Help getting these results starting entirely in ST? Amazing results on data started elsewhere, not so much all in ST...

Questions and answers about processing in StarTools and how to accomplish certain tasks.
xonefs
Posts: 64
Joined: Sun Jul 18, 2021 7:30 am

Help getting these results starting entirely in ST? Amazing results on data started elsewhere, not so much all in ST...

Post by xonefs »

So I first started with startools on some combined RGB data in pixinsight for final touches and had amazing results just with the sharpen module and color panel to fine tune.

I hate working in pixinsight- it's a horribly unintuitive, slow, and clunky workflow. I would love to eliminate it and be able to work all in startools... if I can get it to work properly.

So my initial work in pixinight was crop- run EZ denoise suite on each channel, combine them into RGB file. Stretch, then some SCNR to remove 70% of the green.

Then I loaded non linear file into startools and abused the sharpen module, maybe some contrast I forget, and just clicked the hubble palette color preset and it got rid of more green.

The result was this, which already looks like my best image so far:
Image
I now try to start from scratch in starttools to take advantage of tracking and all ST has to offer. I run autodev, crop, autodev again. I tried wipe but it did weird things and made it look crazy so I skipped since data is so clean already. Then went contrast, sharpen. Then when I get to color seems like theres barely any color there. I hit HST presets and have to jack up saturation high and still the best results I get look like this, which is comparatively horrible
Image
aside from color seems like whole features/depth of nebula structures aren't there compared to the other one. I can tweak color to change it- but it's just different shades of bad and if I try anymore saturation it turns parts to mustard while the rest is washed out.

So what's going on here? seems maybe the individual narrowband channels aren't getting stretched evenly and contributing. Working with narrowband I would want each channel to get stretched to comparable levels and combined. I'm not sure how ST operates- is it only operating on the extracted luminance? how are the individual color channels contributing?

I'm just not sure where the issues with workflow are. i would really like to see some start to finish tutorials on good monochrome narrowband data from people who know what they are doing that would be very helpful.
Mike in Rancho
Posts: 1141
Joined: Sun Jun 20, 2021 10:05 pm
Location: Alta Loma, CA

Re: Help getting these results starting entirely in ST? Amazing results on data started elsewhere, not so much all in ST

Post by Mike in Rancho »

Your first image is very nice. I don't quite understand green removal or capping in false color when green actually means Ha (or in the case of duoband, OIII locked to blue), but, art is art and maybe it makes for more appealing coloration in the end?

We can probably play with a workflow if you want to post the stacks.

I also just downloaded some SHO data of the same target provided by a CN user, so I will also go through that in a little while and fully in ST. I'll for sure let you know if what I create ends up more like your first picture or your second one! :lol:
User avatar
admin
Site Admin
Posts: 3367
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: Help getting these results starting entirely in ST? Amazing results on data started elsewhere, not so much all in ST

Post by admin »

xonefs wrote: Sun Aug 01, 2021 4:47 am I tried wipe but it did weird things and made it look crazy so I skipped since data is so clean already.
Wipe is not an optional step! (see recommended workflow charts, etc.) Could you expand on what "made it look crazy" means? Unless there are some severe dark anomalies in your image (e.g. big dust donuts/specks or stacking artifacts), Wipe should be rather mild mannered. How did you ascertain this?

For an extremely light touch for narrowband data that is otherwise free of gradients (indeed not uncommon; it is usually less affected by artificial sources of light), use the Narrowband preset.
Then went contrast, sharpen.
aside from color seems like whole features/depth of nebula structures aren't there compared to the other one.

The ST rendition definitely shows far more detail/structure (luminance-wise) in the shadows and midtones, with the exception the shock fronts, which stand out much more in the PI version.
So what's going on here? seems maybe the individual narrowband channels aren't getting stretched evenly and contributing.
Not using Wipe means the biases in the different channels are not being modelled and subtracted, so when it is time to color balance, these (typically large) biases are still present, and therefore will be subject to the same multiplier that is applied to the actual chrominance signal. Don't. Skip Wipe. :evil:
Working with narrowband I would want each channel to get stretched to comparable levels and combined. I'm not sure how ST operates- is it only operating on the extracted luminance? how are the individual color channels contributing?
StarTools indeed famously processes luminance entirely separate from chrominance from the very start. You are in control of how much each channel contributes to luminance (e.g. by providing it exposure times).
I would really like to see some start to finish tutorials on good monochrome narrowband data from people who know what they are doing that would be very helpful.
This is "older" (just under 2 years) tutorial does just that, however with ST evolving as quickly as it is, things may look a bit different.

If you could post the stacks, I'd be happy to do a quick workflow with the data at hand though!
Ivo Jager
StarTools creator and astronomy enthusiast
Mike in Rancho
Posts: 1141
Joined: Sun Jun 20, 2021 10:05 pm
Location: Alta Loma, CA

Re: Help getting these results starting entirely in ST? Amazing results on data started elsewhere, not so much all in ST

Post by Mike in Rancho »

Along with Ivo's tips, xonefs, saturation does need to be adjusted, sometimes quite a bit, and also there are numerous SHO blends to choose from in the identity matrix even after you've pressed the HST preset. I did a SHO earlier today and the colors had the look of your first image. But it took some playing around with.

I also just tried the NB Wipe preset just a minute ago, and it made a big difference! Too bad I had already posted what I came up with earlier. :lol: I suppose that's what I get for not reading the manual.

The only thing I can't seem to replicate that the PI guys come up with (though uncertain whether proper or not) is the orange and teal coloration to SHO, no matter which matrix I try or the channel and saturation adjustments. Is that something that might come from altering the weighting in compose? Or should one even change those weights from the default, when the integration time is the same for each NB filter?
xonefs
Posts: 64
Joined: Sun Jul 18, 2021 7:30 am

Re: Help getting these results starting entirely in ST? Amazing results on data started elsewhere, not so much all in ST

Post by xonefs »

Thank you. I'll have to try it again... I was trying this out while traveling and camping so couldn't view everything with my reception and not the best place to be trying to process and learn. I guess I just had some confusion about the purpose of the wipe module from the initial videos/guide on the site since it seemed more about touching up sensor defects and gradients from what I saw so I assumed that was the main purpose.

Just trying to understand better how exactly startools works - and ho w people use it

Also I was curious about the exposure time in compose. I get for RGB data it is important to weight for the synthetic luminance- but how does that apply for narrowband? There will be less signal from Sii and Oiii, even with longer exposure times than Ha. So if I have 5 hours Oiii and 3 hours Ha and put that- wouldn't it just make Oiii contribute less to luminance which is not what you would want? seems it would be unnecessary and I could be better off leaving them default even weighting.
User avatar
admin
Site Admin
Posts: 3367
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: Help getting these results starting entirely in ST? Amazing results on data started elsewhere, not so much all in ST

Post by admin »

Mike in Rancho wrote: Sun Aug 01, 2021 11:45 pm The only thing I can't seem to replicate that the PI guys come up with (though uncertain whether proper or not) is the orange and teal coloration to SHO, no matter which matrix I try or the channel and saturation adjustments. Is that something that might come from altering the weighting in compose? Or should one even change those weights from the default, when the integration time is the same for each NB filter?
The weighting in the Compose module is purely responsible for the luminance (brightness). It does not impact coloring whatsoever (except of course, the way a color may be psychovisually perceived due to brightness of the luminance once composited - this is countered as much as possible though if using one of the CIELAB modes in the Color module).

Typically, it is really easy to get the famous golden/turquoise rendering using the HST preset and dialling the green bias (look at MaxRGB to see if green is very dominant if that helps). Green dominance is rather normal/expected due to the strength of Ha.

What StarTools will not allow you to do however (via the Color module that is), is fudging emission color where they simply do not exist.
Too often, I see "tutorials" for PS or PI where people go ham on their colors with totally inappropriate methods (stretching channels separately before combining, selective color manipulations). There is absolutely no need for "make believe" coloring; the real thing is almost always achievable through legit means that respect the signal and yields true, useful information about relative emission concentrations.
xonefs wrote: Mon Aug 02, 2021 12:26 am I guess I just had some confusion about the purpose of the wipe module from the initial videos/guide on the site since it seemed more about touching up sensor defects and gradients from what I saw so I assumed that was the main purpose.
You are right of course that most of Wipe's functionality indeed pertains to more advanced calibration problems, nevertheless measuting and removing a single constant ("bias") from a channel is also a form of (needed!) calibration (think of it as gradient removal where the gradient doesn't change).
Just trying to understand better how exactly startools works - and ho w people use it
Having some PI experience under your belt will greatly help! Things should hopefully fall into place quickly. This PI <-> ST translation table (let's just say you are not the first :) ) may also be enlightening when trying to relate things you learnt in PI to ST's workflow or terminology.
Also I was curious about the exposure time in compose. I get for RGB data it is important to weight for the synthetic luminance- but how does that apply for narrowband? There will be less signal from Sii and Oiii, even with longer exposure times than Ha. So if I have 5 hours Oiii and 3 hours Ha and put that- wouldn't it just make Oiii contribute less to luminance which is not what you would want? seems it would be unnecessary and I could be better off leaving them default even weighting.
Signal is signal - more is always better! :twisted: Signal from all the bands are added up and normalized (which is equal to averaging them).
The assumption and - often - reality, is that the vast majority of noise in astrophotography we deal with is "shot noise" (aka "Poissoninan noise"). This type of noise is part of the signal). It is applied noise (not added noise). What this means is that, if you have no signal, you have no noise; noise is proportional to the signal collected. Ergo adding up signal from separate bands (and the signal's noise component) is entirely possible (and typically recommended). Adding up all the signal is not harmful; if the weaker O-III attenuates the stronger Ha's signal that's fine; the important thing is that noise will have been lowered more than that the signal was attenuated. Simply stretch the Ha + O-III signal back up, if you wish, to the the previous level of the Ha. Though this time, if you do that, you should find the signal is less noisy at that brightness level.

There are, of course, exceptions, for example when most of the noise in a very-low signal stack is thermal noise (or some other type of "fixed" noise). In any such cases where the other form of noise is not negligible, the assumption this is all based on, falls apart of course. In such cases, a fixed amount of noise is present ("added noise"), and the "no signal means no noise" assumption falls apart.

Hope this helps!
Ivo Jager
StarTools creator and astronomy enthusiast
xonefs
Posts: 64
Joined: Sun Jul 18, 2021 7:30 am

Re: Help getting these results starting entirely in ST? Amazing results on data started elsewhere, not so much all in ST

Post by xonefs »

Oh yeah I think there was some confusion on my last question- I certainly know more signal/integration time is better. I was talking specifically about what you tell startools the exposure time is in proportion to the other narrowband channels, and how that affects the extracted luminance. Let's say you shoot double the Oiii compared to Ha, and you tell startools they are the same exposure time in compose vs the real proportion of exposure time and how that affects the extracted luminance. But not actually changing the actual exposure time.

I went back and did the wipe and it is much improved and a better starting point- but I'm still struggling to get those nice colors/look I somehow got in the first one, but maybe just need to experiment more.

I do still notice a lack of detail on the nebula ridges on top left and bottom even after wipe. I'm wondering if this is because SII is not contributing as much to luminance and washed out by the Ha signal because of how luminance is combined in the beginning? Typically and I know in this data Sii has a lot of very clear structure which is not always as visible in Ha since the nebulosity is so overwhelming in Ha and often covers the more detailed structure. Is this detail from Sii getting obscured in the beginning somehow with how L is combined or wiped when mixed with Ha? I'm not sure how that works.
Mike in Rancho
Posts: 1141
Joined: Sun Jun 20, 2021 10:05 pm
Location: Alta Loma, CA

Re: Help getting these results starting entirely in ST? Amazing results on data started elsewhere, not so much all in ST

Post by Mike in Rancho »

admin wrote: Mon Aug 02, 2021 2:50 am
Typically, it is really easy to get the famous golden/turquoise rendering using the HST preset and dialling the green bias (look at MaxRGB to see if green is very dominant if that helps). Green dominance is rather normal/expected due to the strength of Ha.
Thanks Ivo,

Jonny B posted up some short-integration NA+Pelican data on CN; my first attempt was a bit of a clunker and flawed, but after trying the NB Wipe preset after reading your post here, it went a lot smoother and the dust detail came out nicely. :thumbsup:

I still can't quite match the deep gold and turquoise, but maybe more experimentation, or as you say, some colors being created and posted aren't really real?

Xonefs if you are on CN, I got kind of close to your first color scheme above at this post here: https://www.cloudynights.com/topic/7822 ... ry11265989

A little darker overall might have created more contrast, but I'm kind of ok with it, especially as I have no SHO experience.
User avatar
admin
Site Admin
Posts: 3367
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: Help getting these results starting entirely in ST? Amazing results on data started elsewhere, not so much all in ST

Post by admin »

xonefs wrote: Mon Aug 02, 2021 4:06 am Oh yeah I think there was some confusion on my last question- I certainly know more signal/integration time is better. I was talking specifically about what you tell startools the exposure time is in proportion to the other narrowband channels, and how that affects the extracted luminance. Let's say you shoot double the Oiii compared to Ha, and you tell startools they are the same exposure time in compose vs the real proportion of exposure time and how that affects the extracted luminance. But not actually changing the actual exposure time.

I'm not 100% I understand the question, but this is from the Compose module docs if it helps;
On synthetic luminance generation

For practical purpose, synthetic luminance generation assumes that, besides possibly varying total exposure lengths, all other factors remain equal. E.g. it is assumed that bandwidth response is exactly equal to that of the other filters in terms of width and transmission, and that only shot noise from the object varies (either due to differences in signal in the different filter band from the imaged object, or due to differing exposure times).

When added to a real (non synthetic) luminance source (e.g. the optional source imported as 'Luminance File'), the synthetic luminance's three red, green and blue channels are assumed to contribute exactly one third to the added synthetic luminance. E.g. it is assumed that the aggregate filter response of the individual three red, green and blue channels, exactly match that of the single 'Luminance File' source. In other words, it is assumed that;

Code: Select all

red filter response + green filter response + blue filter response = luminance filter response
If the above is not (quite) the case and your know exact filter permeability, you can prorate the filter response by varying the Total Exposure sliders.

Finally, in the case of the presence of an instrument with a Bayer matrix, the green channel is assumed to contribute precisely 2x more signal than the red and blue channels.
I went back and did the wipe and it is much improved and a better starting point- but I'm still struggling to get those nice colors/look I somehow got in the first one, but maybe just need to experiment more.

I do still notice a lack of detail on the nebula ridges on top left and bottom even after wipe. I'm wondering if this is because SII is not contributing as much to luminance and washed out by the Ha signal because of how luminance is combined in the beginning? Typically and I know in this data Sii has a lot of very clear structure which is not always as visible in Ha since the nebulosity is so overwhelming in Ha and often covers the more detailed structure. Is this detail from Sii getting obscured in the beginning somehow with how L is combined or wiped when mixed with Ha? I'm not sure how that works.
If Ha is the stronger signal in the luminance frame, it may of course 'compete' with S-II detail. However, nothing should prevent you from brining out the detail through local dynamic range optimisation. E.g. if the Ha signal shows a amorphous patch with little detail, but the S-II adds detail on top (however faint), you should be able to bring it out.

FWIW, this is super quick process of JohnnyBravo's data from the CN thread you posted;
StarTools_2771.jpg
StarTools_2771.jpg (443.19 KiB) Viewed 3034 times
Good detail should be visible throughout the dynamic range.
According to the coloring, yellow corresponds to areas of strong Ha and S-II emissions; you should be able to see detail in those areas is not particularly compromised vs any other renditions.

I'm not sure what other people in the thread did in their workflows, so I can't really comment on rendition/color validity from a documentary view. I did notice someone mention a way in PI to get rid of purple using SCNR. That same technique can be performed in StarTools (though Tracking will have to be off!).

Go into the Layer module, choose the Invert Foreground Layer mode, Keep. Launch the Color module (Tracking off!), set Cap Green to however much you wish to reduce the purple (you can nuke it altogether with 100%). Keep. Go into the Layer module, choose the Invert Foreground Layer mode, Keep.

Hope that helps!
Ivo Jager
StarTools creator and astronomy enthusiast
xonefs
Posts: 64
Joined: Sun Jul 18, 2021 7:30 am

Re: Help getting these results starting entirely in ST? Amazing results on data started elsewhere, not so much all in ST

Post by xonefs »

Thanks ivo. I like your processing of that data there and with the hint of more green too. Yes I did mention scnr as to just what I did in pix and am used to doing- I would rather not turn tracking off in startools if I can get colors like that just from the color module and use the proper startools workflow. I'll have to look at those settings and try to replicate.

I can't seem to get deconvolution to work at all and am using the public alpha but that's another story and probably user error (I was the one having issues with the regular stable version on mac m1).

One last question I had is if there was any way to deal with halos in the wipe stage. I just have them in oiii sometimes with antlia filters, so I would think it would be ideal if there was a way to deal with them on that channel in the early stages.
Post Reply