NGC 1333

[Mike in Rancho]

duo NB data of mine. And here SVD worked great

Maybe there are issues with the PSFs of my broadband stars Ron's data is also dual NB

I can at least contribute that there is no problem with broadband stars in general. I recently added some data to my Iris and took an image of the Ghost and in both cases there has been no problem with ringing. But I still have the impression that it depends on the processing done before engaging SVDecon. I had several runs with one image and in some cases there has been ringing and other times not. But I cannot say in which cases right now.

I wonder if there's something we can do to help Ivo to figure out in which cases there are problems. And to do some coordinated testing with new beta versions.

Best regards, Dietmar.

Ivo's busy.

But yeah maybe when time is right to get to it. Plenty of stuff in the beta thread already (probably jotted down ). My theory of apod+background support auto masking could be affected by prior steps in processing, maybe particularly bin and O/D.

[Mike in Rancho]

Yes, vdb 152 is really hard. I tried once with 20 hrs and failed...not a very nice image was the result (https://www.astrobin.com/zgf136/). I guess that NGC 1333 will be difficult from your location

That sounds reasonable. You are right about the altitude. I think in zenith it should be Bortle 4.

Well, NGC 1333 does get to near zenith (literally!), so maybe my Bortle will be better there too. Ya know, maybe SQM 17.99.

I'll see if there's anything more I can dig out of my vdB 152, probably with major binning for SNR boost, and put it up in another thread. Then see if perhaps adding time would be reasonable. Like, doubling or tripling it.

Though they were different targets and so not actually comparable, the SQM readings from ASTAP seem to say I would need 11.5x more integration to achieve similar SNR to you. And you had 27 hours!

That's based on 2.512^(dark SQM - bright SQM).

But, faster optics of f/3.8 compared to f/5.0 would give me a 1.73x advantage. (Slow f / Fast f)^2.

Meaning 6.6x more integration. So let's see, 27 hours x 6.6 =

[decay]

Ivo's busy.

Yeah, we already noted that

maybe when time is right to get to it.

And that's exactly what I meant Perhaps it would already be helpful if every one of us documents his data sets having problems and their respective ST logs. When later on there is a new beta version, we all could check the impact of this new version to the problematic data sets?! Just a thought ...

Best regards, Dietmar.

[decay]

To put things in full context, Ivo's quote also includes -

StarTools is renowned for its Color Constancy feature, rendering colours in objects regardless of how the luminance data was stretched, the reasoning being that colours in outer space don't magically change depending on how we stretch our image. Other software sadly lets the user stretch the colour information along with the luminance information, warping, distorting and destroying hue and saturation in the process.

And I'm also just making an inference from the default settings ST puts in place for us - Scientific Constancy for broadband, and Artistic for Narrowband (probably because there's more leeway to alter this in a false color setting).

I already thought about this, too. What I do not understand is the (strong) impact of the Scientific Constancy to the dark (brown) background nebulosity. My understanding of Ivo's explanation is that without Scientific Constancy bright parts would loose saturation. OK, so with Scientific Constancy the brighter parts will keep their saturation and that obviously happens. But I would assume that the darker background parts would show no difference between Scientific Constancy and Artistic Mode?

Best regards, Dietmar.

[Stefan B]

Now, I can think of several times where I was processing broadband and put the Style and LRGB Method back (sometimes just one), but I can't think of any image where I took Broadband out of Scientific and went into Artistic.

Just FYI: My latest NGC 1333 in this thread also uses the artistic option and has brown dust. You can bring it back by adjusting the saturation, especially dark saturation. You also could do it the other way round by reducing the saturation in Scientific as Mike suggested.

But to my eye the artistic style often is the better starting point for adjusting saturation. As Carles noted it often has less color noise (or it hides it better). AND bright stars are much easier to tame. Or are actually tamed by the artistic style. The two bright stars in my Iris/Ghost image drove me nuts. Especially the red one looked terrible. It dominated the whole image

Changing the style to artistic solved the issue immediately. Now both stars look fine. So my recent images almost all were processed with the artistic style. Though I often processed broadband images with Scientific and also got great results. Also depends which story you want to tell with your image.

I'll see if there's anything more I can dig out of my vdB 152, probably with major binning for SNR boost, and put it up in another thread. Then see if perhaps adding time would be reasonable. Like, doubling or tripling it.

Though they were different targets and so not actually comparable, the SQM readings from ASTAP seem to say I would need 11.5x more integration to achieve similar SNR to you. And you had 27 hours!

That's based on 2.512^(dark SQM - bright SQM).

But, faster optics of f/3.8 compared to f/5.0 would give me a 1.73x advantage. (Slow f / Fast f)^2.

Meaning 6.6x more integration. So let's see, 27 hours x 6.6 =

You can't replace a (relatively) dark sky in this hobby, right? I doubt if you can get satisfying results of dark nebulae under Bortle 8 conditions. Though I've heard people had some success with fast scopes (think RASA) and really short subs, like 10 to 30 sec but LOTS of them. That might also apply to your f3.8 system?

Regards
Stefan

[decay]

Hi Stefan,

sorry for the late post to this thread - I will soon reply to your Heart & Soul Nebula.

I reread this thread and I wonder if there's a conclusion regarding the initial point you described - the unexpected problem with your dataset. Are the other stacks (PI and Siril) significant better than your original ASTAP stack? Do you still needed/used Correlation Filtering for your last rendition?

BTW: I think you succeeded well with your last rendition. As I saw Carles two renditions I immediately thought that the solution must lie somewhere in between. And as Mike pointed out, the darks clouds still show some colour noise and/or are a bit clumpsy due to the use of Super Structure module. So maybe a little bit less aggressive stretch could help?

Anyway: Great object, great image!

Best regards, Dietmar.

[Stefan B]

Hi Dietmar!

I reread this thread and I wonder if there's a conclusion regarding the initial point you described - the unexpected problem with your dataset. Are the other stacks (PI and Siril) significant better than your original ASTAP stack? Do you still needed/used Correlation Filtering for your last rendition?

Both stacks are significantly better than the ASTAP stack since the ASTAP stack also had patterns of monochromatic pixels all over the field. Looked like colorful bear claws scratching over the image. As if hot pixels hadn't been calibrated out and were smeared. My darks were already 1.5 years or so old (upgraded them in the meantime) so that might be an issue. But I also dithered every second frame and that should be enough to get rid of that even without darks. AND the siril and PI stacks didn't have that problem at all even though they were calibrated with the same old darks. I used the same stacking parameters (though with new darks) for the Soul image and all was fine.

The Siril and especially the PI stack were also better in my eyes with regard to correlated noise. I didn't need such a high value for the PI stack than for Siril or ASTAP. Also the Siril and the ASTAP stack had a greater green tint after Wipe even when applying a DAF of 15 pixels and the PI stack did not. I think the subframe weighting has something to do with it so that bad subs had less impact. (Though in ASTAP I culled the 27 hrs down to 20 hrs or so based on the star count...the stack was as bad as with 27 hrs).

To cut a long story short, I don't have a definite answer. I guess it might be two or more factors contributing to the problems. I work in quality control and often experienced that a root cause analysis is quite easy if you have a single factor contributing but it gets very complicated when it's more...

BTW: I think you succeeded well with your last rendition. As I saw Carles two renditions I immediately thought that the solution must lie somewhere in between. And as Mike pointed out, the darks clouds still show some colour noise and/or are a bit clumpsy due to the use of Super Structure module. So maybe a little bit less aggressive stretch could help?

Thanks! You may be right about the less aggressive stretch. In my first rendition I lowered shadow linearity to 20% while leaving it at 50% for the second rendition. But it's an image whose appearance really depends on the screen. On my TV the second rendition appears overstretched. On one of my working PC screens it's almost too dark and on my working note book it looks great. So for the moment I settled on the second rendition

Regards
Stefan

[decay]

Thanks for your detailed answer, Stefan!

I don't have a definite answer. I guess it might be two or more factors contributing to the problems. I work in quality control and often experienced that a root cause analysis is quite easy if you have a single factor contributing but it gets very complicated when it's more...

True! So hopefully this simply does not happen again and everything is fine At least I learned that it sometimes can make sense to try a different stacker and to have a look, what comes out. I will probably not buy PI, but I still like DSS and I played a bit with Siril some time ago while processing C/2022 E3 (ZTF). OK, the outcome regarding the comet was not what I had hoped for, but I liked working with Siril. Especially the possibility to do background gradient removal from single subs seems to be promising for processing of broadband images having strong gradients due to LP or moon.

But it's an image whose appearance really depends on the screen.

Jepp, that's nearly always a problem - for me, too. I do image processing on my PC and then the result often is too bright on my tablet which I use as digital image frame and it's too dark on my mobile phone. But my impression is that it is better to do small brightness or gamma corrections afterwards using Film Dev or another external tool. Trying to make images 'brighter' while processing in ST (final stretch) often leads to unacceptable noise or clumpy background appearance.

Best regards, Dietmar.

[dx_ron]

So there I was, poking around Stellarium looking for a reasonable OSC target @910mm to fill new moon time until IC 405 rises high enough, when the Embryo Nebula caught my eye. NGC 1333 - hmm sounds familiar. Oh yea, Stefan's recent post.

27 hours Bortle 4??? Oh my. Time to have a re-think. (My SQM meter reads 19.1 at best at zenith here at home)

[Stefan B]

Hi Ron,

So there I was, poking around Stellarium looking for a reasonable OSC target @910mm to fill new moon time until IC 405 rises high enough, when the Embryo Nebula caught my eye. NGC 1333 - hmm sounds familiar. Oh yea, Stefan's recent post.

But there should be better targets which are already higher. Actually I had parking targets to fill the time until NGC 1333 was high enough (It's been the double cluster and Caroline's Rose.) Maybe something in Cepheus or Cassiopeia?

27 hours Bortle 4??? Oh my. Time to have a re-think. (My SQM meter reads 19.1 at best at zenith here at home)

Yes, it's probably not suited for a filler. But you don't need 27 hrs. There are lots of images around with less integration time and but better final result. But dark skies certainly help...

Regards
Stefan