NGC 1333

[Rodd]

TOA 130 at F7.7, SBIG STT-8300 with self guiding filter wheel.

Lum: 30 20 min subs

Red: 43 10 min subs

Green:  37 10 min subs

Blue:  15 10 min subs

I had hoped that additional lum would rescue the image...not to be.  As I add data, the image quality does improve, but very slowly.  I am calling it quits with almost 26 hours.  This is the best I can do.  I have a feeling that the difficulties I have with this target are related to sky conditions.  I have tried at F4.3 and now F7.7.  Don't think one was an advantage over the other.  I'll know when I try F3. Someday I will do this target justice.  The image looks ok small--buts its definitely no suited for full resolution mode.  

[Filroden]

This is my favourite object in the Northern sky. I love your image which is stunning at screen size. You’ve really captured the range of emission, reflection and dark nebulosity really well.

As you say, if you pixel peek then there’s definitely something going on. It looks like a processing artefact rather than an image capture issue. 

[ollypenrice]

Before reading your text (and knowing that you always shoot serious quantities of data) I thought 'sky trouble.' I've done this object twice and I believe you've put in the time to deserve a smoother result. (We also shot a full night's colour on it in a comparable instrument to yours last night so let's see what that gives.) My previous efforts were at 2.4 metres and 450mm. It is such a seductive object, but elusive.

Olly

[wimvb]

That's a great image. As @Filroden noted, processing may account for the artefacts. Maybe it's noise reduction that is to blame?

How are your sky conditions? Dark skies make a HUGE difference when you're trying to go deep. When I image at my dark site, away from Stockholm's skyglow (and a lamppost across the street), I find it much easier to get to the faint dust.

Thanks for sharing this image.

[cfpendock]

As usual, Rodd, another stunning image.  It may be my screen, but it seems just a tad magenta-ish for my taste.  But how I envy your possibilty to image for 26 hours on one target.....

Chris

[RichLD]

Rodd, that is absolutely gorgeous! I know that we all strive for our own version of perfection and can be extremely self critical but I for one think you should give yourself a hefty pat on the back! STUNNING!

[Rodd]

7 hours ago, Filroden said:

This is my favourite object in the Northern sky. I love your image which is stunning at screen size. You’ve really captured the range of emission, reflection and dark nebulosity really well.

As you say, if you pixel peek then there’s definitely something going on. It looks like a processing artefact rather than an image capture issue. 

Thanks Ken.  The artifacts are there right from the start--what I have been calling "faceting" is actually posturization I think (whatever that is).  Its there before I do any processing.  If I use noise control to smooth it out--the image looks like it has way too much noise control.  Its a strange area of the sky--when I shoot galaxies say, or other things that are in dark backgrounds (low light levels), I get a smooth background.  But this target gives me posturization--so I think its a situation where there is structure there, but so faint that it can almost be picked up but not quite.  Just a guess.  But I don't think its hardware, because subs of M33 taken the same night with the same filter are fine--10 degrees to the east at this target and I get nasty stuff--even when the target is at zenith.  Its very frustrating

Rodd

[Rodd]

7 hours ago, ollypenrice said:

Before reading your text (and knowing that you always shoot serious quantities of data) I thought 'sky trouble.' I've done this object twice and I believe you've put in the time to deserve a smoother result. (We also shot a full night's colour on it in a comparable instrument to yours last night so let's see what that gives.) My previous efforts were at 2.4 metres and 450mm. It is such a seductive object, but elusive.

Olly

It is a great target that I am determined to either "get right" or at least figure out why I can't.  Sky is no doubt an issue--but the target moves to the meridian, and in regions of my sky that are good enough to yield much better results on other targets.  Its faint--but not that faint--not like the squid, or soap bubble.  I just don't understand it.  

Rodd 

edit:  I am sure yours will look amazing.  Looking forward to seeing it.

[Rodd]

6 hours ago, wimvb said:

That's a great image. As @Filroden noted, processing may account for the artefacts. Maybe it's noise reduction that is to blame?

How are your sky conditions? Dark skies make a HUGE difference when you're trying to go deep. When I image at my dark site, away from Stockholm's skyglow (and a lamppost across the street), I find it much easier to get to the faint dust.

Thanks for sharing this image.

Thanks Wim.  The artifacts are there right from the beginning--and the goal in processing this image for me has to be minimizing the posturization artifacts.  A stack of 30 20 min lums is very grainy, with checkerboard pixelization.  I have processed this image so many times in an effort to remove the artifacts.  The above is the best balance I get .  I have attached the 10 hour lum stack.  No processing has been done.  At first blush, like the image above, it looks pretty good.  But if you zoom in a bit, you ask yourself "does this look like a 10 hour stack".  The other stacks are the same or worse.  

Rodd

[Rodd]

6 hours ago, cfpendock said:

As usual, Rodd, another stunning image.  It may be my screen, but it seems just a tad magenta-ish for my taste.  But how I envy your possibilty to image for 26 hours on one target.....

Chris

Thanks Chris.  Yes there is a color issue--among other things.  In this case, I am afraid, I am left with the feeling of 26 hours that could have been better spent!  At least I can now look forward to my next target!

Rodd

[Rodd]

6 hours ago, RichLD said:

Rodd, that is absolutely gorgeous! I know that we all strive for our own version of perfection and can be extremely self critical but I for one think you should give yourself a hefty pat on the back! STUNNING!

Thanks Rich....aw, heck....why not.  The door may not break first, but determination  I suppose is commendable   Don't get me wrong, I like the image for the same reasons I like all of the images that get posted.  They astound me and I will forever be amazed that we can produce images that not very long ago would have not been possible.  But that little shimmy in the steering wheel of your new car when you turn to the left is annoying, and should not be there.  

Rodd

[Filroden]

30 minutes ago, Rodd said:

I have attached the 10 hour lum stack.

This stack does not have the same degree of artefacts. There are a few dark pixels throughout and I wonder if you need to remove more of these (cosmetic correction) before stacking? I wonder if these, when stacked and combined into a colour image, are causing the effect.

I also see small dark spots on the edges of some of the stars, again suggesting these are added at calibration, stacking or post processing. What do individual subs look like?

[Rodd]

1 minute ago, Filroden said:

This stack does not have the same degree of artefacts. There are a few dark pixels throughout and I wonder if you need to remove more of these (cosmetic correction) before stacking? I wonder if these, when stacked and combined into a colour image, are causing the effect.

I also see small dark spots on the edges of some of the stars, again suggesting these are added at calibration, stacking or post processing. What do individual subs look like?

Cosmetic correction does not remove them (I tried).  Remember this is the best stack--a 10 hour stack--the color stacks look much worse.  When you combine them--Its even worse.  If you remove the dark pixels--you get a overly noise controlled looking image. I did it--There are no alignment issues with the FITs file (Stars don't have black pixels on sides).  maybe the JPEG compression did that.    I used  noise control  in the linear state that did a very good job of only removing the pixellation--I thought it was the answer. I think that image looks to smeary from noise control--virtually no noise control was used after combination.   I was going to post it as an example but I am swimming in versions and I don't want to get onto a run away train.  Some of them look better than the one above in some respects, some don't.  But all have problems of one sort or another.  Its like whittling a square piece of wood to fit in a circular hole--can you do it? Yes...but the level of precision will be far lower than if the 2 parts were machined to fit when created.  Just in case you want to try. here at the fits (xisf actually) files.  4 stacks--RGB and L.  If you can't work with xisf and you want to--I can convert them--I just noticed it (darn PI anyway!).  I am not suggesting to take them time to do it--just getting a jump on the down-load time in case you do.  Anyone feel free.

Rodd

B-15.xisf

G-37.xisf

L-30.xisf

R-43.xisf

[Davey-T]

Hi Rodd, guess you must be disappointed with the result knowing what a perfectionist you are, the main feature looks good but as you say the background is weird, I've seen this effect caused by a dodgy " Remove bad pixel map" but you say there is no processing done on that Lum' image.

Dave

[Rodd]

17 minutes ago, Davey-T said:

Hi Rodd, guess you must be disappointed with the result knowing what a perfectionist you are, the main feature looks good but as you say the background is weird, I've seen this effect caused by a dodgy " Remove bad pixel map" but you say there is no processing done on that Lum' image.

Dave

I agree--the higher signal areas look good, and overall, not unpleasant to look at.  And yes...no processing.  I know I am in trouble when I use DBE or ABE and I get a posturized, pixelated "faceted" look.  But in this case--I knew even before that point.  Oh well...I guess the best medicine is to shoot another target.  Fortunately it is supposed to be clear tonight (maybe a bit breezy though0.

Rodd

[Davey-T]

I don't use PI and no idea of the actual process that DBE uses to " Extract " the background, perhaps the authors could shed some light on it.

Dave

[Rodd]

1 minute ago, Davey-T said:

I don't use PI and no idea of the actual process that DBE uses to " Extract " the background, perhaps the authors could shed some light on it.

Dave

There is notation--but remember those math examples that have answers in the back of the book.....for even numbers?  Well, as was always the case, I have a problem with an odd problem

Rodd

[carastro]

Well I think it's a great image, but I can see your concerns when looking at it close up.  Might explain why I am struggle with this target myself.  

Will watch this thread with interest.

Carole 

[Rodd]

2 minutes ago, carastro said:

Well I think it's a great image, but I can see your concerns when looking at it close up.  Might explain why I am struggle with this target myself.  

Will watch this thread with interest.

Carole 

Its a hard one for sure.  Unfortunately I have pittered out on this target for a while.  This is the second time I have tried it.  Maybe the 3rd will be the charm

Rodd

[wimvb]

Here's my attempt at your data.

The masters are noisy for this long integration time. And the luminance data was very bright. To me, this indicates light pollution.

The combined rgb image had strong gradients, and I'm not sure I got them all out.

DBE is tricky on this target. There's so much dust that it is hard to find a background in the unstretched image. My usual workflow is to have a best guess at where to put the samples (15 pixels in size). Then process the image to reveal as much dust as possible. Next, I open the DBE dialog again and reposition the samples on a clone of the image. I save the process history of the original image, and reset it to before DBE. I then apply the new DBE, and redo the process from the saved process history. Since I've already done the process once, this second iteration is quite fast.

[Rodd]

14 minutes ago, wimvb said:

Here's my attempt at your data.

The masters are noisy for this long integration time. And the luminance data was very bright. To me, this indicates light pollution.

The combined rgb image had strong gradients, and I'm not sure I got them all out.

DBE is tricky on this target. There's so much dust that it is hard to find a background in the unstretched image. My usual workflow is to have a best guess at where to put the samples (15 pixels in size). Then process the image to reveal as much dust as possible. Next, I open the DBE dialog again and reposition the samples on a clone of the image. I save the process history of the original image, and reset it to before DBE. I then apply the new DBE, and redo the process from the saved process history. Since I've already done the process once, this second iteration is quite fast.

I had a bunch that were really dark too.  Its going from there to a better final brightness that is tough.  Nice effort on tough data.  I think you got the gradients.  I always have bad gradients in my RGB data.  Never in NB though. I have decided that no amount of data will result in this image being what I am looking for.  Unfortunately, its time to cut my losses.  

Rodd

[Rodd]

22 minutes ago, wimvb said:

And the luminance data was very bright. To me, this indicates light pollution.

Remember too that the lums were 20 min subs  30mof them

Rodd

[wimvb]

59 minutes ago, Rodd said:

I had a bunch that were really dark too.  Its going from there to a better final brightness that is tough.  Nice effort on tough data.  I think you got the gradients.  I always have bad gradients in my RGB data.  Never in NB though. I have decided that no amount of data will result in this image being what I am looking for.  Unfortunately, its time to cut my losses.  

Rodd

Stretching more starts to increase the "mottle" in the dust. It's also difficult to get any colour in this dust, so I backed off a little. Are you familiar with Gerald Wechselbergers PI videos? He has one about increasing brightness in dusty areas without blowing out stars:

http://www.werbeagentur.org/oldwexi/PixInsight/PixInsight.html

https://www.dropbox.com/s/dw6imbwpq8brvdt/Howto_enhance_nebuala_without_pushing_stars.wmv?dl=0

Rogelio Bernal Andreo also has a few interesting tutorials about lifting weak signal areas:

http://www.deepskycolors.com/tutorials.html

44 minutes ago, Rodd said:

Remember too that the lums were 20 min subs  30mof them

Rodd

What LP zone are you imaging from? The average background in your L master was 10k on a 64k (16 bit) scale, while the standard deviation (noise indicator) was around 500. Maybe your sky conditions just don't support 20 minute luminance frames??

Just a guess.

[Rodd]

5 minutes ago, wimvb said:

Stretching more starts to increase the "mottle" in the dust. It's also difficult to get any colour in this dust, so I backed off a little. Are you familiar with Gerald Wechselbergers PI videos? He has one about increasing brightness in dusty areas without blowing out stars:

http://www.werbeagentur.org/oldwexi/PixInsight/PixInsight.html

https://www.dropbox.com/s/dw6imbwpq8brvdt/Howto_enhance_nebuala_without_pushing_stars.wmv?dl=0

Rogelio Bernal Andreo also has a few interesting tutorials about lifting weak signal areas:

http://www.deepskycolors.com/tutorials.html

What LP zone are you imaging from? The average background in your L master was 10k on a 64k (16 bit) scale, while the standard deviation (noise indicator) was around 500. Maybe your sky conditions just don't support 20 minute luminance frames??

Just a guess.

Probably not--probably didn't help that I had a few subs in from nautical dark.  But M33 was shot with 20 min luminance subs, as was the Cacoon.  In fact M33 was shot simultaneously with NGC 1333--even earlier in the night while I waited for NGC 1333 to rise.  Subs for M33 came out great.  Maybe because it is much brighter.  I'll have a look at those videos.  Thanks

Rodd

[gorann]

Rodd,

I think it is generally a great image but you have taken out more dust and detail than the data can support without showing all that noise. If you just back off a bit in brightness, like in Wim's version, it is a really fine image. I have found that I often go too far in trying to get as much structure and dust out of the dark areas as possible and then, when I am about to give up on the image, I back off in brightness, and it often turns out to be a quite agreeable image. Still, it is worth to see where the limit is and then back off, having to realize that the dust that you can see in there is just too noisy to be presentable. Possibly, about half way between your bright version and Wim's darker one would be quite nice. Here is a 40:60 mix between yours and Wim's version after I applied a bit noise reduction to yours.

EDIT: and I also turned down the magenta saturation a bit and suppressed the lower end of the green i curves in PS