Help with M3 for a StarTools Noob

[smartz1]

Hello all. I'm a new StarTools user (coming from PI). I must say, I am really impressed so far. I'm currently spreading the word with my AP friends and am hopeful they will adopt it as well. Up to this point, I've just been trying to get used to the workflow and some of the basics for the modules. I know I have a ton to learn

Having said that, I was hoping to get some advice and any pointers people may want to share. I have a ton of data on my backlog that I want to process. Given what I've seen, I very likely will re-process at least some of my previously processed data with ST. But to begin with, I decided to start with a pretty simple data set of M3.

This data was acquired with an Officina Stellare RH200 with a Trius 814X. We captured a bit more Green data than is used in my image, but several of the exposures had some hideous moonglow so I pitched them for stacking. I stacked in PI with normalization turned off and used the recommended Winsorized Sigma Clipping. We typically do not capture Lum so I created a Syn Lum in PI. I understand that ST can interpolate the Lum, but I didn't want to get too fancy on my first real go at processing, lol. If doing this could bugger anything up, please let me know The totals that were used in the image were (in hours and minutes; 10 min subs):

Red - 4:10
Green - 2:50
Blue - 4:30

https://www.dropbox.com/s/qdk4q5bxt9hnm ... .tiff?dl=0

I have several questions, so I guess I will just shoot and hope someone can help me with them. They are below:

1. Is there any issues with me creating the Lum rather than just having ST interpolate it?
2. Assuming question 1 is ok, what would I use for the exposure time? In this image I used the longest exposure time listed.
3. What can I do to ameliorate the orange stars with the really blown out cores? I tried using the heal module, but did not really get too far with it. Probably more from ignorance that anything else as it is the first "heal" type processing that I've used.
4. Why do the stars have seem to have light x's in their cores wshen zoomed in? I know the obvious answer is don't zoom in so far, but just curious
5. Are there any other tips/tricks you can share to up my game in ST? I plan on making time to sit around and really play with each module in depth to see that they do. I've read many of the guides, the manual, etc. But I am pretty hands on when it comes to learning.

I'll include my log so you can see how I got to here anyway. Maybe you guys can also tell me things I could do better workflow wise. Any and all help is greatly appreciated!

Code: Select all

-----------------------------------------------------------
StarTools 1.7.461
Mon Aug 02 21:24:32 2021
-----------------------------------------------------------
Loading luminance channel data
File loaded in LRGB module [F:\M3 in Startools\Master Lum.fit].
Loading red channel data
File loaded in LRGB module [F:\M3 in Startools\Master Red_r.fit].
Loading green channel data
File loaded in LRGB module [F:\M3 in Startools\Master Green_r.fit].
Loading blue channel data
File loaded in LRGB module [F:\M3 in Startools\Master Blue_r.fit].
--- Compose
Parameter [Luminance, Color] set to [L + Synthetic L From RGB, RGB]
Parameter [Color Ch. Interpolation] set to [On]
Parameter [Lum Total Exposure] set to [4h30m (270m) (16200s)]
Parameter [Blue Total Exposure] set to [4h30m (270m) (16200s)]
Parameter [Green Total Exposure] set to [2h50m (170m) (10200s)]
Parameter [Red Total Exposure] set to [4h10m (250m) (15000s)]
Image size is 6776 x 5424
Type of Data: Linear, was not Bayered, or was Bayered + white balanced
--- Bin
Parameter [Scale] set to [scale 50.00% / +2.00 bits / +1.00x SNR improvement]
Image size is 3388 x 2712
--- Crop
Parameter [X1] set to [47 pixels]
Parameter [Y1] set to [63 pixels]
Parameter [X2] set to [3322 pixels (-66)]
Parameter [Y2] set to [2648 pixels (-64)]
Image size is 3275 x 2585
--- Wipe
Parameter [Synthetic Dark/Bias] set to [Off]
Parameter [Gradient Edge Behavior] set to [Absorb 50%]
Parameter [Synthetic Flats] set to [Off]
Parameter [Sampling Precision] set to [256 x 256 pixels]
Parameter [Dark Anomaly Filter] set to [1 pixels]
Parameter [Gradient Falloff] set to [0 %]
Parameter [Synth. Bias Edge Area] set to [100 %]
Parameter [Gradient Aggressiveness] set to [75 %]
--- Auto Develop
Parameter [Ignore Fine Detail <] set to [Off]
Parameter [Outside RoI Influence] set to [15 %]
Parameter [RoI X1] set to [1328 pixels]
Parameter [RoI Y1] set to [1004 pixels]
Parameter [RoI X2] set to [1941 pixels (-1334)]
Parameter [RoI Y2] set to [1549 pixels (-1036)]
Parameter [Detector Gamma] set to [1.00]
Parameter [Shadow Linearity] set to [50 %]
--- Contrast
Parameter [Expose Dark Areas] set to [Yes]
Parameter [Brightness Retention] set to [Off]
Parameter [Precision] set to [256 x 256 pixels]
Parameter [Shadow Detail Size] set to [10 pixels]
Parameter [Locality] set to [50 %]
Parameter [Shadow Dyn Range Alloc] set to [50 %]
--- HDR
Parameter [Small Detail Precision] set to [Max]
Parameter [Channels] set to [Brightness Only]
Parameter [Algorithm] set to [Reveal DSO Core]
Parameter [Dark/Bright Response] set to [Full]
Parameter [Detail Size Range] set to [40 pixels]
Parameter [Strength] set to [1.2]
--- Wavelet Sharpen
Parameter [Structure Size] set to [Large]
Mask used (BASE64 PNG encoded)

--- SNR-aware Wavelet Sharpening
Parameter [Scale 1] set to [100 %]
Parameter [Scale 2] set to [100 %]
Parameter [Scale 3] set to [100 %]
Parameter [Scale 4] set to [100 %]
Parameter [Scale 5] set to [100 %]
Parameter [Mask Fuzz] set to [4 pixels]
Parameter [Amount] set to [300 %]
Parameter [High SNR Size Bias] set to [85 %]
Parameter [Low SNR Size Bias] set to [0 %]
Parameter [Dark/Light Enhance] set to [50% / 50%]
Mask used (BASE64 PNG encoded)

--- Deconvolution
Parameter [Image Type] set to [Deep Space]
Parameter [Secondary PSF] set to [Off (Primary Only)]
Parameter [Primary PSF] set to [Moffat Beta=4.765 (Trujillo)]
Parameter [Primary Radius] set to [2.0 pixels]
Parameter [Iterations] set to [11]
Parameter [Error Diffusion] set to [49 %]
Parameter [Mask Fuzz] set to [8.0 pixels]
Parameter [Deringing] set to [50 %]
--- Fractal Flux: Setup
Parameter [Wavelet Library] set to [Small]
--- Fractal Flux
Parameter [Brightness Mask Mode] set to [Off]
Parameter [Algorithm] set to [Add Detail]
Parameter [Negative Flux] set to [0 %]
Parameter [Positive Flux] set to [25 %]
Parameter [Detail Filter] set to [2.5 pixels]
Parameter [Filter Fuzz] set to [1.0 pixels]
Parameter [Filter Radius] set to [1.5 pixels]
Parameter [Filter Amount] set to [200 %]
Parameter [Mask Fuzz] set to [1.0 pixels]
Parameter [Brightness Mask Power] set to [1.0]
--- Color
Parameter [Bias Slider Mode] set to [Sliders Reduce Color Bias]
Parameter [Style] set to [Scientific (Color Constancy)]
Parameter [LRGB Method Emulation] set to [Straight CIELab Luminance Retention]
Parameter [Matrix] set to [Identity (OFF)]
Parameter [Dark Saturation] set to [2.0]
Parameter [Bright Saturation] set to [Full]
Parameter [Saturation Amount] set to [200 %]
Parameter [Blue Bias Reduce] set to [1.18]
Parameter [Green Bias Reduce] set to [1.12]
Parameter [Red Bias Reduce] set to [1.00]
Parameter [Mask Fuzz] set to [1.0 pixels]
Parameter [Cap Green] set to [0 %]
Parameter [Highlight Repair] set to [Off]
Mask used (BASE64 PNG encoded)

--- Filter
Parameter [Filter Mode] set to [Conservative Nudge]
Parameter [Sampling Method] set to [3x3 Average]
Parameter [Filter Width] set to [5]
Parameter [Mask Fuzz] set to [1.0 pixels]
--- Entropy-driven Detail Enhancement
Parameter [Resolution] set to [High]
Parameter [Channel Selection] set to [All]
Parameter [Strength] set to [100 %]
Parameter [Midtone Pull Filter] set to [20.0 pixels]
Parameter [Midtone Pull Strength] set to [50 %]
Parameter [Dark/Light Enhance] set to [50% / 50%]
--- Super Structure
Parameter [Detail Preservation] set to [Min Distance to 1/2 Unity]
Parameter [Compositing Algorithm] set to [Power of Inverse]
Parameter [Brightness, Color] set to [Process Both]
Parameter [Brightness Retention] set to [Off]
Parameter [Mask Fuzz] set to [1.0 pixels]
Parameter [Airy Disk Radius] set to [50 %]
Parameter [Gamma] set to [0.50]
Parameter [Detail Preservation Radius] set to [40.0 pixels]
Parameter [Saturation] set to [100 %]
Parameter [Strength] set to [75 %]
--- Unified De-Noise
Parameter [Grain Size] set to [7.0 pixels]
Parameter [Walking Noise Size] set to [1.0 pixels]
Parameter [Walking Noise Angle] set to [0]
--- Unified De-Noise
Parameter [Scale 1] set to [95 %]
Parameter [Scale 2] set to [95 %]
Parameter [Scale 3] set to [95 %]
Parameter [Scale 4] set to [95 %]
Parameter [Scale 5] set to [50 %]
Parameter [Grain Equalization] set to [0 %]
Parameter [Scale Correlation] set to [50 %]
Parameter [Color Detail Loss] set to [50 %]
Parameter [Brightness Detail Loss] set to [50 %]
Parameter [Grain Dispersion] set to [7.0 pixels]
Mask used (BASE64 PNG encoded)

--- Heal
Parameter [Quality] set to [Medium]
Parameter [New Darker Than Old] set to [No]
Parameter [Grow Mask] set to [0 pixels]
Parameter [Neighbourhood Samples] set to [0]
Parameter [New Must Be Darker Than] set to [Off]
Parameter [Neighbourhood Area] set to [200 pixels]
Undo
Image size is 3275 x 2585
File saved [F:\M3 in Startools\M3 LLRGB.tiff].

I'm sure I'll have plenty more (hopefully not too stupid) questions. Thanks in advance for any and all help! If any clarification is needed on anything I've posted, I am happy to do so.

Clear skies!

[smartz1]

Sorry, I meant to post the image and not just a link:

[Mike in Rancho]

Hi smartz1,

Well for a 50MB tiff, I was expecting to be able to zoom in deep and see some real clarity, but something seems to have gone sideways.

Also your second post/link requires google login - I didn't do so but am guessing it may also not be a public share?

I'll let Ivo or others discuss your compose module use and integration with PI as the stacker/pre-processor. For globular clusters I really don't tend to throw a lot of modules at it. Depends on the data and what seems necessary, but I'll often skip contrast and sharpening.

I noted no use of ignore fine detail in autodev, but depending on your ROI it may have been okay anyway. For HDR on a glob I like using the Reveal Core preset and adjusting strength from there to taste. This will tame down a bright cluster core and allow deeper detail and color.

Pretty much just deconvolution then, adjusting radius to clarify detail but avoid ringing and artifacts. For color, did you use star sampling to set the balance? (If not the auto balance upon entry is usually pretty good too). The golden orange, blue, and white seem reasonable to me, though I did not consult a reference image.

I'm not sure what you nudged in the Filter module? Also I haven't really used Flux or Entropy so cannot speak to those.

If you can link your data (personally I would probably just use the R, G, and B if otherwise unmolested by PI) we might be able to walk through them and evaluate what happened and if it is from the stacks or something done after).

But pending that - what I do for oversaturated or blown out star cores is this: Use the Mask module to select the star core(s) in question. Up to you to decide how much is needed or how much to include, if any, of the outer halo coloring. You'll get the hang of it. Then I enter the Layer module. Zoom/scroll so you can see one of your selected stars. Increase the kernel radius setting bit by bit, and if needed the mask fuzz. In the third (results/blended) panel you should see the harsh core start to blur over and soften. Just a little bit is usually all that's needed, though it depends just how badly the star is blown out. Differently-sized stars may need different levels of this, so it may take a couple go-throughs.

It's also a good method of taming an overly-bright galactic central core - common for M31, for example.

The rest, we'll have to see. I'm not good enough to tell what's going on from just the final image and the workflow. But I'm sure other assistance will pour in.

Good luck!

[smartz1]

Sorry for the difficulties with the image. I admit I am more of a forum lurker than poster, lol. I'll post dropbox links to the data when I get home from work. Thanks for taking a look!

[smartz1]

As promised, the links to the data are below.

Blue - https://www.dropbox.com/s/n58ihv703g3bp ... r.fit?dl=0
Green - https://www.dropbox.com/s/m3m3bggc1gdo3 ... r.fit?dl=0
Lum - https://www.dropbox.com/s/yskmt8ivnjmd7 ... m.fit?dl=0
Red - https://www.dropbox.com/s/03pnh8nlgp55k ... r.fit?dl=0

Please let me know if I screwed anything up when linking them. If so, I will endeavor to fix it.

Thanks!

[admin]

Hello all. I'm a new StarTools user (coming from PI). I must say, I am really impressed so far. I'm currently spreading the word with my AP friends and am hopeful they will adopt it as well. Up to this point, I've just been trying to get used to the workflow and some of the basics for the modules. I know I have a ton to learn

Hi, and welcome! I hope you will enjoy the improvements StarTools brings to your images.

I understand that ST can interpolate the Lum, but I didn't want to get too fancy on my first real go at processing, lol. If doing this could bugger anything up, please let me know The totals that were used in the image were (in hours and minutes; 10 min subs):

Red - 4:10
Green - 2:50
Blue - 4:30

You can indeed create your own synthetic luminance if you wish, though it's actually easier in StarTools.

If you have your own luminance you wish to use in the Compose module, just set "Luminance, Color" to "L, RGB" (not "L + Synthetic L From RGB, RGB", otherwise StarTools will add RGB to that again!). This willl purely use your own luminance for detail, while using any RGB data you import purely for the coloring. You don't have to bother with the Exposure times.

The easier option, however, is to let StarTools create the the synthetic luminance for you from the RGB data. Just import the data in the relevant channels, and set the exposure times. Then leave "Luminance, Color" at "L + Synthetic L From RGB, RGB".

1. Is there any issues with me creating the Lum rather than just having ST interpolate it?

If you know what StarTools does with it, and what you should be doing in PI to create a synthetic luminance, not a problem at all!

2. Assuming question 1 is ok, what would I use for the exposure time? In this image I used the longest exposure time listed.

Exposure time is ignored if you provide your own synthetic luminance and "Luminance, Color" is set to "L, RGB". Exposure time is only used to calculate the proper weighting for any synthetic luminance StarTools creates from your R, G and B data.

3. What can I do to ameliorate the orange stars with the really blown out cores? I tried using the heal module, but did not really get too far with it. Probably more from ignorance that anything else as it is the first "heal" type processing that I've used.

What sort of issues are you trying to correct specifically?

I don't see any particular issues with the stars (particularly once deconvolved) in the data, except that they do have some rather odd stellar profiles at full resolution.

E.g. a quick processing run-through;

NewComposite.jpg (314.67 KiB) Viewed 4696 times

...or with the "Legacy" preset (e.g. as apps like PixInsight would typically render coloring);

NewComposite2.jpg (309.96 KiB) Viewed 4696 times

Is the issue you are trying to correct still visible here?

4. Why do the stars have seem to have light x's in their cores wshen zoomed in? I know the obvious answer is don't zoom in so far, but just curious

Exactly what I meant! This appears to be a "feature" of your optics...

5. Are there any other tips/tricks you can share to up my game in ST? I plan on making time to sit around and really play with each module in depth to see that they do. I've read many of the guides, the manual, etc. But I am pretty hands on when it comes to learning.

Wipe: There appears to be a modest amount of vignetting present, the Vignetting preset will nicely flatten the background a bit better in this case.

HDR: For this particular object (and any other clusters that have a core with tightly packed stars), in the HDR module, try Reveal Core with a small Detail Size Range. This will resolve the core better. Deconvolution will do the rest.

Flux: not typically used/needed in a workflow. Its premise is using self-similarity at different scales (the idea behind good old fractals!) to detect features and manipulate the image that way. This tends to work best in big, image-filling areas of nebulosity (gas flows and filaments tend to behave the same over large scales, from the small clouds here on earth to massive multi-lightyear clouds of gas in outer space), but not so much for stars (which look different at different scales).

Entropy: also not typically used on star clusters, but rather used on images that show large, contiguous areas of nebulosity in one specific channel. For example, it usually does a great job at enhancing the deep red dust-filtered hydrogen emissions in visual spectrum renditions of the Horse Head/Flame nebula complex. Similarly, it can work quite well on narrowband data, which also usually shows good, big differences between the three channels.

Heal: not typically used unless you have distracting defects in your image or wish to create starless images (many times faster than StarNet++ and much more control!).

Hope this helps,

[Mike in Rancho]

Hey Smartz1,

Ivo did a great job already, of course, but I did give your data a try earlier tonight.

I checked the individual fits files first. No crosses but many stars had odd, even blocky shapes to them. Maybe a guiding/tracking thing? In compose, my laptop/ST chewed on the files for a while, odd as they were only ~150MB each. I only used the RGB and set the exposure times per your post above.

After crop and bin to about 2.3K wide, in Wipe things were looking like they maybe weren't calibrated. Were they? Regardless the uncalibrated 1 preset worked fine. Then, using 1.8.506a, I added some correlation filtering. AutoDev, HDR Reveal Core - which was a bit harder here than other glob cores I've done, deconvolution, and color. In color I used a star mask with the glob unselected and hit sample. Then denoise.

And really that's it! Pretty quick. Because of the blocky star shapes and the bin down I did on the scale, some of the stars were too square for my taste. So I gave their cores a slight blur in the Layer module to soften that up. I also used both Layer and Color to tame that one big orange star, though I may have overdone it.

Anyway, I really didn't end up with any major problem with the crossed star shapes in the final image, even though the original fits files do have a hint of that around the stars.

smartz1 m3 st8 1a.jpg (448.77 KiB) Viewed 4689 times

[smartz1]

Thank you guys for all the help! Ivo, your explanations make a lot of sense. Thank you for those. I really need to get out of the mindset of using a linear workflow. But screwing up is part of learning I imagine, lol.

The issue I was referring to was the orange stars that are pure white in the center going out to an orange halo. Kinda what I was looking for was what Mike did to the big orange star. It really softens the transition and makes it look less like a donut with a white center if that makes sense.

To answer your question Mike, yes they were calibrated with Flats, Darks, and Bias frames. However, we did have an issue with SkyX flipping the images in the red channel in relation to the flats (i.e. the vignetting profile was reversed). I thought we went through and corrected all the affected frames, but we may have missed some. Having said that, I will try using the uncalibrated 1 profile and see what I get.

A couple follow-up questions for Mike. What is the correlation filtering you refer to? Also, you said you tamed the big orange star with layer and color. That's exactly the type of thing I was interested in for the ones with blown out cores. Could you perhaps elaborate on your technique for that?

I'll probably take another go at it tonight and see what I can do with this help. Thanks a ton guys.

[Mike in Rancho]

Correlation filter is a new function in the Wipe module of the 1.8 alpha. Probably a better explanation of it on the linked development module description page, but I in essence you can filter out "detail" that you know must be noise or not real - because their pixel scale is below your ability to resolve that finely. Or something like that.

There's several ways to tame stars, and I probably use them all depending on what is needed. For example you can use the Filter module with a star/star-halo mask and the Fringe Killer setting, then click on colors until you are happy. It can be done globally too with a full mask.

A little layer blur can soften hard edges of cores, or blend in some halo color, it just depends what is needed and how you create your mask.

For your big orange star though I thought some extra work was needed. For one, there is a very large outer halo along with the harsh white central core and an orange donut in between. Fringe killer actually does a decent job but I didn't want to lose as much color in the extended halo as that caused. Lastly, in the version of your M3 I posted above I think I oversoftened this star way too much, so here's a second try at it.

This mask was created using the default lighter pixels and clicking far outside the star. I then inverted it to select the smaller stars the same way, then inverted again.

Star and halo mask select.jpg (22.66 KiB) Viewed 4642 times

Then in Color I zeroed out the bright saturation but kept dark saturation high. This tamed some of the donut but maintained a nice hue throughout the extended halo.

Star halo desat.jpg (67.24 KiB) Viewed 4642 times

Then I made a new mask, pretty much just for the donut. I left the central core alone this time as I wanted the star to maintain it's bright pop - I'm just looking to soften the hard edged lines, harshness, and unsightly donutting here that doesn't really make sense (to me) for a star core and radiating halo.

Star core mask select.jpg (15.1 KiB) Viewed 4642 times

Now taking that into the Layer module (you'll need to learn the 3 panels and so on but far right is the end result) and adding the amount that looked good to kernel radius and mask settings, softens it just a bit to this-

(Place image here? Uh oh, there seems to be a limit!)

Oh I also added a couple of the surrounding stars to the mask also because they were kind of squared off, but just for this screenshot lol. In reality I would do all the little star defects separately.

Just a matter of practice and trying things out. And only do the minimum amount needed to address a defect or artifact. As you can see in the overblurred version I did at first, the star is beginning to look too out of focus.

[Mike in Rancho]

The missing fourth image! Anyway pretend this is up above where I was discussing the layer module.

Star core-donut fuzz.jpg (61.48 KiB) Viewed 4642 times

Also the difference in the third panel may be hard to see in this scaled jpg. Perhaps I scaled too much. But if you try it you will see.