M81 and M82 OSC

[Richard_]

I received an adjustable reducer earlier in the week for my refractor which saw first light last night. I'm happy that I was able to get the adjustable distance and back focus right the first time which resulted in round stars look across the whole field. Although the moon was about 50% last night I decided to go for M81 and M82 with my OSC camera just for fun as I've shot this before so wanted to compare star size etc. I've got just under 3.5hrs of useable data so I had a quick play around with it today. It turns out I didn't have any darks at this gain and exposure time in my library so I contibuted to calibrate without them, but it doesn't look like it's made much difference which is nice!

I'm quite happy with the results, although I don't have a great amount of star colour. I took separate 10 x 30s subs just to capture star colour but they appeared the same as my final image comprising 180s subs so I didn't bother introducing them.

You'll find my processing work flow and a FIT version of my master light should anyone wish to play with my data. Comments and constructive criticism are welcome 🙂

M81.fit

 

M81 – One Session

 

*** EQUIPMENT ***

Camera: ASI533MC-Pro

Telescope: WO FLT120 (x0.8 reducer)

Mount: EQ6-R

Guide Scope/Camera: WO 50mm Guide scope / ASI290MM

Filter: Baader UV-IR

Peripherals: Beelink Mini-PC, PPBA, Dew Heaters, Sesto Senso 2

 

*** ACQUISITION DETAILS ***

Gain: 0

Sensor temp: -10°C

Subs:  67 x 180s (3.4hrs)

Darks: 0

Flats: 20 (45% of histogram)

Dark Flats: 20

Dates: 10-Apr-2022

Area: Northern Hemisphere, Bortle 5 zone

 

*** PRE-PROCESSING ***

• WBPP
  • Manual subframe selector
  • Register
  • Light Stack
  • Save as XISF

 

*** PROCESSING ***

• Dynamic Crop
• Background extraction
  • DBE, division
  • ABE, division
• Photometric Colour Calibration with Background Neutralisation
  • Select background preview and set upper limit based on preview
• EZ Denoise
• GHS Stretch
• Starnet++ v2
• On starless version
  • Range selection to mask galaxies
  • HDR and LHE
  • SCNR green 
  • Curves transformation to improve saturation of galaxies
  • Flip mask and reduce background saturation
• Pixel Math to insert stars
• Apply star mask
  • Slight increase in colour saturation
  • Unsharp mask
• Save as JPEG

Edited April 10, 2022 by Richard_
Added in missing processing steps and equipment

[juno16]

Beautiful image Richard. Wow, what a closeup!

I appreciate all of the fine processing information that you provided. It really helps me learn different techniques in Pixinsight.

Concerning your stars, did you do anything with the star mask before adding the stars to the starless image (like Curves - Saturation)?

 

[Richard_]

11 minutes ago, juno16 said:

Beautiful image Richard. Wow, what a closeup!

I appreciate all of the fine processing information that you provided. It really helps me learn different techniques in Pixinsight.

Concerning your stars, did you do anything with the star mask before adding the stars to the starless image (like Curves - Saturation)?

 

Thanks for your kind words Juno!

I think I did increase the saturation of stars just before the unsharp mask step (forgot to jot down in my work flow), but the cores of the stars looked a bit "deep fried" with chroma noise before I could present a nice hue to them so I couldn't push it much farther.

I've recently been using the Generalised Hyperbolic Stretch script for stretching which is doing an excellent job at reducing star bloat but perhaps I'm doing something wrong which is limiting my ability to bring out star colour later on? 

[juno16]

46 minutes ago, Richard_ said:

Thanks for your kind words Juno!

I think I did increase the saturation of stars just before the unsharp mask step (forgot to jot down in my work flow), but the cores of the stars looked a bit "deep fried" with chroma noise before I could present a nice hue to them so I couldn't push it much farther.

I've recently been using the Generalised Hyperbolic Stretch script for stretching which is doing an excellent job at reducing star bloat but perhaps I'm doing something wrong which is limiting my ability to bring out star colour later on? 

I wish that I could help more. I am "only" about a year into using PI and have a long way to go.

I have toyed around with the Generalised Hyperbolic Stretch, but have yet to use it on an actual image that I am processing. I usually use the EZ Soft Stretch, but have found that different images produce very different results using some of the other stretching techniques such as STF/HT and Masked Stretch. I make several copies of the linear image and try different stretches to see which I like the best.

I have found that Masked Stretch does a better job at protecting the stars than other stretching methods, but again, I am still experimenting quite a bit.

Good luck and if you find a suitable method to produce star colors that you desire please share your methods.

[ONIKKINEN]

Image looks pretty good to me, if a bit green. Try running SCNR green? Worked great when i tried it with just the JPEG. I tried processing your file and didn't get the green cast with SiriL photometric colour calibration so not sure how to advice on how that side of thing would go in PI. But SCNR rescues these anyway, so whatever the reason that there is an easy fix.

I can get plenty of star colour out of the image and without much of a saturation boost even so i guess one of the processes in your workflow brought the issue in. Try without starnet or any other layer trickery. Star removal AIs leave a bit of fat stars behind and the resulting star layer can look like you described, chroma noisy, after saturation attempts. If this happens to me i try with a lower stretched version of the image and see if the artifacts go away.

Stars in your image do look a bit fat and soft to me anyway, and measuring FWHM gets concerning results of around 5-6'' FWHM (depending on measurement tool). This seems like focus, seeing, guiding issues. Could be any one of the former or a mix of all of them. You could try being more strict with the subframe selector tool and see if some of the data is better?

Part of the issue, one that is easy to fix by restacking with colour channel alignment if this exists in PI (pretty sure it does) is that your colour channels are a bit misaligned. Happens when the target is low in the sky. Lower than 50 degrees and this gets very noticeable. Lower than maybe 40 degrees and it starts getting into a big problem, which could also explain the big stars if you did shoot when the target was at these altitudes. Also, plays a part in how much saturation you can push the stars through, since quite early on you start to notice this colour separation issue.

Example below, reds and blues are on opposite sides because they refract at different rates through the atmosphere.

[Richard_]

29 minutes ago, ONIKKINEN said:

Image looks pretty good to me, if a bit green. Try running SCNR green? Worked great when i tried it with just the JPEG. I tried processing your file and didn't get the green cast with SiriL photometric colour calibration so not sure how to advice on how that side of thing would go in PI. But SCNR rescues these anyway, so whatever the reason that there is an easy fix.

I can get plenty of star colour out of the image and without much of a saturation boost even so i guess one of the processes in your workflow brought the issue in. Try without starnet or any other layer trickery. Star removal AIs leave a bit of fat stars behind and the resulting star layer can look like you described, chroma noisy, after saturation attempts. If this happens to me i try with a lower stretched version of the image and see if the artifacts go away.

Stars in your image do look a bit fat and soft to me anyway, and measuring FWHM gets concerning results of around 5-6'' FWHM (depending on measurement tool). This seems like focus, seeing, guiding issues. Could be any one of the former or a mix of all of them. You could try being more strict with the subframe selector tool and see if some of the data is better?

Part of the issue, one that is easy to fix by restacking with colour channel alignment if this exists in PI (pretty sure it does) is that your colour channels are a bit misaligned. Happens when the target is low in the sky. Lower than 50 degrees and this gets very noticeable. Lower than maybe 40 degrees and it starts getting into a big problem, which could also explain the big stars if you did shoot when the target was at these altitudes. Also, plays a part in how much saturation you can push the stars through, since quite early on you start to notice this colour separation issue.

Example below, reds and blues are on opposite sides because they refract at different rates through the atmosphere.

Ops, another process I missed out in my log! I did include SCNR but only during the galaxy mask (I guess I should have done this without a mask on to reduce background green). 

Wow, you can get good star colour? Awesome, that means I can salvage the image. 

I know my star size is a bit pants. At the beginning of the night whilst the target was still high, I was getting FWHM of around 2-3" for my subs but this increased up towards 5-6" for the latter subs (probably about 75% of subs). As you say, this is due to the target lowering throughout the night. I have an electronic focuser which auto focuses after 10% increase in HFR so I'm confident that the focus was as good as it could be. Guiding is about 0.8" RMS which is also OK. Sadly, I know my seeing isn't amazing (averages 2-5" FWHM) so I can't do much about that! If I have lots of data, I normally reject high FWHM values but in this case I left them in just to have a play with the new data so I expected a little bit of star bloat. 

Excellent sopt on the atmospheric dispersion too! It was extremely noticeable without a reducer when shooting at native 780mm and is difficult to deal with when using a OSC. I previously tried splitting RGB channels, registering and recombining but this had little effect. I suspect its due to blue being out of focus slightly as well as channel misalignment during imaging. 

Please do share any images you create if you decide to continue playing with my data. Thanks for taking the time to provide good feedback too, it's much appreciated 😊

Edited April 10, 2022 by Richard_

[ONIKKINEN]

13 hours ago, Richard_ said:

Please do share any images you create if you decide to continue playing with my data. Thanks for taking the time to provide good feedback too, it's much appreciated 😊

I did do this as a quick process to see the star colour issue first hand:

I binned it 2x2 in ASTAP, but it could be binned more without much or any loss of detail, not that i focused much on detail in this shot (mostly colour processing). 5-6'' fwhm stars will do that to an image. Not much you can do about the seeing other than try to only shoot near zenith, but that's not always possible of course.

Thats getting close to the limit of where the atmospheric dispersion effects start being noticeable at a glance, but there is nice variation in star colours already so its saturated enough IMO. I started stretching with an asinh stretch with a power of 1000 in Siril. Dont quite know how this works but the asinh stretch also stretches colours and the resulting image needs less, or no saturation afterwards. The regular histogram stretch function usually results in an image that needs a fair bit of saturating afterwards to get anything out of stars.

The data is quite nice, apart from the big stars. Would love to see this with more of the good stuff and the seeing effected subs removed from play, weather willing of course.

[Richard_]

On 11/04/2022 at 12:41, ONIKKINEN said:

I did do this as a quick process to see the star colour issue first hand:

I binned it 2x2 in ASTAP, but it could be binned more without much or any loss of detail, not that i focused much on detail in this shot (mostly colour processing). 5-6'' fwhm stars will do that to an image. Not much you can do about the seeing other than try to only shoot near zenith, but that's not always possible of course.

Thats getting close to the limit of where the atmospheric dispersion effects start being noticeable at a glance, but there is nice variation in star colours already so its saturated enough IMO. I started stretching with an asinh stretch with a power of 1000 in Siril. Dont quite know how this works but the asinh stretch also stretches colours and the resulting image needs less, or no saturation afterwards. The regular histogram stretch function usually results in an image that needs a fair bit of saturating afterwards to get anything out of stars.

The data is quite nice, apart from the big stars. Would love to see this with more of the good stuff and the seeing effected subs removed from play, weather willing of course.

Thanks Onikkinen, I'm glad to hear that my data is good as I've spent quite some time trying to get my flat frames right with my UV-IR filter!

I just had a play with my stack again, this time I used the Arsinh stretch function within the GHS script, adjusted the order of executing HDR and saturating my stars with a star mask. I'm able to introduce a lot more colour into the stars as before but the colour gradient of the stars caused by atmospheric dispersion are exaggerated with the Arcsinch stretch function so don't zoom in haha! I'll have to break the bad news to the missus that I'll need a mono camera to avoid this The green cast is also gone although I only had to use SCNR once whilst in the linear stage.

I'm happy with this re-process but I think M82 looked nicer in my first attempt, so it sounds like I'll have to refine the stretching and masking a bit more. Thanks for the suggestions

[ONIKKINEN]

On 14/04/2022 at 21:53, Richard_ said:

I'm able to introduce a lot more colour into the stars as before but the colour gradient of the stars caused by atmospheric dispersion are exaggerated with the Arcsinch stretch function so don't zoom in haha! I'll have to break the bad news to the missus that I'll need a mono camera to avoid this

Actually you can resample and the problem goes away. Unfortunately cant advice on how this goes with Pixinsight, but i would be shocked if there is no feature like this there. In DSS there is a simple "align colour channels" option which uses resampling and attempts to fix the colours and it works ok-ish. In SiriL, which i use, i do a bayer split for all the subs (dividing each sub to its 4 raw colour channels and stacking these as mono subs, which is the most "true" way to treat data from a colour sensor camera) and then stack those with one sub selected as the reference frame for all the subs. This stacking on one reference frame for all the different colour channels will align them during stacking and the atmospheric dispersion issue goes away.

That is of course if you dont want to buy a mono camera. If you do want a mono camera then best to tell the missus that there is no way to fix the issue without getting one 😉.

[Richard_]

I revisited this target with some additional data. Originally, I kept the bad subs in (i.e. too high FWHM) just to have a go on creating a final target. Originally I had 3hrs of poor data but now I have 2.5hrs of "OK" data after filtering out the worst subs. Processing was similar as the above but with some extra steps:

• Aligned my R,G,B channels to a single reference image (best Green sub) before stacking
• Final step was to downsample the image to 2x2 (Pixinsight > IntegerResample).

The end result is a dramatic reduction in visible atmospheric dispersion in the stars. I also didn't need to push the colour saturation too hard to get some nice colour out of the stars and the subject so given the small amount of data I have, I'm happy with the progress I've made. Thanks @ONIKKINEN for the advice

 

[Lee_P]

Very nice!

[wimvb]

I downloaded the original TIFF (in your first post) and had a go at it in PI

I used Photometric Colour Calibration with an Sb galaxy as white reference. Then Arcsinh stretch to get some colour. MMT noise reduction on chrominance to reduce the colour noise.

Cleaned up the atmospheric/abberation effects in the stars with MMT (removed layers 1 - 4 in chrominance).

Extracted luminance from the linear image, and sharpened detail with Deconvolution. I found that the stars PSF was a bit "fat", either poor seeing or oversampling.

Repeated histogram stretch followed by HDR Multiscale Transformation to restore detail in the cores of M81 and M82. I used different settings for M81 and M82 with HDRMT (creative masking).

LRGB combination

Replaced LRGB stars with RGB stars.

90 degrees rotation.

[Richard_]

Nice work @wimvb! The reason why I will continue to attach my FIT files is that it allows a fair comparison between different processing work flows using the same data (admin, if you're reading this post, it could be an idea for SGL competitions! 😉).

Yeah, the PSF is going to be on the "fat" side of things as my seeing conditions aren't that good plus I didn't reject bad FWHM subs in my first data set. Resampling to 2x2 works OK with this data. 

Thanks for taking the time to process and submit. I need to better understand better how MMT works.

[wimvb]

2 hours ago, Richard_ said:

admin, if you're reading this post, it could be an idea for SGL competitions! 😉

They are aware of this. There is an image processing competition with data from FLO’s/IKI’s own rig.

https://stargazerslounge.com/forum/294-iki-observatory/

Edited May 6, 2022 by wimvb