StarTools

So I am still getting to grips with StarTools. I was wondering if anyone would be kind enough to try processing my data and see if they can get a better result than I did!

This is my first time ever capturing, stacking and processing a narrowband image. I know that there's more there to get and I don't particularly like the colour palette I ended up with (trying for "Hubble" but it is too green biased). The lights were taken between about 1 and 3:15am, when it started to get lighter (and I started to get sleepy), so I left the rig taking darks while I got some shuteye.

NGC6960, the Witch's Broom Nebula, also known as the Western Veil

Exposure:
Total Lights 2 hours 15 minutes
3 x 900 seconds each of Ha, Oiii, Sii
20 Flats individually for each filter
20 corresponding DarkFlats for each filter
19 x 900 seconds Darks

Hardware:
Altair Hypercam 183M
Celestron 80ED
SkyWatcher Motor Focuser
SkyWatcher HEQ5
Xagyl USB Filter Wheel
Altair 1.25" 7nm Ha Filter
Altair 1.25" 6.5nm Oiii Filter
Altair 1.25" 6.5nm Sii Filter
SkyWatcher ST80 Guide Scope
QHY5L-iic Guide Camera
Altair 0.6x Focal Reducer (for guide cam)

Software:
SharpCap
Cartes du Ciel
PHD2 Guiding
Deep Sky Stacker
StarTools

The raw stacked data is here:

Ha - https://www.dropbox.com/s/lk2mx0doy738kka/Ha.fits?dl=1
Oiii - https://www.dropbox.com/s/76e4sv69m78ln ... .fits?dl=1
Sii - https://www.dropbox.com/s/zbqva3r3turj91j/Sii.fits?dl=0

My result is:



StarTools log is:

https://www.dropbox.com/s/ssyqv7od1x4e7 ... s.log?dl=1

Many thanks!

-simon

I realised that there isn't much Sii in the Witch's Broom, so I thought it might be interesting to use the Sii as a Luminance channel for a Bicolour image with synthetic green. Here's my result, again I'm sure it can be done so much better!

Loading red channel data
File loaded in LRGB module [C:\Users\szymon\Desktop\SharpCap Captures\broom\Ha.fits].
Loading blue channel data
File loaded in LRGB module [C:\Users\szymon\Desktop\SharpCap Captures\broom\Oiii.fits].
Loading luminance channel data
File loaded in LRGB module [C:\Users\szymon\Desktop\SharpCap Captures\broom\Sii.fits].



Log file: https://www.dropbox.com/s/jllqpncucr1wl ... g.txt?dl=1

Hi,

Would you be able to gives us the link to the Sulphur II stack? (I think you accidentally gave use the Ha link twice!)

Thank you!

Doh! I mispasted the link, apologies! It's https://www.dropbox.com/s/zbqva3r3turj91j/Sii.fits?dl=1 (have also edited the first post).

This is what "the competition" does with the Ha/Oiii data:



(My friends jumped on the data and pulled it into Pixinsight -- I made the same mistake posting the Ha link twice so this was all they had). I am sure StarTools can do better!

Another competitive rendering of just the Ha/Oii data:

Oof! I won't lie - that's pretty challenging data to work with.

First off, you're absolutely right; though sulphur is definitely abundant, emissions through excitation are somewhat harder to spot/acquire in this nebula.
Consequently, many just pour all their effort in acquiring Ha and O-III and go for a bi-color (because - hey- it's narrowband and you have that freedom!)

To process this, you'd obviously use the LRGB module (soon to be superceded by the "Compose" module). Since they are all the same exposure time, you'd load simply S, H and O as R, G and B respectively. If they had been different exposure lengths, you could have created a synthetic luminance frame (ST 1.5. will do this automatically for you). If you'd like to do a bi-color, a synthetic luminance frame would allow you to "cheat" add signal from the SII stack to your Ha + O-III luminance dataset, while only Ha and O-III take care of the color.

Before doing anything, however, you'd have to align the 3 channels. This is really best done with your stacker. You first stack one channel, say Ha. Then once you have that one stacked, you use that stack as a reference frame when stacking the other two stacks. You should then end up with three perfectly overlapping stacks. I aligned them manually in StarTools using the Lens module (which is really more meant for Lens-related shifts and aberrations).

From there you can process the dataset roughly as normal, bearing in mind that the dataset is rather severely oversampled and noisy, though quite well calibrated ( ). Binning actually allowed me to eek out just enough signal to perform a small amount of deconvolution in ST 1.5.

This is what I did in the latest 1.4;
--- LRGB
Load S as red
Load Ha as green
Load O-III as blue
--- Lens
Parameter [Red Shift X] set to [18.9 pixels]
Parameter [Red Shift Y] set to [22.9 pixels]
Parameter [Blue Shift X] set to [29.2 pixels]
Parameter [Blue Shift Y] set to [14.9 pixels]
--- Auto Develop
To see what we're working with. We can see heavy noise and oversampling, as well as some alignment artefacts we introduced around the edges (from using the Lens module).
--- Bin
To make use of oversampling for noise reduction.
Parameter [Scale] set to [(scale/noise reduction 25.00%)/(1600.00%)/(+4.00 bits)]
--- Crop
Get rid of alignment artefacts.
Parameter [X1] set to [20 pixels]
Parameter [Y1] set to [17 pixels]
Parameter [X2] set to [1290 pixels (-69)]
Parameter [Y2] set to [891 pixels (-20)]
--- Wipe
Now, I think I can spot a slight bit of vignetting, but if it's there, it'd be only slight.
Because of my hunch I use the Vignetting preset, but back off the Corner Aggressiveness.
Parameter [Corner Aggressiveness] set to [92 %]
Parameter [Dark Anomaly Filter] set to [6 pixels] to help Wipe ignore dark noise better.
--- Develop
The dataset is too noisy for AutoDev to get a lock onto the real detail, so a manual Develop will have to do.
Parameter [Digital Development] set to [89.30 %]
Parameter [Dark Anomaly Filter] set to [10.6 pixels]
--- Color
Anything goes here really, but this should emulate a Hubblish-palette (though there is a reason the Veil Nebula is not image too often in SHO).
Parameter [Bias Slider Mode] set to [Sliders Reduce Color Bias]
Parameter [Style] set to [Artistic, Detail Aware]
Parameter [LRGB Method Emulation] set to [RGB Ratio, CIELab Luminance Retention]
Parameter [Dark Saturation] set to [1.00]
Parameter [Bright Saturation] set to [3.80]
Parameter [Saturation Amount] set to [500 %]
Parameter [Blue Bias Reduce] set to [1.68]
Parameter [Green Bias Reduce] set to [1.61]
--- Wavelet De-Noise
Parameter [Scale 1] set to [90 %]
Parameter [Scale 2] set to [90 %]
Parameter [Scale 3] set to [90 %]
Parameter [Scale 4] set to [90 %]
Parameter [Scale 5] set to [81 %]
Parameter [Mask Fuzz] set to [1.0 pixels]
Parameter [Scale Correlation] set to [6]
Parameter [Color Detail Loss] set to [12 %]
Parameter [Brightness Detail Loss] set to [12 %]
Parameter [Grain Size] set to [23.0 pixels]
Parameter [Read Noise Compensation] set to [19.75 %]
Parameter [Smoothness] set to [82 %]

You'd get this; For fun, below are some test runs with ST 1.5 as well (I'm currently throwing anything at it I can get my hands on ). These runs use the Compose module to create synthetic luminance from the three stacks, while using the new engine to process color and luminance separately, yet simultaneously. As a result, the workflow is quite close to the one above, yet it's doing full luminance and color processing in one hit (cross-consulting the datasets for noise reduction and detail enhancement purposes). You can see the difference applying Decon makes, though the stretch here was a little more conservative;
And a bi-color (just Ha as red and O-III mapped as blue with green interpolated . Many AP'ers like to add O-III to green as well though, rather than interpolate green as was done here;
Any questions, remarks, let me know. Hope this helps!

EDIT: one more, showing you the more typical H:O:O (R:G:B) bi-color rendering, more popular because O-III emission really do look more teal green in real-life;

Those are wonderful results -- especially the last one! I don't suppose you have a beta program for people to test the latest latest bleeding edge v5? Happy to send my qualifications if you're interested, I believe I could provide useful input and am happy to keep pulling in updates and won't complain if things don't work

I am surprised that you found you had to align the three channels -- I thought I actually did that properly in DSS! I first marked one of the Ha images as a reference image:



and then in the Oiii and Sii I added that image, set it as reference, but didn't select it for stacking:





When I put them together in StarTools, it seemed to me that they were correctly aligned. What did I do wrong?

I am interested as to why this is a challenging data set -- in particular, I would like to learn what I can do to make it better next time. I have to say, I actually thought that what I got was relatively "clean" by my standards! The flat and dark frames appeared to me to have done their jobs pretty well (as you say, it is "well calibrated")! Is it just that this is a low amount of data captured, or are there actual mistakes somewhere that I could improve? When you say that it is "rather severely oversampled and noisy", would the fix for that for me to capture binned 2x2? My camera is a Hypercam 183M which has very small (2.4um) pixels and lots of them (5440x3648). If I binned at 2x2 or more, I'd have bigger pixels and more sensitivity. As for the noise, it's likely that this is at least in part due to the analogue gain settings -- the gain in SharpCap on this camera goes from 0 to 5000, apparently (according to "the internet") unit gain is 400, these were taken with gain 1600. I'll try going at 400 gain next time, I just wanted to make sure I actually capture some data (my previous attempt was 300 second frames of Ha with gain 400, and there was no nebulosity visible at all!). I suppose there would also be less noise if I get more data (45 minutes each is really not a lot!).

(My favourite of all of your renderings is the HOO bi-colour at the end; when I can get those kinds of fantastic results myself from StarTools I'll be over the moon!

I'm not familiar with sharpcap's gain settings or the hypercam's so take this with a grain of salt. What I do know is that ZWO uses the same chip in their ASI183MM. The gain goes from 0 to 300 and unity is at 120. If we assume a similar relation of the two scales of 0-5000 and 0-300, I would assume unity to be more like 2000 on a 0-5000 scale. (120*5000/300) incidentally, 2000/5000 is 0.4, so maybe that's where the "400" came from????
I shot from a fairly light polluted area and use a gain of 200 (on the 0-300 scale) i'd assume that would be around 3333 on a scale of 0-5000?
Another thing i'm curious about is the location you shot from. 15 minute exposures are very long for this camera unless you have a dark sky. My NB exposure time is 3 minutes per sub. ( 8" F/4 )
I'd try bumping the gain up to around 3500 (out of 5000) and perhaps 8 minute exposures. And take a lot of them! I grab at least 100 per channel, but I like to get 200 for each channel if I can. For me that 5 to 10 hours total exposure per channel. Again, maybe you shot from a dark location and don't need to get that much data. These CMOS cameras work best with lots of short exposures, higher gain, and dithering .... IMHO

I live in London, England, known fondly to those of us who astroimage as “LPLondon”. The light pollution is basically at the levels of Victorian Smog. My back garden is Bortle 8. I do not have Dark Skies. That said, this is why I have 7/6.5nm narrowband filters, so that I can ignore the light pollution!

The figure for “unity gain” comes from Nick at Altair Astro so I am pretty sure if he says it’s 400 then it’s 400 . That said I previously tried 300 second captures at gain 400 and got basically no visible nebulosity. This is why i went up in gain...