clicking this or that command' is somewhat challenging!!! 😞

Anyway, for now I'm still holding out for that n2! Hopefully in the next week or so now that the moon is getting later & the weather improving... But will try to experiment with some 'dummy files' in the meantime.

I am wondering though, is it not better to have two completely separate images so as to edit them differently in different layers? For example to edit the nebulosity separately? Or there's no way to align the two images unless you 'stack' them together?

Other quick question was about the file format. As I mentioned, I've been exporting from Siril as 32bit, but can only use the Pyastro plugins in GIMP if I export it as a 16bit. Anyone else seen that problem?

Thanks again,

Rob

 

[alacant]

 

8 hours ago, Robculm said:

what does pre-process actually mean?

It means 'get your images into a state where they can be stacked'. In your case this would consist of manipulating your light frames with the bias and flat frames; best to lose the dark frames with an 800d. It's what's done inside the OSC black box you mentioned.. To get the best out of Siril, I'd recommend learning how to use it manually before using the processing scripts;)

8 hours ago, Robculm said:

I haven't needed to understand the conversion, sequence, pre-processing, registration tabs

Now you do!

But seriously, to do anything other than basic processing (for which you may as well just use DSS), you need to understand what's going on in the scripts. The tabs are easy to understand and there are loadsa tutorials.

8 hours ago, Robculm said:

two completely separate images

You could try both methods. Use a sequence with just your two stacked images > register > stack

Yet another way would be to layer the two stacked images in GIMP using screen/lighten/WHY.

8 hours ago, Robculm said:

Pyastro plugins in GIMP if I export it as a 16bit.

Not a problem. Unless you've been using the HST, you'll see no difference. 16 bit output in Siril is fine.

if you like Siril, you'll most probably like StarTools too. Instead of stretching using the levels and curves of the 1990s, it uses mathematical modelling to achieve a cleaner result. Your image is held linearly in a database throughout, thus obviating the need to distinguish between said and non-linear versions of the same.

HTH

Edited March 29, 2021 by alacant

[prusling]

Take a look at the Sirilic front end to Siril - it simplifies preprocessing whilst supporting multiple sessions. 

[Robculm]

I've seen some references to Sirilic, will definitely explore, thanks.

HTH, ha, I chuckled at your politely suggesting I learn how to use the tools properly 😆. You're absolutely right, lack of computer skill is indeed no excuse for laziness! Will endeavour to get a better understanding of the processes. Siril is a great piece of software & seems to cover a significant amount of the whole process, so definitely worth the effort I think...

Regarding Startools, actually I did take a look, but almost ashamed to admit that I couldn't get it loaded up properly on my computer! 🙄 ('Hardware guy' as I may have mentioned!!!) I think it was something to do with folder permissions... 

Anyway, the moon is passing & hopefully some clear skies soon... Will try for the OHC filter imaging on M101 & then I'm keen to switch my attention to some globular clusters. That was always a favourite target for 'visual'. I've made an attempt with M3, next planning some 3 - 4am sessions with M4 & M5!

Thanks again,

Rob

[alacant]

1 hour ago, Robculm said:

3 - 4am

Also around the same time you have m13 and m92. Superb with your 200p:)

[AstroMuni]

On 28/03/2021 at 21:33, alacant said:

One of many ways:

Pre process both nights, say folders n1, n2 but do not register and stack.

I am new to imaging and SIRIL and I have heard that its better to stack the stacks. i.e First night's images stacked as s1 and then 2nd nights as s2. Now put these into a separate folder and stack them together s1+s2.  Why the different approaches and does one offer better advantages over the other?

[AstroMuni]

On 28/03/2021 at 12:35, Robculm said:

I will attach the latest endevour (M101) (pre-belt mod, but having deleted the bad frames!).

Lovely image. How many images and what is the duration of each exposure?

[alacant]

Hi

1 hour ago, AstroMuni said:

I have heard that its better to stack the stacks.

Could you please post the link/reference and the target example?

1 hour ago, AstroMuni said:

does one offer better advantages over the other

AP. The dark art indeed. 

Until you've tried the methods available, you don't know which will afford the most pleasing result. The advantage being that you have only your own criteria by which to judge:)

Cheers.

Edited March 30, 2021 by alacant

[Robculm]

Hi AstroMuni,

Thanks 🙂. This was with 24 x 180s exposures at ISO400 (started with 30 or so, but deleted out the ones with elongated stars) with x20 each flats, darks & biases.

I'm still very much experimenting with different exposures & ISO's etc, but from the handful of sessions to date & lots of single test shots at different settings, this worked out the best combination so far.

I'm thinking to try and push the exposure a little longer ( 4 or 5 minutes maybe) in future, now that I've got the Rowan belt mod sorted (assuming that works to fix the star elongation problem). I read that the 'mountain' in the histogram on the camera should be 1/4 - 1/3 in from the left, I've not managed to get any further than about an 1/8th. Although I'd imagine it depends somewhat on how 'bright' the particular target is?! 

Cheers,

Rob

Edited March 30, 2021 by Robculm

[AstroMuni]

18 hours ago, Robculm said:

I'm thinking to try and push the exposure a little longer ( 4 or 5 minutes maybe) in future, now that I've got the Rowan belt mod sorted (assuming that works to fix the star elongation problem).

What tool do you use to drive the mount? I use Ekos + EQMOD + NO guide scope and I have the same star elongation problem. Workaround for me at the moment is taking lots of short bursts (limited by star elongation), followed by a realign to bring image back in centre.

[AstroMuni]

22 hours ago, alacant said:

Could you please post the link/reference and the target example?

It was through friends on my local astronomy group, so no link

[Robculm]

1 hour ago, AstroMuni said:

What tool do you use to drive the mount? I use Ekos + EQMOD + NO guide scope and I have the same star elongation problem. Workaround for me at the moment is taking lots of short bursts (limited by star elongation), followed by a realign to bring image back in centre.

I'm driving directly from the SynScan controller, no external apps & no guiding.

Initially I did need to spend a lot of time getting things set up, levelling the mount, polar alignment, the longitude / latitude / altitude etc & in particular the 'cone error' caused me problems & took some figuring out. I do always perform the 3 star alignment routine, but everything seems fairly repeatable now, even I dismantle everything between sessions (I have permanently marked the mount position on my patio though!).

 

[Robculm]

On 29/03/2021 at 13:31, alacant said:

best to lose the dark frames with an 800d

Hi HTH,

I didn't pick up on this  specific comment at the time, but can you explain more about losing the dark frames? 

Thanks, Rob

[alacant]

Hi

We find they introduce noise/banding/artifacts. Purely hands on subjectivity. Try with and without dark frames. Try with and without dark optimisation.

With modern eos's, we get better results without dark frames of any type. As above, YMMV.

Summary: flat and bias frames with a 10 or so pixel dither between each light frame. Stack with linear clip or similar.

Cheers

Edited July 16, 2021 by alacant

[Robculm]

Thanks HTH, 

I have started dithering now that I have the guiding capability (ASIair pro seems to be working a treat in that regard), so will definitely give this a try.

Only problem is, it's brings me back to the problem that without darks, the OSC_Preprocessing option in Siril won't run, so as you've mentioned several times, I'm going to have to learn to use it 'manually'!!! 😆

Cheers, Rob

[alacant]

1 hour ago, Robculm said:

 I'm going to have to learn to use it 'manually'

+10!!!

[AstroMuni]

On 17/07/2021 at 08:10, Robculm said:

Only problem is, it's brings me back to the problem that without darks, the OSC_Preprocessing option in Siril won't run,

You can edit the script yourself to remove the darks or you can do what I do, use Sirilic. Gives you a lot of control to automate

Edited July 19, 2021 by AstroMuni

[Robculm]

Yes, I did install Sirilic at the weekend and had a try, but not quite there yet, ended up with crazy coloured patterns all over the image! 

However, while I was reading on this topic I discovered that you can download additional Siril scripts, including OSC preprocessing without Dark

https://free-astro.org/index.php?title=Siril:scripts

😁

[AstroMuni]

18 hours ago, Robculm said:

Yes, I did install Sirilic at the weekend and had a try, but not quite there yet, ended up with crazy coloured patterns all over the image! 

However, while I was reading on this topic I discovered that you can download additional Siril scripts, including OSC preprocessing without Dark

https://free-astro.org/index.php?title=Siril:scripts

😁

Yes those are there too, but Sirilic allows you to do much more on the fly rather than be restricted by the scriots available. As Sirilic generates a script you could save that and use that in future if you wish.