IC5070 Pelican Nabula

[jiberjaber]

I'm impressed with this 2x2 binning, first time using it - current work in progress but over the moon to see the initial result..

All processed in Pixinsight, this is just ABE after integration then combination using the SHO-AIP script with only the Ha, OIII, SII data and clicked the remove pink halos! .  No other processing so far other than a crop, rotate and flip so I could see the Pelican itself.  I might grab some more SII if I get the chance and see if that brings out anything further.

This is about 12 hours of data, though not sure if I should count the 2x2 as double time

36 Ha 600s 1x1 (Green)

23 OIII 600s 2x2 (Blue)

14 SII 600s 2x2 (Red)

ASI1600 & ZWO unmounted filters.

[geoflewis]

That's looking great Jason. I've never got my head around mixing binned 2x2 and 1x1 data, though I had to do it once when I get my filter bin settings wrong in SGP, so I know that it works.

[jiberjaber]

Thanks Geof, Yeah - it's new territory for me too - I wasn't sure if the stars would be blown out. I think they are a bit on the OIII (some of them are square!) but it doesn't seem to translate through to the combined image from what I can see.... I also needed to do new darks for the 2x2 both light sub length and flat darks. Not too much of a problem but In my head when I built my dark library I wasn't expecting to do 2x2 so it's now in addition

[MarkAR]

Lovely image. Well done.

[sloz1664]

Hi Jason,

Nice data capture and you have some detail there. However, how did you stack your images and what software did you use to process the final image. If you process as a SiiHaOiii image you should get more blues& golds in the image.

Steve

[jiberjaber]

4 hours ago, sloz1664 said:

Hi Jason,

Nice data capture and you have some detail there. However, how did you stack your images and what software did you use to process the final image. If you process as a SiiHaOiii image you should get more blues& golds in the image.

Steve

Detail is in the original post, Pixinsight is my tool of choice for calibration , alignment and stacking etc. 

 I loaded the 3 integrated NB filter images in to the SHO-AIP script. using the channel combination I am getting an error " source image dimension" on the SII file, though it seems OK on the script version. I was expecting more colour though... 

 

ETA: I aligned all the data against one good Ha Sub (1x1) so I was expecting the 2x2 sub integration to scale to that. If I channel combine SHO I get a very blue image so something isnt quite right! Welcome hints on where to look, error on the console is about the source dimension mentioned above, but when I look at all 3 main files they are all the same size as far as I can see (fits view) ???

Edited June 27, 2020 by jiberjaber

[jiberjaber]

Aha - with a quick click of the unlink on STF we are in a more colourful area, just need to resolve the magenta stars!

 

[jiberjaber]

Hmmm so a bit more playing and the Pelican seems to have gained weight! Not sure what happened to transpose the X/Y all I did was a crop then rotate 180, mirror horizontal and a rotate 90 clockwise! Anyhow running it through the script gives an option to remove some of the magenta in most (but not all) of the stars.

 

 

[5haan_A]

Hi,

 

I would like to ask what does 2x2 binning do for you with this picture? I am aware of what binning is but have never been sure on how to apply it. It looks like you have applied it to great effect because it is a stunning picture, but I am just not sure how and why it helped. 

Best,

[Sunshine]

WOW! that is one of the richest images of a nebulae i have ever seen, beautiful, almost looks like jade. The contrast is just outstanding, amazing work.

Edited June 27, 2020 by Sunshine

[carastro]

Hi Jason, that has come out really well, except for the colour and the magenta stars.   When I say colour, I am sure you can bring out more colour but as I don't use Pixinsight I can't advise you as I use solely Photoshop.

I had a play with your image in Photoshop.  The colour is there.  I will message it to you as don't want to mess up your thread. 

There must be a tool in Px for reducing the magenta stars.

Carole    

Edited June 27, 2020 by carastro

[x6gas]

I actually really like the original version - a take on this object that I've not seen before and it looks very sympathetically processed.

[sloz1664]

Hi Jason, As you are using Pixinsight to process your image, try using SCNR to transform your greens into a nice blue. I have some narrowband images of the Pelican. Here's a quick rendition of the colours as an SiiHaOiii with SCNR green removal:_

Steve

[jiberjaber]

Carole & Steve - thanks guys it's good to see what good looks like so I can see where I can push the image to - I'll have a play again today with the data and see what I can pop out - I did continue messing with it last night but my bottle of wine got the better of me LOL 

8 hours ago, 5haan_A said:

Hi,

 

I would like to ask what does 2x2 binning do for you with this picture? I am aware of what binning is but have never been sure on how to apply it. It looks like you have applied it to great effect because it is a stunning picture, but I am just not sure how and why it helped. 

Best,

As I understand it, it means that with the 2x2, in a shorter time I have more data to work with due to the increased sensitivity though it is at a lower resolution, so compensating for the brightness of the target in the different filters. 

I may benefit from some more SII data perhaps.

 

[jiberjaber]

Not sure how much more I can wring out of this.

Linear fit against SII then combined SHO, after which I have been playing with the curves all day, tweaked Saturation, Hue, A, B C components as well as the actual RGB & L channels. Had a play with some masks to try and target areas but couldn't quite get what I wanted just yet.

Overall image is rotated 180, flipped horizontally and then 90 deg clockwise to put the beak pointing downwards.  Final operation was to run the darkstructureenhance script which added a small amount of contrast to the pilar area

Ideally I'd like to preserve the Pelican beak detail as well as have the bright colours, suspect there might be some pixelmath in the future for this image to bring in something to show the beak... 

[sloz1664]

2 hours ago, jiberjaber said:

Not sure how much more I can wring out of this.

Linear fit against SII then combined SHO, after which I have been playing with the curves all day, tweaked Saturation, Hue, A, B C components as well as the actual RGB & L channels. Had a play with some masks to try and target areas but couldn't quite get what I wanted just yet.

Overall image is rotated 180, flipped horizontally and then 90 deg clockwise to put the beak pointing downwards.  Final operation was to run the darkstructureenhance script which added a small amount of contrast to the pilar area

Ideally I'd like to preserve the Pelican beak detail as well as have the bright colours, suspect there might be some pixelmath in the future for this image to bring in something to show the beak... 

Hi Jason,

Now doesn't that look better than the fist image. There are plenty of colours now which makes the Pelican "pop"

Steve

[geoflewis]

Wow Jason, what a transformation, you definitely spent your time on it today very well 👌

[tooth_dr]

Hi Jason

I really really like the smoothness and the detail of the first green image, but I do prefer the colours in the second one. Unfortunately it looks a lot noisier that the original version though.

I hope you dont mind I had a go bash at your data.  I made the first 'green' image mono, and used it as a luminance layer for the second image.

 

Adam

 

[x6gas]

I do like the most recent image you posted Jason, but the first one (the jade one) is something different and sometimes it's good to be different!

[jiberjaber]

24 minutes ago, tooth_dr said:

Hi Jason

I really really like the smoothness and the detail of the first green image, but I do prefer the colours in the second one. Unfortunately it looks a lot noisier that the original version though.

I hope you dont mind I had a go bash at your data.  I made the first 'green' image mono, and used it as a luminance layer for the second image.

 

Adam

 

Yes, there's no treatment for noise in the latest image, I spent all my time on messing with the colours!  I've done a fair bit of processing of OSC (DSLR) but for some reason I'm struggling to make the leap of using that workflow adapted for mono shots! Feels like the sequence of workflow for mono and OSC are a bit reversed in the linear part of it in some tutorials and in others they align with that I am used to for OSC.... 

Did you use Pixinsight to mix in the new lum layer? I tried messing with adding a luminance layer in pixelmath but couldn't get it right

Got to Thank @carastro and @sloz1664 for giving me something to aim for in terms of what it might look like when done, though I still couldn't quite pull out the reddish dust in the lower left of the image like Carole did!

@x6gasYes, the first view of that jade image was quite something though I like the newer colour versions, I liked the jade one as you can see a Pelican in the image, it was a struggle to resist the urge to keep the Pelican whole vrs the colour - perfecting the image would for me show the Pelican's beak better and colour too... I am sure it's possible with some magic of blending images etc... 

[tooth_dr]

8 hours ago, jiberjaber said:

Yes, there's no treatment for noise in the latest image, I spent all my time on messing with the colours!  I've done a fair bit of processing of OSC (DSLR) but for some reason I'm struggling to make the leap of using that workflow adapted for mono shots! Feels like the sequence of workflow for mono and OSC are a bit reversed in the linear part of it in some tutorials and in others they align with that I am used to for OSC.... 

 

I’m the same, went from DSLR to mono.  Funny I find NB easier, but I find LRGB is tricky! Unfortunately once the noise gets put in during processing it’s not easy to remove it at the end.  

I don’t use PI.  I literally scaled the RGB to match the Lum and added it in a few goes at 20% opacity.  Done in photoshop  

 

Edited June 28, 2020 by tooth_dr

[jiberjaber]

10 hours ago, tooth_dr said:

I’m the same, went from DSLR to mono.  Funny I find NB easier, but I find LRGB is tricky! Unfortunately once the noise gets put in during processing it’s not easy to remove it at the end.  

I don’t use PI.  I literally scaled the RGB to match the Lum and added it in a few goes at 20% opacity.  Done in photoshop  

 

Thanks for the detail, Yeah - I'm not sure how I might do that in PI, sounds like it is something I might be able to do in pixelmath but the problem with PI is it is too scientific sometimes when really the ability to be more 'arty' (i.e. just move a slider till it looks right, rather than guess a number for a formula) would be easier (though there are some scripts which go some way towards this)  My Adobe subscription has run out a few weeks ago and I've not seen any discounts lately for it so might have to persevere trying to work it in PI till a discount turns up!  

[jiberjaber]

I grabbed another 12 SII subs last night which prompted me to give another reprocess a go - there's is certainly more detail in the dust cloud in the bottom left now but there's still a fair amount of noise in the blue 'pool' below the off centre right star and additionally a new type of artifact/noise in the darker solid clouds above HH555 - like brush stroke marks, I suspect I might have to go back to post stretch and treat some noise reduction with a luminance mask but what I don't want to lose is the interplay between orange/red and black in the bottom left corner... perhaps some of the new artifacts are because this is the none drizzle version? anyway - here's "version 5" 

Weighted batch preprocessor with cosmetic correction, all subs aligned to the best in their filter.
(Selected bottom-up FITS)
Sub-frame selector, reject anything with score less than 67, approved subs separated out into own directory.
1x1 = 1.5 arcsec/pixel
2x2 = 2.99 arcsec/pixel
Gain 0.652
12bit camera res
Ha IC_5070_Light_H_Alpha_600_secs_2020_06_22T00_57_53_006_c_cc
OIII IC_5070_Light_OIII_600_secs_2020_06_26T00_28_51_008_c_cc
SII IC_5070_Light_SII_600_secs_2020_06_29T01_32_37_006_c_cc

Best sub from each filter then used as reference for Alignment and Local Normalisation.
Integration using SSWEIGHT from sub-frame selection for Sub Weights
Muredenoise of integrated images cycle-spin = 48
Gain 1x1 = 0.06
Gain 2x2 = 0.240
Gaussian Noise 2x2 = 24.99 (OIII & SII)
Gaussian Noise 1x1 = 50.65 (Ha)
Combination count SII = 26, OIII= 23 Ha=36

ABE of integrated images subtraction then division.
Align integrated imaged to the best Ha sub
Linear fit to SII (lowest median, can only do at this stage after images resize to Ha size)
dynamic crop on all
channel combination SII,Ha,OIII
correct magenta stars script (amount = 0.8)
90deg clockwise rotation
STF & Histogram stretch
Curves transformation Saturation, L, a, b and c 
Luminance mask applied to mask large cloud and expose the 'pool'
Multilayertransform 4 layers, 2,3&4 = +0.1 to sharpen slightly
Curves tweek with same mask
Inverted mask
Curves tweek (L) to bring out dust cloud detail

Edited June 29, 2020 by jiberjaber

[MarkAR]

Absolutely love it, well done.

[jiberjaber]

4 hours ago, MarkAR said:

Absolutely love it, well done.

Thanks Mark - I am just experimenting with some MLT noise reduction using a mask on the pre-combined mono images to see if I can dampen down some of the paint brush artifacts that seem to be there in the 2x2 images - will see if that makes a difference. Mind you I have noticed them in the HA 1x1 image too