m42

[alacant]

Hi everyone

As part of the is-lots-of-short-frames-better series, this time a dim refractor, so I thought double the frame length to 2 minutes. 

Pushed to just after the noise begins to appear in the dusty bits- Probably too far but after a 1/2 hour struggle, I'd had enough.
Detail and stars courtesy of Ivo's latest Spatially Variant what's-there-stays-there Deconvolution.

Am slowly reaching the conclusion that loadsa short frames are not that much different to fewer longer frames. It would be nice one day to lose the guiding entirely.

Thanks for looking and clear skies.

72x2min.siril 3.0.0a, st1.9.561beta ubuntu 22.04 

Edited December 10, 2023 by alacant
spanglish

[happy-kat]

perhaps short exposures are the natural way to less stars  too, stars are good

[Xilman]

I see you are using Ubuntu (like me) so you may wish to investigate FABADA. It can be quite spectacular when it comes to noise reduction so perhaps you could continue taking longer exposures.

[Elp]

I normally stick to 60s. Don't really see much difference in the stack for signal, just background noise. This was evident in my recent project when I compared my ha stack of 2h Vs 4-5h. The latter was far more noise free. However the o3 stack in comparison was noisy in both.

[ollypenrice]

I never really understand what you're trying to do. My agenda is simple: I try to take the best picture I can, within my means, and I have no preconceived ideas regarding what is good or bad, either in hardware or software, capture or processing, so long as I don't invent.

You seem to want to make astrophotos within a set of constraints of your own making. This approach is good, right, laudable but that is bad, wrong, reprehensible.  Perhaps you might publish a kind of manifesto saying, 'These are the constraints I impose on myself when making astrophotos.'  You are, of course, perfectly entitled to work within whatever constraints you impose upon yourself: we all do this.

I really don't think you need me or anybody else to offer a critique of your image. Its many strong points and its few weak ones are extremely obvious and you must be perfectly well aware of them. If I were to engage with them, though, I fear that I would be engaging with your unpublished manifesto on how astrophotos should be made.

Seriously, why not publish your 'Rules of engagement?'

Olly

[alacant]

9 hours ago, happy-kat said:

short exposures are the natural way to less stars

Could be. I didn't add or remove any stars though. I left them as they were after the first StarTools OptiDev. I'm not usually this lucky with refractors as usually the stars punch unsightly dark holes in the nebulosity. Maybe another advantage of short exposures? Dunno.

8 hours ago, Elp said:

normally stick to 60s

That may indeed be the sweet spot, especially with modern cameras. The only reason I went to 120s for this was that I had only a small aperture. My 60s trial was with a larger reflector. 

9 hours ago, Xilman said:

FABADA

Thanks. Downloaded and installed. I'm by no means a Python expert but promise to feed back my findings. I had an error with the noise switch. Could you give me an example cli for a fits file?

[alacant]

12 hours ago, ollypenrice said:

I never really understand

LOL. It's called experimentation!

Much of what we do both on this forum and on the ground is answer the what if or what should I do type of questions. On the rare occasions where I get a telescope to myself, I like to try and answer some of those questions. If the experiment produces a worthwhile image, that's simply a bonus. It may be irrelevant to the many, but if I can give back just a tiny piece of what this forum has given to me, then surely it is worth sharing.

Edited December 11, 2023 by alacant

[Xilman]

10 hours ago, alacant said:

Thanks. Downloaded and installed. I'm by no means a Python expert but promise to feed back my findings. I had an error with the noise switch. Could you give me an example cli for a fits file?

I call it as "fabada -out smooth.fits raw.fits 100" for an appropriate value of 100.

The noise value is chosen by experiment! Too small and no smoothing is done. Too little and artefacts appear.  I am sure that there must be a better way but I have not yet found it.

What sort of error do you get? There may be a bug in the code. I found a couple when it was first released but the author is very responsive and within 2 days he  told me that he'd fixed them. As I had already implemented my own fixes I didn't bother downloading a fresh copy.

[alacant]

11 minutes ago, Xilman said:

What sort of error do you get?

python3 fabadaCMD.py -out smooth.fits r_pp_cyg-1_Light_stacked.fits  100
Starting smoothing with fabada in r_pp_cyg-1_Light_stacked.fits image...
Warning: Size of array not supported
Warning: Size of array not supported
---

Any ideas?

 

[Xilman]

1 hour ago, alacant said:

python3 fabadaCMD.py -out smooth.fits r_pp_cyg-1_Light_stacked.fits  100
Starting smoothing with fabada in r_pp_cyg-1_Light_stacked.fits image...
Warning: Size of array not supported
Warning: Size of array not supported
---

Any ideas?

 

That is a new one on me.

What is the alleged size of the image? Do the FITS headers agree with the data which follows? I would hope so or something is very seriously wrong.

Is it a colour image? I only ever process monochrome images so there is only ever one HDU. If so, you may need to specify each plane separately and re-merge the three after smoothing. If this is the case, use the -hdu option setting its value (presumably) to 0, 1 and 2 in turn.

If you still can't get it to work, I suggest contacting the author for assistance.

Good luck!

[ollypenrice]

2 hours ago, alacant said:

LOL. It's called experimentation!

Much of what we do both on this forum and on the ground is answer the what if or what should I do type of questions. On the rare occasions where I get a telescope to myself, I like to try and answer some of those questions. If the experiment produces a worthwhile image, that's simply a bonus. It may be irrelevant to the many, but if I can give back just a tiny piece of what this forum has given to me, then surely it is worth sharing.

Certainly. I religiously place experimentation over theory.

OK, the image is, I think, deep and clean and the issue of exposure length is no longer all that critical since the CMOS chip has replaced the CCD. What I would want to do, were it my image, is sort out the saturation around the Trapezium. Whether or not you'd need short exposures for that region depends on what the exposure is like in linear form. It it isn't saturated when linear, all you need is a separate stretch blended in using one of the HDR routines, either ready-made or hand-made, so to speak. However, I do't know how you feel about a processing step like that.

Personally, I'd also lose the star halos but that would take time and involve another layer of processing.

Olly

[Xilman]

On 11/12/2023 at 10:47, Xilman said:

That is a new one on me.

What is the alleged size of the image? Do the FITS headers agree with the data which follows? I would hope so or something is very seriously wrong.

Is it a colour image? I only ever process monochrome images so there is only ever one HDU. If so, you may need to specify each plane separately and re-merge the three after smoothing. If this is the case, use the -hdu option setting its value (presumably) to 0, 1 and 2 in turn.

If you still can't get it to work, I suggest contacting the author for assistance.

Good luck!

OK, I sussed it. The code as written can not handle anything bu 1D (spectra, etc) or 2D (monochrome image) data. A colour image is 3D data, the third dimension being the colour channels.

So what I did was to separate out each channel in your image, smooth them separately, and re-combine into a tri-colour smoothed image.

 

Added in edit: I contacted Pablo about this issue.

 

Edited December 19, 2023 by Xilman

[Marvin Jenkins]

My question first and foremost in a thread as technical as this, with a picture capturing dust around M42, with many paragraphs of text about programs and technical procedures is......

Why is this in Getting Started with Imaging? I am not far from getting started and frankly to a beginner this is nonsense.

Someone beginning the journey would at least on first look would want to see what equipment you are using to see if what is being posted has any relevance.

M

[Xilman]

It started as a discussion about noise in images. I provided a link to my experiments with FABADA, which can do quite amazing things with noise in images. The discussion all rather snowballed from there, sorry.

If a moderator wishes to move some of this thread elsewhere, somewhere more suitable, I would be quite happy with that.

As for equipment, all I have used is the fabada software. The data was provided by @alacant so I can't say anything meaningful about how it was taken.

Paul

[alacant]

27 minutes ago, Xilman said:

move some of this thread

Yes. Of course. If the content is unsuitable, please move or delete it.
Please accept my apologies.

Cheers and clear skies.

[TiffsAndAstro]

On 10/12/2023 at 12:00, alacant said:

Hi everyone

As part of the is-lots-of-short-frames-better series, this time a dim refractor, so I thought double the frame length to 2 minutes. 

Pushed to just after the noise begins to appear in the dusty bits- Probably too far but after a 1/2 hour struggle, I'd had enough.
Detail and stars courtesy of Ivo's latest Spatially Variant what's-there-stays-there Deconvolution.

Am slowly reaching the conclusion that loadsa short frames are not that much different to fewer longer frames. It would be nice one day to lose the guiding entirely.

Thanks for looking and clear skies.

72x2min.siril 3.0.0a, st1.9.561beta ubuntu 22.04 

This is fabulous.

More info on your gear please, for context.