NGC1333 LRGB

[wimvb]

3 minutes ago, Rodd said:

Anybody want to try there hand at 1,400 subs?  Might not get another chance!  I'll post the fits if you have the hankering

Rodd

How many TB zipped? ?

[carastro]

I wouldn't mind having a go at the stacked unprocessed files but I don't think my PC would cope with all those individual files.

By the way, what temperature do you image at Rodd? 

I image at -15 but I wonder whether I would get less noise if I went lower. 

Carp;e 

[Rodd]

6 minutes ago, vdb said:

however 1 min subs is probably to short to get the faint dust noise free,

60 sec at F3?  There is no way Lum can be longer.  maybe RGB.  But 60 sec should be long enough with thousands of subs according to all literature.  Besides--I tried 2min subs when I was shooting SH2-119 and it made no difference.  2min or 1min or 30sec even were all the same basically.  that was a frustrating project too.

Rodd

[wimvb]

7 minutes ago, Rodd said:

That's the trouble with DBE and images like this--at the time you place the markers it is VERY difficult to determine what is background and what is nebula--almost impossible..  I toil at this for hours.  

Rodd

Dynamic crop to the rescue!

Seriously though, if you use your image to place dbe markers on the background, but then apply that dbe instance to the original data, you may have more success. I've done this on a few occassions.

[Rodd]

4 minutes ago, carastro said:

I wouldn't mind having a go at the stacked unprocessed files but I don't think my PC would cope with all those individual files.

By the way, what temperature do you image at Rodd? 

I image at -15 but I wonder whether I would get less noise if I went lower. 

Carp;e 

I use to shoot at -40 with the STT-8300.  Now I shoot at -20 because there were some days that the asi1600 just could not reach the -40 and I wanted to pick a temp that could be reached all the time so I don't have to worry about shooting new recalibration files each time.  I have always been able to reach -20.  -15 is good.  I don't think you'll see much difference.  I don't think my issue is noise (from camera/scope).  I think its sky related, so temp won't matter.

Rodd

[Rodd]

6 minutes ago, wimvb said:

Dynamic crop to the rescue!

Seriously though, if you use your image to place dbe markers on the background, but then apply that dbe instance to the original data, you may have more success. I've done this on a few occassions.

That's an idea--it means processing once for DBE then again for real.  I guess that is the best option.  I  often clone a stack and stretch it hard to differentiate background......but when the dynamic range is this small, even that fails me.  I will try your suggestion.  I will process this data again......one more time.  I am not hopeful though.

Rodd

[wimvb]

3 hours ago, ollypenrice said:

I'll see if I have the original widefield data but I suspect I'll have canned it.

 

28 minutes ago, ollypenrice said:

I've found my original data and will try a reprocess since this is an interesting problem bit of sky and my existing process is whacky.

Do you mean this data? ?

These are the master frames you shared 2 years ago. They're still sitting on my hard drive.

[Rodd]

20 minutes ago, wimvb said:

How many TB zipped

 

18 minutes ago, carastro said:

I wouldn't mind having a go at the stacked unprocesb302.fitsed files

Not the subs silly people.  Here are the FITs stacks.  Please put me to shame.  The lum stack is SLlrgb1348 (super luminance made from lrgb subs totaling 1,348)

r353.fit

g302.fit

b302.fit

SLlrgb1348.fit

[wimvb]

33 minutes ago, Rodd said:

Anybody want to try there hand at 1,400 subs?  Might not get another chance!  I'll post the fits if you have the hankering

Rodd

 

8 minutes ago, Rodd said:

Not the subs silly people.

I guessed that, but you actually did write 1,400 subs. ?

You put me in a difficult position here: Grade student work, or have fun?

I'll grade first and have fun with your data after that. This saves my weekend.

Btw, could you post the L master as well (not just the synthetic L)? I want to try to remove the artefacts.

[Rodd]

9 minutes ago, wimvb said:

 

I guessed that, but you actually did write 1,400 subs. ?

You put me in a difficult position here: Grade student work, or have fun?

I'll grade first and have fun with your data after that. This saves my weekend.

Btw, could you post the L master as well (not just the synthetic L)? I want to try to remove the artefacts.

I Meant 1,400 subs of data....but now that I think about it, the number of subs makes no difference.  And regarding responsibilities (or shirking them) I know how you feel....I am at work (shhh..)

Rodd

[wimvb]

1 hour ago, Rodd said:

That's an idea--it means processing once for DBE then again for real.  I guess that is the best option.

As I said, you can use the image you already have. For future projects, you can just process the L master first for this. No need to do a full lrgb with all the trimmings.

You can also download a wider field image from the interweb, star align that to your master, and then use that for dbe markers. This isn't cheating, because you don't use any parts of the image in your product. You only use it as a reference.

[Rodd]

10 minutes ago, wimvb said:

As I said, you can use the image you already have. For future projects, you can just process the L master first for this. No need to do a full lrgb with all the trimmings.

You can also download a wider field image from the interweb, star align that to your master, and then use that for dbe markers. This isn't cheating, because you don't use any parts of the image in your product. You only use it as a reference.

But won't the dbe points be reflective of the image on the screen--not your data?  

[ollypenrice]

2 hours ago, wimvb said:

 

Do you mean this data? ?

These are the master frames you shared 2 years ago. They're still sitting on my hard drive.

Yup, I'd forgotten that but I've found them. Thanks Vim.

Olly

[wimvb]

Guess what I prioritised?

 

As always, messed up in PixInsight.

DBE: division on L, subtraction on RGB

Noise control: tgvdenoise in linear and nonlinear stage on L only. MMT on chrominance in linear stage of RGB image. SCNR before lrgb combination.

Masked stretch of rgb. Histogram stretch of L. No star reduction.

I also got that nasty edge after stretching, so I took to some old bonsai wisdom: "when in doubt: cut".

I'll go back to this later to try some more fancy stuff (removing the coloured streaks, star reduction)

And now back to grading.

[wimvb]

For comparison, this is how I mistreated Olly's data back then.

[Rodd]

13 minutes ago, wimvb said:

As always, messed up in PixInsight.

DBE: division on L, subtraction on RGB

Noise control: tgvdenoise in linear and nonlinear stage on L only. MMT on chrominance in linear stage of RGB image. SCNR before lrgb combination.

Masked stretch of rgb. Histogram stretch of L. No star reduction.

I also got that nasty edge after stretching, so I took to some old bonsai wisdom: "when in doubt: cut".

I'll go back to this later to try some more fancy stuff (removing the coloured streaks, star reduction)

And now back to grading.

While better (could use some star control), it still is very noisy for 1,348 subs.  I don't get it.  unless it's not noise--but LP.  I am beginning to despise the Balrog

Rodd

[Rodd]

4 minutes ago, wimvb said:

For comparison, this is how I mistreated Olly's data back then.

Pretty close to the same.........why doesn't that ameliorate me?

Rodd

[vdb]

4 hours ago, Rodd said:

60 sec at F3?  There is no way Lum can be longer.  maybe RGB.  But 60 sec should be long enough with thousands of subs according to all literature.  Besides--I tried 2min subs when I was shooting SH2-119 and it made no difference.  2min or 1min or 30sec even were all the same basically.  that was a frustrating project too.

Rodd

I'm still  convinced longer is beter, even with CMOS camera's, in very signal poor area's like this one, you need 5 minute,

the question is can your skies handle this, because LP will ruin any kind of duration, LP swamps the signal, so whatever you do it's futile.

Maybe doing only RGB and longer subs is the answer (on the condition your RGB filter have an LP stop in them, which with LED is becoming less and less effective anyway ...)

 

/Yves

 

[Rodd]

8 minutes ago, vdb said:

I'm still  convinced longer is beter, even with CMOS camera's, in very signal poor area's like this one, you need 5 minute,

the question is can your skies handle this, because LP will ruin any kind of duration, LP swamps the signal, so whatever you do it's futile.

Maybe doing only RGB and longer subs is the answer (on the condition your RGB filter have an LP stop in them, which with LED is becoming less and less effective anyway ...)

 

/Yves

 

According to "theory" electrons hit the sensor and signal is the measure of the number of photons.  As long as you are above the noise threshold so that photons are striking the sensor and can be measured, the duration of the sub is irrelevant.  The difference between 10 1min subs and 1 10 minute sub is negligible.  What is important is the total duration that the sensor is exposed to the sky--it can be at one long go, or in 100 smaller gos.  Same.

Rodd

[carastro]

Well I said my PC wasn't up to it, but that was an understatement.  I couldn't even process it because it kept telling me the scratch disks were full.   So I have spent the best part of 2 hours deleting old files I don't need any more and transferring those I do to my back up drive to make space.  Decided to get myself a 2nd back-up drive just for my astro data as the current one is getting full.  I like to keep all the Raw and fits files for a few years in case I need to go back to the data for either reprocessing or adding to new data which i did only in the last few days.  So it piles up somewhat.  

Then when I did manage to get back to it, my result was even more noisy than yours so have given up for tonight.

Carole 

[Rodd]

15 minutes ago, carastro said:

Well I said my PC wasn't up to it, but that was an understatement.  I couldn't even process it because it kept telling me the scratch disks were full.   So I have spent the best part of 2 hours deleting old files I don't need any more and transferring those I do to my back up drive to make space.  Decided to get myself a 2nd back-up drive just for my astro data as the current one is getting full.  I like to keep all the Raw and fits files for a few years in case I need to go back to the data for either reprocessing or adding to new data which i did only in the last few days.  So it piles up somewhat.  

Then when I did manage to get back to it, my result was even more noisy than yours so have given up for tonight.

Carole 

The stacks are no larger than any other stacks (8-10mb I think).  You must have had a full computer!.

Rodd

[wimvb]

Totally forgot: what gain do you use?

Cmos cameras have variable gain, and a dynamic range that decreases as you increase gain. You can get dr back by increasing the number of subs, but I don't know if there's a tradeoff.

There is an astrophotographer on astrobin who used the same camera I use (he now has a 1600MM) and he stubbornly sticks to very low gain, and 4 minutes exposures. But he delivers. While read noise is higher at low gain, full well is also higher, and dr is higher also. And btw, he also images from the suburbs. He compensates for that by NEVER imaging less than 8 hours on a target. (8 x 60 / 4 = 120 low gain, high dr subs as a minimum)

Just a thought.

[Rodd]

34 minutes ago, wimvb said:

Totally forgot: what gain do you use?

Cmos cameras have variable gain, and a dynamic range that decreases as you increase gain. You can get dr back by increasing the number of subs, but I don't know if there's a tradeoff.

There is an astrophotographer on astrobin who used the same camera I use (he now has a 1600MM) and he stubbornly sticks to very low gain, and 4 minutes exposures. But he delivers. While read noise is higher at low gain, full well is also higher, and dr is higher also. And btw, he also images from the suburbs. He compensates for that by NEVER imaging less than 8 hours on a target. (8 x 60 / 4 = 120 low gain, high dr subs as a minimum)

Just a thought.

I have only ever used unity gain (139).  If I raise the gain to 200, will 60 sec subs look brighter?  or dimmer?

Rodd

[wimvb]

1 hour ago, Rodd said:

I have only ever used unity gain (139).  If I raise the gain to 200, will 60 sec subs look brighter?  or dimmer?

Rodd

The subs will be brighter. Increasing gain is  like increasing ISO on a DSLR.

If you look at the diagram (the section with "gain" heading), you'll see that at gain 200 you have doubled the real gain (0.5 e/ADU vs 1 e/ADU, that is, each captured photon will increase the ADU count by two). You will also have lower read noise. But most important of all: dynamic range is one stop lower than at unity gain (139). This means that either will detail fall below the noise floor, or stars will become overexposed. Otoh, if you lower the gain to 0, you will have 12.5 stops dynamic range. This means a lot mare room for detail between the noise floor and bloated stars.

But I wouldn't blindly trust the numbers. As always, best is to experiment to find the gain settings and exposure time that best suits your conditions and equipment.

[Rodd]

41 minutes ago, wimvb said:

The subs will be brighter. Increasing gain is  like increasing ISO on a DSLR.

If you look at the diagram (the section with "gain" heading), you'll see that at gain 200 you have doubled the real gain (0.5 e/ADU vs 1 e/ADU, that is, each captured photon will increase the ADU count by two). You will also have lower read noise. But most important of all: dynamic range is one stop lower than at unity gain (139). This means that either will detail fall below the noise floor, or stars will become overexposed. Otoh, if you lower the gain to 0, you will have 12.5 stops dynamic range. This means a lot mare room for detail between the noise floor and bloated stars.

But I wouldn't blindly trust the numbers. As always, best is to experiment to find the gain settings and exposure time that best suits your conditions and equipment.

So if I wanted to capture fine scale details, I really need to use low gain and take long subs