NGC6995 L-eNhance attempt

Questions and answers about processing in StarTools and how to accomplish certain tasks.
Burly
Posts: 387
Joined: Mon Aug 31, 2015 7:06 am
Location: Northamptonshie uk

Re: NGC6995 L-eNhance attempt

Post by Burly »

Also if using the tracing panel , I have read the light in some can be the wrong spectrum and can mess colours up I just use a tablet dimmed to lowest setting on my 70 mm refractor..


Regards Dave
deanST
Posts: 17
Joined: Wed Mar 10, 2021 6:35 pm

Re: NGC6995 L-eNhance attempt

Post by deanST »

thank-you all for responding: really appreciate your insights.
There must have been an issue with one of my master bias/flat/dark. I redid them all from scratch in Siril
and then re-calibrated the lights. When I look at the final stack in Siril or when I import into ST, no grid pattern
weirdness anymore.
I processed my last and hopefully final stack after adding more subs from last night for a total of 13.6 hrs of
OSC L-eNhance data. I also applied the current workflow to the original mis-calibrated data: the background
definitely cleaned up but overall the differences were not astounding. I am pleased with this bicolor
result (C&C welcomed). Final stacked fits: https://tinyurl.com/y3ptjn4f, and processed:

Image
Mike in Rancho
Posts: 1142
Joined: Sun Jun 20, 2021 10:05 pm
Location: Alta Loma, CA

Re: NGC6995 L-eNhance attempt

Post by Mike in Rancho »

1500+ lights and dithered every frame? Wow! :shock:

I saw this post but couldn't really get to it earlier. Your follow-up does indeed look better to me. But when I looked it over and processed through, I couldn't manage to find any matrix or grid like problem. Now I have run into that - in fact the West Veil I did had a really bad grid noise, which I discovered was caused by using spiral dithering. Luckily it was just the first night and 3 hours of data. Eventually at 9 hours of data, the grid patterns were mostly washed out by the extra integration using random dithers.

Fairly standard ST processing for HOO bicolor, maybe other than my crop and rotate choices. Wipe was more aggressive than default but not crazy, and I also used the correlation filter. AutoDev stretch was conservative. Did not use Sharpen or Shrink. Backed off SS to 75% strength. In denoise I bumped up the bottom scale partway, and also reduced the scale correlation. Then with tracking over I did a 90% gamma in FilmDev. No other touchups.

Your data seems pretty impressive to me. :obscene-drinkingcheers: Might be I'm just not able to find the issues that others can see? Or could be because I didn't shrink stars to get them out of the way.

dean st E Veil ST8 1B.jpg
dean st E Veil ST8 1B.jpg (491.41 KiB) Viewed 2489 times
deanST
Posts: 17
Joined: Wed Mar 10, 2021 6:35 pm

Re: NGC6995 L-eNhance attempt

Post by deanST »

cheers indeed Mike! thank-you! For me the pattern was really evident when looking at the individual channels in Siril and
sometimes at certain zoom scales in ST. After diligent re-calibration, it all looks smooth(er).

Dithering every frame in Ekos, which the docs merely state is "random", definitely helps with pattern noise in my experience.
I used to dither every 5 frames and that was not enough. So it adds at minimum 1.5x the actual exposure time, but
not 2x. I can live with that: I don't have the luxury of desert skies or a remote obsy. Maybe when
I retire to the country ...

Given that this is tri-band data, do you think it would enhance the final result by loading the
stack in Compose as r, g, b and accent data and using the new NBAccent module ?
Mike in Rancho
Posts: 1142
Joined: Sun Jun 20, 2021 10:05 pm
Location: Alta Loma, CA

Re: NGC6995 L-eNhance attempt

Post by Mike in Rancho »

Hi Dean,

Ok, well yeah I guess. I also dither every 1x with the L'eNhance, but my subs are more like 360s, not 30! Which my rig is doing right now in fact. For RGB data last night I was dithering every 3x at 120s.

Uncertain on the NB Accent, I've only played around with it once. Only one way to know! Well, two -- Ivo tells us. :lol:

I'm also not convinced on the tri/quad band thing. They have two lumps for bandpasses, so seems like duoband to me, just a matter of how wide. Each channel will only know a photon hit it, so say between Hb and OIII, unless you can parse out the particular ratio of B and G to separate them...? And anyway for a duoband composite I think B and G are blended into one O as it is.

Off the top of my head I wouldn't think throwing the same file into NB would be real helpful, since how would you be differentiating things? However, the NB module does permit a number of settings as well as a new AutoDev stretch, so....?

But being the exact same data, I'm not sure it will accomplish anything that you couldn't do better with the other modules?

If you find out let us know! Might be a backdoor way of blending two different stretches?
Post Reply