Advice on path to take - integrating frames of different exposure lengths with dark corners

Discuss how you are able to get those fantastic images!!!
Post Reply
User avatar
ram United States of America
Saturn Ambassador
Articles: 0
Offline
Posts: 349
Joined: Tue Jun 11, 2019 3:21 am
4
Location: Youngstown, NY, USA
Status:
Offline

TSS Photo of the Day

Advice on path to take - integrating frames of different exposure lengths with dark corners

#1

Post by ram »


My query is about what strategy to use to integrate 153 frames of 5 minutes each taken at a resolution of 1.5 arcseconds/px and 204 frames of 7 minutes each taken at a resolution of 1 arcseconds/px. Due to differences in framing, there are black corners on some of these images after registration (10% of pixels total says). No, it cannot be cropped out.

Here's what I'm trying:

1. Just do normal image integration of all 357 frames using GESD test - doesn't work great but faint borders can be seen.

2. Do drizzle integration of all 357 frames using GESD test - seems to work better but can still see borders. It's not bad at all actually and may work and is so far the best I've gotten with the best SNR.

3. Try to do AN on 1-2, which doesn't work due to the dark corners. I could try LocalNormalization. Should I?

4. Do 1-2 separately on the set of 153 and 204 frames and then combine using HDRComposition, GradientHDRComposition, and GMM. Previously when I experimented with 1/4 of the data, I found GHDRC worked like a charm and produced the best SNR after combination but now it doesn't work at all and GMM does work with a marginal SNR improvement.

Thoughts? I really would've thought AN would've worked but perhaps I should try LN.

You may ask why not just do 4 which I've tested with a smaller number of frames and I think GHDRC really worked well. But I'm not sure it's the optimal path to take. 90% of the pixels are common in all 357 frames so why not take advantage of this data? The statistics would be more robust with 357 frames instead of half the frames, no?

Thanks!

--Ram

PS: About the AN not dealing with 0 (all black) value pixels, can I use PixelMath to write an expression that would go through the all black pixels and give it a tiny value of 1 and have it work? Or should I use LN instead?
Tubes: Celestron 9.25" 235mm f/10 XLT EdgeHD SCT; Meade ETX 80mm f/5 achromat; Coronado SolarMax II 60mm f/6.6 Hα <0.7Å BF10 solar; Stellarvue 70mm f/6 triplet apochromat; Obsession UC18 457mm f/4.2 with Argo Navis & ServoCAT; Takahashi FS128 5" f/8.1 and FC100DF 4" f/7.4 fluorite doublet apochromats. Mounts: AVX; LXD75; Paramount MyT. Eyepieces: 2" Tele Vue Ethos 4.7/13/21mm, Paracorr, 2,4x Powermate; Stellarvue 0.8x, Takahashi 0.7x, 0.66x reducer/corrector. Cameras: ZWO ASI120MC-S; Lodestar X2c; X2m; Canon T7i; QHY163M; QHY247C; QHY294M-Pro. Filters: 1.25" Astrodon 5nm Ha, 3nm O3 and S2; Chroma LRGB.
Image
User avatar
STEVE333 United States of America
Inter-Galactic Ambassador
Articles: 0
Offline
Posts: 3466
Joined: Sat May 11, 2019 5:01 pm
4
Location: Santa Cruz, Ca, USA
Status:
Offline

TSS Awards Badges

TSS Photo of the Day

Re: Advice on path to take - integrating frames of different exposure lengths with dark corners

#2

Post by STEVE333 »


Hi Ram - I don't understand some of the acronyms:

GESD
AN
LN

Steve
Steve King: Light Pollution (Bortle 5)
Telescope + Mount + Guiding: W.O. Star71-ii + iOptron CEM40 EC + Orion Magnificent Mini AutoGuider
Camera: ASI 1600MM Pro + EFW Filter Wheel + Chroma 3nm Siii, Ha, Oiii + ZWO LRGB Filters
Software: PHD2; APT; PixInsight ***** My AP website: www.steveking.pictures
Image
Image
User avatar
ram United States of America
Saturn Ambassador
Articles: 0
Offline
Posts: 349
Joined: Tue Jun 11, 2019 3:21 am
4
Location: Youngstown, NY, USA
Status:
Offline

TSS Photo of the Day

Re: Advice on path to take - integrating frames of different exposure lengths with dark corners

#3

Post by ram »


Generalised Student Deviate test - one of the image integration options for pixel rejection

AN - adaptive normalisation
LN - local normalisation

I usually assume once I first write LocalNormalization then the rest of the time people can follow when I say LN but perhaps I should put in parenthesis (what we do in formal writing).

--Ram
Tubes: Celestron 9.25" 235mm f/10 XLT EdgeHD SCT; Meade ETX 80mm f/5 achromat; Coronado SolarMax II 60mm f/6.6 Hα <0.7Å BF10 solar; Stellarvue 70mm f/6 triplet apochromat; Obsession UC18 457mm f/4.2 with Argo Navis & ServoCAT; Takahashi FS128 5" f/8.1 and FC100DF 4" f/7.4 fluorite doublet apochromats. Mounts: AVX; LXD75; Paramount MyT. Eyepieces: 2" Tele Vue Ethos 4.7/13/21mm, Paracorr, 2,4x Powermate; Stellarvue 0.8x, Takahashi 0.7x, 0.66x reducer/corrector. Cameras: ZWO ASI120MC-S; Lodestar X2c; X2m; Canon T7i; QHY163M; QHY247C; QHY294M-Pro. Filters: 1.25" Astrodon 5nm Ha, 3nm O3 and S2; Chroma LRGB.
Image
User avatar
ram United States of America
Saturn Ambassador
Articles: 0
Offline
Posts: 349
Joined: Tue Jun 11, 2019 3:21 am
4
Location: Youngstown, NY, USA
Status:
Offline

TSS Photo of the Day

Re: Advice on path to take - integrating frames of different exposure lengths with dark corners

#4

Post by ram »


In the past when I was experimenting with 1/4 the data, I found some great solutions with GradientHDRComposition that had the best SNR and was seamless. Yet when I did the same thing with the full data, it didn't work!

And perhaps this would've worked then but I'm finding if I do LN, then it is also seamless but the SNR goes down. Anyone know why? But perhaps this is worth it. Or perhaps I can patch the high SNR seamed version with areas where the seam is from the LN version.

--Ram
Tubes: Celestron 9.25" 235mm f/10 XLT EdgeHD SCT; Meade ETX 80mm f/5 achromat; Coronado SolarMax II 60mm f/6.6 Hα <0.7Å BF10 solar; Stellarvue 70mm f/6 triplet apochromat; Obsession UC18 457mm f/4.2 with Argo Navis & ServoCAT; Takahashi FS128 5" f/8.1 and FC100DF 4" f/7.4 fluorite doublet apochromats. Mounts: AVX; LXD75; Paramount MyT. Eyepieces: 2" Tele Vue Ethos 4.7/13/21mm, Paracorr, 2,4x Powermate; Stellarvue 0.8x, Takahashi 0.7x, 0.66x reducer/corrector. Cameras: ZWO ASI120MC-S; Lodestar X2c; X2m; Canon T7i; QHY163M; QHY247C; QHY294M-Pro. Filters: 1.25" Astrodon 5nm Ha, 3nm O3 and S2; Chroma LRGB.
Image
User avatar
ram United States of America
Saturn Ambassador
Articles: 0
Offline
Posts: 349
Joined: Tue Jun 11, 2019 3:21 am
4
Location: Youngstown, NY, USA
Status:
Offline

TSS Photo of the Day

Re: Advice on path to take - integrating frames of different exposure lengths with dark corners

#5

Post by ram »


There's one more thing I'd like to add here: whenever I compare the images I've taken of these very long total exposures, i.e., 30+ hours each filter, I see that I have better contrast than most people's images for reasons I can't quite understand. I would've thought more data would on average just increase the median lightness but at least with my set up, it is contrast that gets visibly better. There's no more "smear" of light but rather the dark and light areas stand out clearly. It's happening right now with a lot of Ha data of the Bat at ~37h (for half the frame).

--Ram
Tubes: Celestron 9.25" 235mm f/10 XLT EdgeHD SCT; Meade ETX 80mm f/5 achromat; Coronado SolarMax II 60mm f/6.6 Hα <0.7Å BF10 solar; Stellarvue 70mm f/6 triplet apochromat; Obsession UC18 457mm f/4.2 with Argo Navis & ServoCAT; Takahashi FS128 5" f/8.1 and FC100DF 4" f/7.4 fluorite doublet apochromats. Mounts: AVX; LXD75; Paramount MyT. Eyepieces: 2" Tele Vue Ethos 4.7/13/21mm, Paracorr, 2,4x Powermate; Stellarvue 0.8x, Takahashi 0.7x, 0.66x reducer/corrector. Cameras: ZWO ASI120MC-S; Lodestar X2c; X2m; Canon T7i; QHY163M; QHY247C; QHY294M-Pro. Filters: 1.25" Astrodon 5nm Ha, 3nm O3 and S2; Chroma LRGB.
Image
User avatar
STEVE333 United States of America
Inter-Galactic Ambassador
Articles: 0
Offline
Posts: 3466
Joined: Sat May 11, 2019 5:01 pm
4
Location: Santa Cruz, Ca, USA
Status:
Offline

TSS Awards Badges

TSS Photo of the Day

Re: Advice on path to take - integrating frames of different exposure lengths with dark corners

#6

Post by STEVE333 »


Hi Ram - I can't help with the GESD/AN/LN questions as I haven't use any of them.

As for the "contrast" question. If you integrate N images together the following happens:
  • The N images are added together (with hot/cold/excessively-noisy pixels being excluded) and then the sum is divided by N. Thus the signal levels do not increase. However, the noise is reduced by sqrt(N). Thus the image signal-to-noise increases.
  • Adding more data will further reduce the stacked image noise.
  • Images are typically stretched to the point where noise is just visible or maybe just a little less so the noise isn't visible. So, when the noise is reduced the image can be stretched further before the "noise limit" is reached. This produces a higher contrast image.
Hope this makes sense.

Steve
Steve King: Light Pollution (Bortle 5)
Telescope + Mount + Guiding: W.O. Star71-ii + iOptron CEM40 EC + Orion Magnificent Mini AutoGuider
Camera: ASI 1600MM Pro + EFW Filter Wheel + Chroma 3nm Siii, Ha, Oiii + ZWO LRGB Filters
Software: PHD2; APT; PixInsight ***** My AP website: www.steveking.pictures
Image
Image
User avatar
ram United States of America
Saturn Ambassador
Articles: 0
Offline
Posts: 349
Joined: Tue Jun 11, 2019 3:21 am
4
Location: Youngstown, NY, USA
Status:
Offline

TSS Photo of the Day

Re: Advice on path to take - integrating frames of different exposure lengths with dark corners

#7

Post by ram »


Hi Steve, thanks for the explanation! That makes a lot of sense.

I'm surprised you don't use GESD test - I find it superior to all the other algorithms like Winsorized sigma and linear fit clipping at rejecting outliers. However it does require 25+ frames to be optimal.

AN won't work in my case because I have a lot of black areas but it turns out LN works, at least at seamless integrating all 357 frames (153 x 5 and 204 x 7) in one shot. Unfortunately LN reduces the SNR quite a bit relative to the undrizzled version and makes the image look fuzzier compared to the non-LocalNormalized version. I may just have to use your method of mixing and matching between images to only mix and match the seamed areas from the LN version onto the non-LN version.

--Ram
Tubes: Celestron 9.25" 235mm f/10 XLT EdgeHD SCT; Meade ETX 80mm f/5 achromat; Coronado SolarMax II 60mm f/6.6 Hα <0.7Å BF10 solar; Stellarvue 70mm f/6 triplet apochromat; Obsession UC18 457mm f/4.2 with Argo Navis & ServoCAT; Takahashi FS128 5" f/8.1 and FC100DF 4" f/7.4 fluorite doublet apochromats. Mounts: AVX; LXD75; Paramount MyT. Eyepieces: 2" Tele Vue Ethos 4.7/13/21mm, Paracorr, 2,4x Powermate; Stellarvue 0.8x, Takahashi 0.7x, 0.66x reducer/corrector. Cameras: ZWO ASI120MC-S; Lodestar X2c; X2m; Canon T7i; QHY163M; QHY247C; QHY294M-Pro. Filters: 1.25" Astrodon 5nm Ha, 3nm O3 and S2; Chroma LRGB.
Image
Post Reply

Create an account or sign in to join the discussion

You need to be a member in order to post a reply

Create an account

Not a member? register to join our community
Members can start their own topics & subscribe to topics
It’s free and only takes a minute

Register

Sign in

Return to “Image processing”