Page 1 of 2
Processing Challenge 34 - Astrobee's Wizard
Posted: Wed Oct 12, 2022 11:34 am
by Jockinireland
This months challenge is some great looking data (I've not processed it yet!) from Forum Member
@AstroBee - Greg
Greg has recently invested in a ASI 2600mm and this data comes from that camera. Being the delighted owner of a 2600mc I am excited to see how the mono version of this great sensor works out.
We have
RGB,
Ha and Oiii data here so there are plenty options for combination. Greg provided the following capture details
Explore Scientific ED152 CF
APO
ZWO ASI 2600mmPro Camera
Chroma Filters
iOptron CEM70EC mount
Captured with
NINA
Processed with PixInsight
25 x 2min ea
RGB = 2hrs 30min
15 x 10min
Ha = 2hrs 30min
21 x 10min OIII = 3hrs 30min
Total Integration Time = 8hrs 30min
The image will solve at
FL 1216mm and Pixel size 3.76micron
Files are available at
https://1drv.ms/u/s!Alj6NhuYfZtWgP9bAvo ... g?e=f0VskY
I've put in Gregs original .XISF but also 32 bit .fit and 16 bit .tiff (NB. There was an issue on the 16b tiff in last months challenge but I THINK they are ok this time -if not let me know asap)
As always, download, process and post back in this thread with details of how you did it.
Enjoy the challenge.
Take care
David
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Fri Oct 14, 2022 3:39 am
by starfield
Alrighty. This was a pretty data set to play with. And since the Southwest summer monsoons have decided to return, I may be working thru more of these. Here's my take. I decided to go for a pale red/pinkish hue here to follow what's already there in the RGG. This is an HOO combo with
RGB Stars blended in.
Basic workflow was as follows in
PI
Split my HO the and
RGB into two different workspaces.
For HO, deconvolute both the
HA and OIII, go starless with StarXterminator, and then run NoiseXterminator.
Light Masked Stretch on both channels. Combine with Pixel math
R = "1.075*
Ha"
G = ".80*OIII+.2*
Ha"
B = ".15*
Ha+.85*OIII"
Second stretch with Histogram Transformation
Tweak colors in curve
MLT to slightly sharpen
DBE to neutralize background
For
RGB - Combine, DBE, PCC. Light stretch and clip to get happy balance of stars in Histogram Transformation. Increase color saturation. Run StarXterminator to extract stars with Unscreen option.
Combine Stars with Starless HOO. Run color corrected HDR script. Light local histogram transformation. Final noise reduction pass. Save as Tiff for final tweaking in Photoshop using Camera Raw Tool.
Thanks, Greg, for providing this nice data set.
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Fri Oct 14, 2022 10:29 am
by Jockinireland
Thats really nice @starfield ,
Is there a reason you did not do background extraction on the Ha/Oiii?
I'm working on my attempt just now.
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Fri Oct 14, 2022 2:49 pm
by starfield
I didn't run DBE on the narrorwband since I couldn't see any gradients. Even when super stretched. They may be there but since I couldn't see them I just left it alone. Now that I think about it, I actually did run DBE on the combined HA/OIII starless. This is a trick I learned from an Adam Block video to help neturalize the background when working with a starless image.
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Sat Oct 15, 2022 3:13 am
by STEVE333
I really like Starfield's treatment of this data.
Here is my approach using only HOO stars. I haven't learned how to add
RGB stars to an NB image yet.
PROCESSING STEPS:
Did ABE and NoiseXTerminator on both
Ha and O3 images.
Did mild stretch on
Ha with HistogramTransformation (HT).
Boosted O3 to approximately match
Ha star brightness and background levels.
Removed stars from both.
Stretched O3 further with HT to approximately match brightness and background levels of stretched
Ha image.
Combined
Ha and O3 to form HOO image.
Used HT to align R, G and B histograms to make background gray.
Sharpened HOO image using UnsharpMask.
Increased HOO image contrast using LocalHistogramEqualization (Kernel 26 and 110).
Increased color saturation using CurvesTransformation with "S" option selected.
Combined
Ha and O3 stars-images to create HOO_stars image.
Added HOO_stars image to enhanced HOO image.
Reduced stars using Bill's star reduction.
Thanks AB for sharing your nice data with us.
Steve
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Sat Oct 15, 2022 8:41 am
by Jockinireland
I decided that I wanted to try to use the full data set in one image. So I decided to process the
RGB data and enhance the red with
Ha nd the green blue with oiii.
My process was;
All five images cropped, DBE, deconvolution.
RGB combined with channel combination then PCC followed by Starexterminator, noiseexterminator
HA & Oiii starexterminator, noise extreminator.
all 3 images stretched with HT
HA & Oiii adjusted with HT to reach similar background levels then MLT to sharpen and LHE to increase contrast.
I then used Shawn Neilsons (Visible Dark) process icon to enhance the Red with the
HA - worked fairly well
I cant find a similar process icon to put the Oiii into G&B so split the channels, added the Oiii at 50;50 with pixelmath and recombined the channels (not sure it worked so well
) anyone know the proper way to do this - I see the Light Vortex tutorial on it but it was too complicated for me).
I then threw the
Ha on top as a Lum using LRGB combine (getting the most use out of free data - its a Scottish thing
)
Stretched the
RGB stars and added them back with pixelmath
After that it was a few tweaks in CT on sat and contrast and finished by saving as JPG
This the LHaROiiiGB version
And this the version without the Oii added -
I'm going to try a HOO combination - the with the already porcessed images later today
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Sat Oct 15, 2022 5:02 pm
by starfield
STEVE333 wrote: ↑Sat Oct 15, 2022 3:13 am
Here is my approach using only HOO stars. I haven't learned how to add RGB stars to an NB image yet.
Hi Steve, nice take. Regarding the
RGB stars, I'm not sure if you're using StarXterminator, but if you are it will generate a nice clean starless version for you. I stretch/clip a version of
RGB to get the stars where I want them. Then run StarXterminator with UnScreen selected and generate the stars. Then I combine them back with this in pixel math: ~((~Starless)*(~Stars)) (Got this from the tool tip in StarXterminator).
Jockinireland wrote: ↑Sat Oct 15, 2022 8:41 am
I then used Shawn Neilsons (Visible Dark) process icon to enhance the Red with the HA - worked fairly well
I cant find a similar process icon to put the Oiii into G&B so split the channels, added the Oiii at 50;50 with pixelmath and recombined the channels (not sure it worked so well
) anyone know the proper way to do this - I see the Light Vortex tutorial on it but it was too complicated for me).
David, I'll have to check out the Sean Neilson video. I've never been happy with my
HA/
RGB blends. Yours turned out really nice. I'm wondering for the OIII. Could you try splitting your
RGB channels back out and then blending in the OIII in G and B? I've seen Adam Block do something similar with
HA data. It is a lot of work with lots of views to keep track of...
--Steve
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Sat Oct 15, 2022 10:59 pm
by STEVE333
Wasn't happy with the lack of blue in my first attempt, so, stretched the O3 a bit more and processed again. The processing steps were virtually identical and the amount of detail in the new versions is nearly identical also, but, I like the colors better. The result is:
Thanks
Starfield for the tip on how to combine the
RGB stars with the NB, I'll give it a try.
Steve
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Sat Oct 15, 2022 11:04 pm
by starfield
I like the contrast in 2nd version too. I think it helps the red pillars in the bottom 3rd pop more.
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Sun Oct 16, 2022 2:56 am
by STEVE333
starfield wrote: ↑Sat Oct 15, 2022 5:02 pm
Regarding the RGB stars, I'm not sure if you're using StarXterminator, but if you are it will generate a nice clean starless version for you. I stretch/clip a version of RGB to get the stars where I want them. Then run StarXterminator with UnScreen selected and generate the stars. Then I combine them back with this in pixel math: ~((~Starless)*(~Stars)) (Got this from the tool tip in StarXterminator).
--Steve
Hi Steve - Yes I use StarXTerminator. I tried your suggested approach for combining
RGB stars with my starless NB image. Wow, the stars look soooooo much better!!! Thanks so much for sharing that information. The picture below is the result with the
RGB stars added. They are smaller and look like real stars!!!
You said you stretch/clip the
RGB stars image before running StarXTerminator. Could you explain that further?
Steve
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Sun Oct 16, 2022 2:43 pm
by Jockinireland
@STEVE333 I think your second version is really good - much prefer the colour and contrast level and with the RGb stars its even better
@starfield Whats the benefit of doing the screen/unscreen process. youll see in the one I'm about to post i remove stars in linear then stretch them on their own. what could I expect to improve doing it the other way?
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Sun Oct 16, 2022 3:47 pm
by Jockinireland
So heres my HOO version with
RGB stars.
I simply took the final
Ha and Oiii images from my earlier version, made a straight HOO in channel combination and made some tweaks in curves to emphasise the blue and tone down the red a little.
on the stars I
RGB combined the
RGB's which were already deconvoluted and denoised from my previous version, did PCC, Starexterminator.
Then I stretched the stars with HT, SCNR, added to HOO im pixelmath ("HOO+Stars" - my pixelmath is not up to anything more complicated
) ruduced the stars with Bills star reduction then into Photoshop to balance the background and run astronomy tools 2increase star colour2 which i find super easy and useful. saved as JPG.
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Sun Oct 16, 2022 9:14 pm
by starfield
STEVE333 wrote: ↑Sun Oct 16, 2022 2:56 am
starfield wrote: ↑Sat Oct 15, 2022 5:02 pm
Regarding the RGB stars, I'm not sure if you're using StarXterminator, but if you are it will generate a nice clean starless version for you. I stretch/clip a version of RGB to get the stars where I want them. Then run StarXterminator with UnScreen selected and generate the stars. Then I combine them back with this in pixel math: ~((~Starless)*(~Stars)) (Got this from the tool tip in StarXterminator).
--Steve
Hi Steve - Yes I use StarXTerminator. I tried your suggested approach for combining
RGB stars with my starless NB image. Wow, the stars look soooooo much better!!! Thanks so much for sharing that information. The picture below is the result with the
RGB stars added. They are smaller and look like real stars!!!
You said you stretch/clip the
RGB stars image before running StarXTerminator. Could you explain that further?
.
Hi Steve,
In terms of the
RGB, I read that the unscreen option doesn't work well on linear data. So all I do is a histogram transformation and get the stars where I want them. I will black clip the image (although since I use StarXterminator to create a stars only this may be unnecessary.) Then I run StarXterminator. I believe best practice here is to adjust the the saturation and number of stars before you create the starless version, but I could be wrong.
@Jockinireland in terms of unscreen option. The unscreen option evidently keeps you from picking up artifacts when you subtract them out. Here's a great discussion of it in a YouTube video by Adam Block and Russ Crowman. Unscreen option starts at about 00:30:00.
https://youtu.be/otNoptot78M
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Mon Oct 17, 2022 8:53 pm
by STEVE333
starfield wrote: ↑Sun Oct 16, 2022 9:14 pm
Hi Steve,
In terms of the RGB, I read that the unscreen option doesn't work well on linear data. So all I do is a histogram transformation and get the stars where I want them. I will black clip the image (although since I use StarXterminator to create a stars only this may be unnecessary.) Then I run StarXterminator. I believe best practice here is to adjust the the saturation and number of stars before you create the starless version, but I could be wrong.
Thanks Steve. That's pretty similar to what I'm doing, so, now I can feel better about doing it that way
.
I did notice that the big and bright stars end up with a "halo artifact" around them. I've come up with a technique in
PI to remove the halos. I'll be posting it on the Image Processing subforum when I finish some handyman jobs. The two pictures below show before/after repair for the bright star on the right side of the image that has a pretty bad halo.
Thanks so much for your suggestions on inserting
RGB stars into another image.
Steve
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Tue Oct 18, 2022 4:49 pm
by Sky
Here's my straight HaRGB image of the Wizard. Photoshop CS5 was used for image processing. However, Starnet** was used for creating the starless images.
All images were reduced to 1800 x1156 to speed up processing on my dinosaur desktop.
Workflow ....
RGB image
- Red, Green, Blue images combined using the "Construct
RGB Image" from the Astronomy Tools Action set.
- Overall green background removed using the HSVG (green removal) action from the DeepSkyColor Action set.
- Original
RGB image was duplicated and the copy stretched (Curves and Level) to a point where the stars look good. The image was then split into just the stars (nebula was removed) using a "Separate Stars and Sky" action. Copy
RGB image with stars put aside.
- Original
RGB image was stretched using "Equalize" tool ...thanks for the tip, Carole ... and the stars removed using Starnet**. Back to "Curves" and "Level" to further increase nebulosity. Additional adjustments made on Saturation, Color Balance and Vibrance and overall noise reduction applied using the "Dfine2" action from the Nik Collection. Stars from the Copy
RGB image were then "blended" back into the original
RGB image using the "Lighten" option and star colors boosted using "Increase Star Color" action from Astronomy Tools.
Workflow ...
Ha image
- Used "Equalize" to quickly bring out the nebulosity .... thanks again, Carole ... and removed the stars using Starnet**.
- "Curves" and "Levels" used to bring out the maximum amount of nebulosity.
- Sharpening using "Unsharp Mask".
- Global noise reduction using "Dfine2" and selective noise reduction using "Strong Noise" from the Nik Collection.
- Selective Sharpening/Contrast to bring out faint detail.
Workflow ... Final
- Combining the
RGB image and the
Ha image using "Blend - Lighten"
- Some final color adjustments and selective color enhancements using "Saturation" and "Vibrance.
FULL DISCLOSURE ... I made a major mistake while processing this image. I didn't want to go back and make the needed correction, so I did a quick touch-up. However, the mistake is still glaringly apparent. Can you spot it?
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Wed Oct 19, 2022 2:45 pm
by Jockinireland
Sky wrote: ↑Tue Oct 18, 2022 4:49 pm
Here's my straight HaRGB image of the Wizard. Photoshop CS5 was used for image processing. However, Starnet** was used for creating the starless images.
All images were reduced to 1800 x1156 to speed up processing on my dinosaur desktop.
Workflow .... RGB image
- Red, Green, Blue images combined using the "Construct RGB Image" from the Astronomy Tools Action set.
- Overall green background removed using the HSVG (green removal) action from the DeepSkyColor Action set.
- Original RGB image was duplicated and the copy stretched (Curves and Level) to a point where the stars look good. The image was then split into just the stars (nebula was removed) using a "Separate Stars and Sky" action. Copy RGB image with stars put aside.
- Original RGB image was stretched using "Equalize" tool ...thanks for the tip, Carole ... and the stars removed using Starnet**. Back to "Curves" and "Level" to further increase nebulosity. Additional adjustments made on Saturation, Color Balance and Vibrance and overall noise reduction applied using the "Dfine2" action from the Nik Collection. Stars from the Copy RGB image were then "blended" back into the original RGB image using the "Lighten" option and star colors boosted using "Increase Star Color" action from Astronomy Tools.
Workflow ... Ha image
- Used "Equalize" to quickly bring out the nebulosity .... thanks again, Carole ... and removed the stars using Starnet**.
- "Curves" and "Levels" used to bring out the maximum amount of nebulosity.
- Sharpening using "Unsharp Mask".
- Global noise reduction using "Dfine2" and selective noise reduction using "Strong Noise" from the Nik Collection.
- Selective Sharpening/Contrast to bring out faint detail.
Workflow ... Final
- Combining the RGB image and the Ha image using "Blend - Lighten"
- Some final color adjustments and selective color enhancements using "Saturation" and "Vibrance.
FULL DISCLOSURE ... I made a major mistake while processing this image. I didn't want to go back and make the needed correction, so I did a quick touch-up. However, the mistake is still glaringly apparent. Can you spot it?
Wizard_TSS_Challenge.jpg
Thats really nice - I love how much of the outer gas you have showing and the details in the main nebula are nice and crisp.
As to you "major mistake" I'm not really sure, but I note that not all the stars are showing when compared to my
RGB. Did you throw the
Ha stars on top instead of the
RGB stars? In fact, on further inspection i suspect you may have used entirely the wrong stars and then clone stamped the brightest ones back??
But either way its a nice version and i'd never have noticed anything unusual until you said it.!
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Thu Oct 20, 2022 4:50 pm
by starfield
Real pretty take. Curious what the issue is?
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Fri Oct 21, 2022 2:47 am
by STEVE333
Sky wrote: ↑Tue Oct 18, 2022 4:49 pm
Here's my straight HaRGB image of the Wizard. Photoshop CS5 was used for image processing. However, Starnet** was used for creating the starless images.
FULL DISCLOSURE ... I made a major mistake while processing this image. I didn't want to go back and make the needed correction, so I did a quick touch-up. However, the mistake is still glaringly apparent. Can you spot it?
Wizard_TSS_Challenge.jpg
I agree with Jim. Nice job of bringing out the low level nebulosity.
As for the mistake it does look like the wrong stars were added to the image.
Steve
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Sat Oct 22, 2022 2:08 pm
by Sky
OK ... you guys pretty much figured it out. When I blended back in the stars, somehow they were shifted up in the image. I don't know how that happened, maybe I accidentally bumped a key? Anyway, I didn't spot it until after I had finished the final image. I had already deleted all the RGB data and since it would have required a pretty substantial redo ... I decided to just relocate the brighter stars to their correct positions and call it a day.
Sorry, for posting a defective image, but I thought it would be interesting to see if the mistake would be spotted.
BTW, a lot of extremely well processed images were posted this time around. So much variety and color shading that I'm having a hard time picking a favorite. I only wish I had some familiarity of PixInsight to better understand what is happening in the posted workflows.
Re: Processing Challenge 34 - Astrobee's Wizard
Posted: Sat Dec 10, 2022 4:06 pm
by Rasalgethi
This is my go at Astrobee's Wizard following The Lazy Geek's video.
Pixinsight
ABE
EZ Decon
EZ Soft Stretch
LRGB Combination (using
Ha as Luminance)
PCC
Curves - Contrast, Saturation