presentation questions

General discussion about StarTools.
Post Reply
Rkonrad
Posts: 252
Joined: Sun Feb 07, 2016 2:55 am

presentation questions

Post by Rkonrad »

Hi all,

I'm giving a startools workshop at my local astronomy club (RASC in Winnipeg Canada) and have a few questions.

1. What is the difference between "shadow linearity" and "detector gamma"? I find the stretches of autodev to be very conservative and low contrast - especialy in the later versions of startools - so I use these quite a bit. Is the former mid-tones and lower, latter mid-tones and higher?

2. Would anyone have a different way than the website in describing the function of "Super structure"? There is a significant "glow" in the main dso when the airy disk radius is set 50% and higher. (isolate) What is this?

Thanks in advance!

Richard
hixx
Posts: 252
Joined: Mon Sep 02, 2019 3:36 pm

Re: presentation questions

Post by hixx »

Hi Richard,
please have a look at Guy's excellent User Notes with description of all parameters, including best practice
viewforum.php?f=12
If You look for a document, these are also included in the Inofficial guide available on ST download site

Clear Skies!,
Jochen
Mike in Rancho
Posts: 1158
Joined: Sun Jun 20, 2021 10:05 pm
Location: Alta Loma, CA

Re: presentation questions

Post by Mike in Rancho »

Hi Richard,

I'll go backwards. ;)

2. I don't think I can do any better than the ST website and Guy's notes on Super Structure.

1. Yes the latest release can give a feeling of being muted after OptiDev, relative to what we were probably all used to. This came about from increasing bit depth...or something like that. :think: Ron, Dietmar, myself, and the gang had a thread or two going where this was all discussed and experimented with. I need to get back to that actually, if I can find the time. :D But in a nutshell OptiDev was grabbing onto stellar data of questionable linearity, if not outright clipped, and producing its optimal global stretch with that figured in. Alas, that sort of bright star anomaly is pretty darn common. That resulting final stretch, however, often increased DSO detail contrast -- which could look great -- but on the other hand could trash stars with unrecoverable ringing (albeit there's some interesting optical effects involved), and IMO could also cause a harsher or noisier-seeming background.

With that fixed, increasing DSO contrast and bringing out details falls more to the enhancing modules. Or yes, you can pre-tweak the data using detector gamma and shadow linearity, before it gets handed off to OptiDev's data analysis.

Guys' User Notes on AutoDev are again useful here, though I believe there's a typo or error in describing gamma +/- 1.00 that needs fixing. We'll probably have to tell him! Now, I don't completely understand gamma correction, and depending on where you look it up sometimes it seems >1.0 is brighter, other times it's <1.0. Go figure. But more or less I think you can consider it to be a global mid-tones adjustment. If you look at something like levels in Gimp, it would be the middle triangle that you can drag back and forth. The other preliminary control, shadow linearity, compresses the shadows (providing more range to highlights) if below 50, expands if above 50.

Somebody correct me if I botched any of that! :)
decay
Posts: 456
Joined: Sat Apr 10, 2021 12:28 pm
Location: Germany, NRW

Re: presentation questions

Post by decay »

Rkonrad wrote: Fri May 24, 2024 4:25 pm I find the stretches of autodev to be very conservative and low contrast - especialy in the later versions of startools
Mike in Rancho wrote: Tue May 28, 2024 5:10 am Yes the latest release can give a feeling of being muted after OptiDev, relative to what we were probably all used to. This came about from increasing bit depth...or something like that. :think: Ron, Dietmar, myself, and the gang had a thread or two going where this was all discussed and experimented with. I need to get back to that actually, if I can find the time. :D But in a nutshell OptiDev was grabbing onto stellar data of questionable linearity, if not outright clipped, and producing its optimal global stretch with that figured in. Alas, that sort of bright star anomaly is pretty darn common.
I'm afraid, we're not done with this yet. :cry: Stefan (@Stefan B) and I felt independently of each other the need to brighten up our small galaxies using Super Structure's brighten preset, because it was not possible to get them bright enough just using OptiDev. And lately Andy (@AndyBooth) explained that he stretches his images with APP, because he feels it 'gives a smoother, better tonal, a seemingly larger dynamic range, result'. I think, the initial stretch is very important and it should not be up to later following modules like Super Structure to catch things up. And I'm not sure, if Detector Gamma is the way to go. Of course all this is difficult to assess and personal preferences and personal perceiving do play a role. I wanted to add a constructive contribution to this discussion with my stretch-curve-analyzer project and I still think it could be helpful, but I can't find enough time to drive things forward. :( OptiDev is a great tool and very important as it plays a central and indispensable (!) role within ST's self-contained signal-evolution-tracking workflow. And therefore I think this discussion is important, too. (I hope this doesn't sound like 'complaining'. It really isn't meant this way.)

:confusion-shrug:

Dietmar.
Mike in Rancho
Posts: 1158
Joined: Sun Jun 20, 2021 10:05 pm
Location: Alta Loma, CA

Re: presentation questions

Post by Mike in Rancho »

decay wrote: Sat Jun 01, 2024 7:14 pm I'm afraid, we're not done with this yet. :cry: Stefan (@Stefan B) and I felt independently of each other the need to brighten up our small galaxies using Super Structure's brighten preset, because it was not possible to get them bright enough just using OptiDev. And lately Andy (@AndyBooth) explained that he stretches his images with APP, because he feels it 'gives a smoother, better tonal, a seemingly larger dynamic range, result'. I think, the initial stretch is very important and it should not be up to later following modules like Super Structure to catch things up. And I'm not sure, if Detector Gamma is the way to go. Of course all this is difficult to assess and personal preferences and personal perceiving do play a role. I wanted to add a constructive contribution to this discussion with my stretch-curve-analyzer project and I still think it could be helpful, but I can't find enough time to drive things forward. :( OptiDev is a great tool and very important as it plays a central and indispensable (!) role within ST's self-contained signal-evolution-tracking workflow. And therefore I think this discussion is important, too. (I hope this doesn't sound like 'complaining'. It really isn't meant this way.)

:confusion-shrug:

Dietmar.
Hi Dietmar,

Yes I think I remember that from Andy, but thought it was before the latest OptiDev fix? I agreed with his sentiment, but was of the opinion that old OptiDev-AutoDev was causing some of that harshness, and again due to the stellar profile problem as we have found. But I'd have to go back to old threads to be sure. :confusion-shrug:

Unfortunately it's been a busy year. I want to get back to the OptiDev experiments and analysis, but I also haven't even had time for capturing new data - despite plenty of clear SoCal nights. :(

One thing to perhaps keep in mind though is that, despite the results of old OptiDev often seeming well-stretched for detail and contrast, we now know it was oft caused by erroneous star data rather than the DSO data. Lucky coincidence? Or maybe just those "shoulders" of bright stars tend to fall within ADU ranges that we would want contrast in for most DSO anyway? :think:

Until we can get back to the testing, I agree that Detector Gamma may not be enough to get where we want to be. I wonder if a third control to add pedestal (such as SkyGlow in FilmDev) would help. But if so, should it be applied before or after the hand-off to the OptiDev optimal stretch algorithm, or an option for either? I would try to test those, but I don't know if ST would let me bounce around between OD and FD, particularly within tracking. Maybe I can try some stuff outside tracking or turn it on and off. :confusion-shrug:
decay
Posts: 456
Joined: Sat Apr 10, 2021 12:28 pm
Location: Germany, NRW

Re: presentation questions

Post by decay »

Mike in Rancho wrote: Sat Jun 01, 2024 8:18 pm I think I remember that from Andy, but thought it was before the latest OptiDev fix? I agreed with his sentiment, but was of the opinion that old OptiDev-AutoDev was causing some of that harshness, and again due to the stellar profile problem as we have found. But I'd have to go back to old threads to be sure.
You are right, I too think, it was before the last fix, Mike. But I can't remember that we supposed the fix could be a solution for Andy's issues. :think: Will have to re-read, too. But I remember one of my ideas was to compare the stretching curves produced by APP and by ST.
Mike in Rancho wrote: Sat Jun 01, 2024 8:18 pm One thing to perhaps keep in mind though is that, despite the results of old OptiDev often seeming well-stretched for detail and contrast, we now know it was oft caused by erroneous star data rather than the DSO data. Lucky coincidence? Or maybe just those "shoulders" of bright stars tend to fall within ADU ranges that we would want contrast in for most DSO anyway?
Yes, maybe. Who knows? This is something we could try to analyse. And I still haven't re-read the thread and Ivo's explanations. I remember that I didn't understand them fully at that time. :(

And I thought of comparing stretching curves of ST 8 and ST 9 (pre fix). And FilmDev. And ... and .. and ... :lol:

Until then we maybe should keep all this in mind and have a look on how well the stretching of our next images will work.

Best regards, Dietmar.
User avatar
admin
Site Admin
Posts: 3374
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: presentation questions

Post by admin »

I added a "Brightness Tranches" setting in AutoDev to the latest beta. You can now choose which precision is most suitable for your dataset.

It specifies the granularity with which tranches (e.g. brightness continuums) are isolated for curve construction. As a general rule of thumb, start off by matching this setting with the originating camera's sensor bit-depth. This setting impacts mostly how aberrant discontinuities due to non-linearities (such as stars and other strong highlights) are allocated dynamic range in lieu of other detail. Lower settings progressively ignore the discontinuities in these areas for curve construction, while higher settings are careful to allocate more dynamic range to provide smoother transitions to these areas (in lieu of showing other detail).

It defaults back to 12-bit, which works for most datasets and tends to be a good compromise between showing good detail in the shadows and midtones and ignoring "weird" non-linearities (by lobbing them all in the same few tranches). However, if you need the sensitivity, you can now choose to force it to be very careful in those areas, and have it allocate more dynamic range by allocating more tranches to these areas.
Ivo Jager
StarTools creator and astronomy enthusiast
decay
Posts: 456
Joined: Sat Apr 10, 2021 12:28 pm
Location: Germany, NRW

Re: presentation questions

Post by decay »

Thank you much for paying attention to this, Ivo.
I will reprocess my M63 and get back here soon.

Dietmar.
decay
Posts: 456
Joined: Sat Apr 10, 2021 12:28 pm
Location: Germany, NRW

Re: presentation questions

Post by decay »

Here are some impressions on how the new "Brightness Tranches" setting works. Processing was: Bin 71% preset, Crop, Wipe Basic preset but aggressiveness boosted up to 96% because of some gradients. Then OptiDev, IFD about 6 px and ROI as shown. "Brightness Tranches" setting increased from 10 up to 15 bit, left to right. In the first series the bright star was included into the ROI, second series without star:

w-star.jpg
w-star.jpg (391.77 KiB) Viewed 737 times
wo-star.jpg
wo-star.jpg (499.55 KiB) Viewed 737 times
(Best viewed at 100%, then scroll left to right.)

The Brightness Profile Viewer shows the profile of the bight star above its window. At low "Brightness Tranches" values the "Ron-Effect" is clearly visible: Appearing as plateau / shoulder in the brightness profile and as a ring around the star. With increasing bit-depth setting / curve resolution the shoulder moves up and the ring disappears. My EOS 2000D has a 14 bit ADC and indeed at this bit-depth setting the star profile looks quite nice :bow-yellow: Very interesting :think: . And very cool. :)

As expected the overall brightness in the mid-tones (the galaxy disc) decreases at higher bit-depth settings. So for datasets affected of these non-linearities (mine are obviously affected) we now have a means to get decent stellar profiles.

I wonder how to get some brightness / dynamic range into the mid-tones. The galaxy is now much to dim I guess? Is levering the "Detector gamma" the way to go in this case? :think:

Dietmar.
Post Reply