-
- Elephant_trunk_AlpT_209x300s_basic-HOO.jpg (692.4 KiB) Viewed 14182 times
-
- Bubble_80x300s_AlpT-only.jpg (688.7 KiB) Viewed 14182 times
-
- Crescent_2023_AlpT_228x300s_v2_2788x1700.jpg (685.8 KiB) Viewed 14182 times
3 images in progress
3 images in progress
1. Elephant's Trunk @420mm. This is duoband only - about 17 hours worth. I have ~2 hours of RGB, but getting a mask just right to layer in the stars is difficult. Especially with the Garnet Star doing its thing. It's a very interesting star, for many reasons. I've looked at a bunch of astrobin IC 1396 images, and it's common for people to shrink the heck out of the Garnet Star. But some versions show the same sort of "glow" that this does (and Freddy's, too). One of its interesting features is that it has a shell that extends to a radius of a bit over 6 arc-seconds as viewed from Earth. The shell apparently emits fairly strong Sodium lines, which I think are orangish-yellowish. A 12 arc-sec diameter circle in my image would be ~15 pixels without binning, so there's no way I can just claim that all the glow is just that shell.
2. The Bubble Nebula plus M52 at 910mm (AT130EDT). This is ~6 hours - I hope to add another 6 plus RGB for stars. I purposely pushed some of the fainter Ha signal right down into the background. I know we are "supposed to" push our data hard to "bring out" nebulosity, but a) my data don't support that yet; and b) I rather like having the Bubble more isolated.
3. The Crescent Nebula at 420mm. This is also about an 18-hour integration with the AlpT duoband filter, and it also is waiting on gracefully layering in RGB stars. (hints and tricks to getting the mask just right would be greatly appreciated!). The Soap Bubble shows up pretty well without having done any dodgy masking to process it differently. It actually shows up better in grayscale.
2. The Bubble Nebula plus M52 at 910mm (AT130EDT). This is ~6 hours - I hope to add another 6 plus RGB for stars. I purposely pushed some of the fainter Ha signal right down into the background. I know we are "supposed to" push our data hard to "bring out" nebulosity, but a) my data don't support that yet; and b) I rather like having the Bubble more isolated.
3. The Crescent Nebula at 420mm. This is also about an 18-hour integration with the AlpT duoband filter, and it also is waiting on gracefully layering in RGB stars. (hints and tricks to getting the mask just right would be greatly appreciated!). The Soap Bubble shows up pretty well without having done any dodgy masking to process it differently. It actually shows up better in grayscale.
Re: 3 images in progress
Here's a crop of the Soap Bubble before the Color module
-
- soap_bubble_at_SVD.jpg (152.3 KiB) Viewed 14178 times
-
Mike in Rancho
- Posts: 1153
- Joined: Sun Jun 20, 2021 10:05 pm
- Location: Alta Loma, CA
Re: 3 images in progress
Nice work. Those all look to be the appropriate and expected results for HOO. Will be interesting to see what they look like if the RGB stars gets accomplished.
I have no tips there, never got it acceptable the few times I tried. Good mask, color of fg, some mask fuzz. But when there's nebulosity across the field and/or many tiny stars that may be missed by automask settings, it's really a daunting task.
Good that you've been so prolific. Between work, weather, and my patio project, I'm pretty much 0-for-September. Oh and the moon too, can't forget that.
But you did inspire me to go back to my "weirdo" HOO Crescent data from June 2022 and try it out in 559 beta. Originally 1.8. But I think it only looks good (though that's debatable too) at full scale, not downsampled other than the original bin at the beginning of processing. So I'll try to upload it to a-bin tonight after work and maybe you can see how it compares. My FOV is really close to what yours is. No RGB stars, it's just straight HOO, but I do think I got some life and decent-enough color into the star field anyway, despite minimal module usage. It's really just AutoDev, HDR, SVD, Color, and Denoise. I embraced the very busy stellar region and refrained from SS and shrink.
And oh yeah my soap bubble turned out the same. Visible in grayscale, but HOO requires nose-to-the-monitor to find it.
I have no tips there, never got it acceptable the few times I tried. Good mask, color of fg, some mask fuzz. But when there's nebulosity across the field and/or many tiny stars that may be missed by automask settings, it's really a daunting task.
Good that you've been so prolific. Between work, weather, and my patio project, I'm pretty much 0-for-September. Oh and the moon too, can't forget that.
But you did inspire me to go back to my "weirdo" HOO Crescent data from June 2022 and try it out in 559 beta. Originally 1.8. But I think it only looks good (though that's debatable too) at full scale, not downsampled other than the original bin at the beginning of processing. So I'll try to upload it to a-bin tonight after work and maybe you can see how it compares. My FOV is really close to what yours is. No RGB stars, it's just straight HOO, but I do think I got some life and decent-enough color into the star field anyway, despite minimal module usage. It's really just AutoDev, HDR, SVD, Color, and Denoise. I embraced the very busy stellar region and refrained from SS and shrink.
And oh yeah my soap bubble turned out the same. Visible in grayscale, but HOO requires nose-to-the-monitor to find it.
Re: 3 images in progress
Good work,
about the soap bubble, I would say still needs more exposure, but you can create a circular mask ---> then auto --> substract from old, and you get this
After go Color Module, keep the mask and click "Sample", the resulting color is pretty much the same color as the Oiii shell of the crescent, as it should be. You can still tweak a bit, but with sample, Startools "knows" better the balance in that region, that it's a bit mixed. Speciallt soften the Mask (sorry, got it in Spanish to check if there's missing translations to be done ^^; ) but you know where.
Obviously this is from the jpeg you posted here.. results would be better with original dataset.
hope it helps,
CS,
Carles
about the soap bubble, I would say still needs more exposure, but you can create a circular mask ---> then auto --> substract from old, and you get this
-
- mask.jpg (317.41 KiB) Viewed 14107 times
-
- color sample.jpg (115.67 KiB) Viewed 14107 times
hope it helps,
CS,
Carles
Re: 3 images in progress
Thanks, Carles! Yes, of course more time... Though at 18 hours already it may be that "darker skies" is the more viable solution. I assume there's a good reason the thing wasn't noticed until 2008.
Interesting note about re-sampling for color with just the masked Soap Bubble. I'm not normally inclined to use masks to process specific parts of an image just to get the result I "want" - but you can be sure that I will try your trick if only to see how it looks.
Interesting note about re-sampling for color with just the masked Soap Bubble. I'm not normally inclined to use masks to process specific parts of an image just to get the result I "want" - but you can be sure that I will try your trick if only to see how it looks.
-
Mike in Rancho
- Posts: 1153
- Joined: Sun Jun 20, 2021 10:05 pm
- Location: Alta Loma, CA
Re: 3 images in progress
Looks kind of like a green Death Star. That's no moon! 
I'm not sure what viable (to me) way there would be to boost the soap bubble. Throttling Ha in an HOO, at least to any fair extent, often doesn't seem to look so good in my eyes as it just fades the red too much. And the teal often fogs.
I suppose maybe just a great deal of extra OIII integration, probably by mono and discrete filter, which would result in extra SNR and with the exposure settings in compose, it would be weighted accordingly.
I'm not sure what viable (to me) way there would be to boost the soap bubble. Throttling Ha in an HOO, at least to any fair extent, often doesn't seem to look so good in my eyes as it just fades the red too much. And the teal often fogs.
I suppose maybe just a great deal of extra OIII integration, probably by mono and discrete filter, which would result in extra SNR and with the exposure settings in compose, it would be weighted accordingly.
Re: 3 images in progress
The images to follow are just screen grabs. Carles' suggestion certainly works.
It gave me some ideas for experimentation - in the versions I'm posting I started by sampling just a mask around the Crescent itself and applying that to the rest of the image. Not tremendously different from the full-frame sampling, though the Ha gets more throttled.
As long as I felt like I was already just painting a Soap Bubble, I jacked up the saturation in the masked region:
Now we're definitely painting!
OK - what happens if I make a circle mask in the most red (well, pink) Ha region (lower left of the full image). I thought maybe it would look more red.
But...
Well that was unexpected - though completely consistent with the "I'm painting a Soap Bubble into my image" line of thought.
Soooo - what to do? Will I paint a blue bubble in the image I eventually share with friends and family? Maybe - it does a better job of telling the story of the Soap Bubble nebula.
By the way - this is as good a time and place as any to ask: Am I the only one who sees "StarTools Ha/Oiii from duoband" coloring as a fairly unattractive blue-grey and pink? I guess I can see the logic on the Ha side, as the Hydrogen also has other color lines and doesn't really look deep red. Yet I didn't capture *any* data for the other lines with the duoband filter, so it feels a bit like making up what should be there.
Meanwhil, I'm also playing around with ways to maybe avoid the "blue-grey circle around an rgb star" effect that annoys me when trying to layer in the rgb stars. Running Shrink with Color Taming jacked way up looks to do a good job in the Ha nebulosity regions, though I've not yet actually tried all the way through the Layer module.
-
- Soap_Bubble_sample_more-subtle.jpg (249.27 KiB) Viewed 14010 times
As long as I felt like I was already just painting a Soap Bubble, I jacked up the saturation in the masked region:
-
- Soap_Bubble_sample_overcooked.jpg (249.56 KiB) Viewed 14010 times
OK - what happens if I make a circle mask in the most red (well, pink) Ha region (lower left of the full image). I thought maybe it would look more red.
But...
-
- HA-region_sample.jpg (376.15 KiB) Viewed 14010 times
Soooo - what to do? Will I paint a blue bubble in the image I eventually share with friends and family? Maybe - it does a better job of telling the story of the Soap Bubble nebula.
By the way - this is as good a time and place as any to ask: Am I the only one who sees "StarTools Ha/Oiii from duoband" coloring as a fairly unattractive blue-grey and pink? I guess I can see the logic on the Ha side, as the Hydrogen also has other color lines and doesn't really look deep red. Yet I didn't capture *any* data for the other lines with the duoband filter, so it feels a bit like making up what should be there.
Meanwhil, I'm also playing around with ways to maybe avoid the "blue-grey circle around an rgb star" effect that annoys me when trying to layer in the rgb stars. Running Shrink with Color Taming jacked way up looks to do a good job in the Ha nebulosity regions, though I've not yet actually tried all the way through the Layer module.
-
Mike in Rancho
- Posts: 1153
- Joined: Sun Jun 20, 2021 10:05 pm
- Location: Alta Loma, CA
Re: 3 images in progress
Nice. You've created your own "dynamic" narrowband combination. Now you just need a cool name for it like Foraxx. It's the exact same concept (not that anyone on CN seems to realize it).
I thought I might try something on the bubble using the Crescent data I put up in the other thread, and see if the bands could be processed separately and then Layer blended. This throws away the relative emissions, but perhaps would allow some bubble boost. Alas, the data where I extracted the Ha and OIII is pretty weak, even with the 2xGB composition. Pretty much the entirety of my soap bubble structure comes from the mono cam, even with the same filter. Guess the only lesson is just how much better the 2600MM is than the D5300. But it ought to be.
Might work in your case though with a 571 OSC. If you extract the OIII by itself can you isolate and emphasize the bubble?
On the ever-so-appealing salmon pink and teal, have you tried going to color constancy to see if it becomes more red?
Maybe in 1.10 we can request fully variable and selectable narrowband color mapping.
I was mulling over your RGB star issue.
Is it that the O of the HOO diffraction ball/halo is just outsized compared to the RGB version, for which I presume you have processed a separate stack? If it is just for star-masked layer blending or colorizing, I wonder if when you run the RGB-only workflow you could overstretch the stars. Obviously something that wouldn't be done in the normal course, but maybe just fiddle with OptiDev so as to bloat up the stars and the diffraction ball, thus hopefully becoming a better-matched overlay for the HOO.
It's the exact same concept (not that anyone on CN seems to realize it).I thought I might try something on the bubble using the Crescent data I put up in the other thread, and see if the bands could be processed separately and then Layer blended. This throws away the relative emissions, but perhaps would allow some bubble boost. Alas, the data where I extracted the Ha and OIII is pretty weak, even with the 2xGB composition. Pretty much the entirety of my soap bubble structure comes from the mono cam, even with the same filter. Guess the only lesson is just how much better the 2600MM is than the D5300. But it ought to be.
Might work in your case though with a 571 OSC. If you extract the OIII by itself can you isolate and emphasize the bubble?
On the ever-so-appealing salmon pink and teal, have you tried going to color constancy to see if it becomes more red?
Maybe in 1.10 we can request fully variable and selectable narrowband color mapping.
I was mulling over your RGB star issue.
Is it that the O of the HOO diffraction ball/halo is just outsized compared to the RGB version, for which I presume you have processed a separate stack? If it is just for star-masked layer blending or colorizing, I wonder if when you run the RGB-only workflow you could overstretch the stars. Obviously something that wouldn't be done in the normal course, but maybe just fiddle with OptiDev so as to bloat up the stars and the diffraction ball, thus hopefully becoming a better-matched overlay for the HOO.Re: 3 images in progress
I think I have mostly tamed the rgb stars issue, through a combination of large-scale color-taming in Shrink, some tedious tweaking of certain stars in the mask, 7ish pixels of mask fuzz and using a "blown out" version of the rgb image. At some point I will play around to see if all of those are necessary. The rgb image itself is a thing of ugly:
Here's the composite, for now resized to 1600px wide and shrunk to 700kb in IrfanView. I have settled on the name Conaxx palette. I considered Threeaxx or Fiveaxx, then Againstaxx, which ultimately led to Conaxx, which I think conveys the right impression...
As an aside (or a further-aside), while searching a-bin for an image with an annotation for the lower-right knot of Ha (it is Sh2-104, but Stellarium won't label it for me), I ran across a recent image with 20 hours Ha and 20 hours Oiii, at f/3, from Marathon, TX (so B2, probably?) - and the Soap Bubble is only as distinctly visible as in our basic HOO images.
-
- Crescent_OSC_blowout.jpg (679.44 KiB) Viewed 13961 times
-
- Crescent_AlpT_228x300s_soapy_wOSC_stars_1600.jpg (730.95 KiB) Viewed 13961 times
-
Mike in Rancho
- Posts: 1153
- Joined: Sun Jun 20, 2021 10:05 pm
- Location: Alta Loma, CA
Re: 3 images in progress
Very cool! The RGB stars add a lot of life to the image. 
So based on your findings it seems the soap bubble really ought to be left in a just-barely-visible state. Sort of like the squid in a bat. I already don't trust any squid that's more than the faintest of outlines.
On the Ha knot, I presume you mean the bright one near 36 Cyg? Yeah Stellarium won't give me anything there either unless I check the box for DWB. But searching Sh2-104 gives me a different location, closer to Sadr.
Conaxx is perfect. Subtle mocking with plausible deniability. My preferred Fauxaxx would likely bend too many noses out of shape.
So based on your findings it seems the soap bubble really ought to be left in a just-barely-visible state. Sort of like the squid in a bat. I already don't trust any squid that's more than the faintest of outlines.
On the Ha knot, I presume you mean the bright one near 36 Cyg? Yeah Stellarium won't give me anything there either unless I check the box for DWB. But searching Sh2-104 gives me a different location, closer to Sadr.
Conaxx is perfect. Subtle mocking with plausible deniability. My preferred Fauxaxx would likely bend too many noses out of shape.