Underestimating Read Noise - Sculptor

[Addos (Adam)]

So this was another teaching moment for me in my astro journey



Been using 0 gain on the 294mm for a while now but have only really been taking broadband for stars, so wasnt really concerned about resulting noise in stacks, and wanted to keep sub numbers/gigabytes down.


Carry this on into actually having a go at a broadband target and learned the hard way that the read noise of the 294mm in 2x2 nil gain mode requires 'substantial' exposure to swamp it. I did not do that and ended up underexposing in this respect. That said, I couldnt really increase sub time further, as I was already hitting the point of noticeable star bloat.


Thanks to really effective and simple to use noise reduction tools these days, the main impact was that I lost details I should have been able to tease out further in post, particularly background detail. Gradient control too makes you want to drink excessively. Yes, I dithered, 15px. Didnt help much!


Anyway, the take away I wanted to share with others with all of this is if you have the 294, in bin 2 mode, its 120 gain or nothing


Sculptor in LHaRGB from a bortle 6 during the new moon period.


L ~ 10hrs
RGB ~ 3hrs each
Ha ~ 2hrs

[Stephane]

Wow, great image Adam! I love the warm glow in the core and hydrogen details. Stars look great too.

I am curious about your read noise issue, as in the near future I’ll be taking shortest possible subs to avoid star bloating, while at the sane time swamping read noise. A fine balancing act. I noticed that modern CMOS cameras don’t require much time at all to swamp read noise. Typically anywhere between a couple of seconds and half a minute depending on the camera. What was your sub exposure length?

[Addos (Adam)]

Hey Stephane, sorry about that - should have included that in my first post. And thanks! My first time blending Ha in to an image like this, it certainly is tricky finding the right blending point to keep that lovely blue fringe.



L - 60s
RGB - 120s


Lum was closest to being properly 'swamped', but stars were already showing some bloat by then.



My main point is highly specific to this camera compared to other current zwo cmos cameras. Most have between 3.5-4e read noise at 0 gain/lcg, and I got lazy and assumed same for this chip without specifically checking. I've gone this long being ignorant about it because I almost always do narrowband, at hcg where the read noise gets ridiculously low (~2e).



the 294 though has ~7.3e read noise at nil gain in bin2 mode, so you gots a heck of a lot more swamping to do than other popular cmos chips. my experience is that I wouldnt be able to raise the mean adu much further without too much saturation in the brighter parts of the image.



so next time, I'm just going to wear the extra files and preprocessing time and shoot at the low read noise gain I think for my setup that will be ~30s L and stay at 120s for RGB at 120 gain. 10hrs+ of L in 30s subs, hard drive cries.

[Startrek (Martin)]

Adam,
Superb image of the Sculptor , detail and res is tremendous
Nothing beats a cooled Mono with specific filters that’s for sure ( a 2600MM Mono is on my list for 2023 )
Well done !!!
As far as Read Noise goes , ZWO provide performance charts for most cameras which do help in determining your Gain setting
Attached is the performance chart for the 294MM
You can see the HGC ( High Gain Mode ) switches on at Gain 120 where your read noise drops dramatically from over 6.0e to 1.9e but still delivers excellent Dynamic range at 13 stops. Well depth is still quite high at nearly 20k
You can also see from the charts that increasing Gain higher than 120 doesn’t achieve any real further reduction in read noise but it does proportionally sacrifice Dynamic Range and well depth ( probably more for lucking imaging with super short subs )

My 2600MC has a similar HGC mode where it kicks in at Gain 100 to achieve a low read noise
I did some testing with Gain 0 and Gain 100 under Bortle 7/8 skies and Gain 100 made a huge difference.Plus I dropped my sub exposure lengths down to 2 minute for Narrowband and 90 sec for Broadband and the results were really positive. At my dark site down south I’ll probably only use Gain 0 during the new moon period imaging dim galaxies pushing +5 min subs

Cheers
Martin

[Dave882 (David)]

Great image mate. Yes looks like you’ve taken the harder route but still.. super smooth and stacks of depth and detail! Lovely composition and the Ha is really well blended too. Excellent work all round!!!

The issue of large stacks and 100s GBs of data is not insignificant. I’ve actually right now got a bunch of data on the good ol’ Sculptor turning over in APP. it’s been going for about 24hrs and only 30% done… and I’m not sure whether the stack settings will be ok and might have to go back and do it all again…

[RyanJones]

That’s a lovely sculptor Adam. Finding the optimum gain and exposure lengths is a subject that has seen a bit of a revival of late and it’s interesting to not only hear everyone’s views but to see the results. You’ve admitted to getting it wrong on this particular image but have proven that you can over come it in post. Thanks for postung your experience. It add to the food for thought.

Cheers

Ryan

[Ryderscope (Rodney)]

Your efforts have produced a very respectable Sculptor Galaxy Adam. The outer halo around the galaxy looks good and plenty of detail and structure in the galaxy core.

[Retrograde (Pete)]

Excellent 253 there Adam - great depth in the galaxy & lovely star profiles.

To my uncalibrated eye it could perhaps do with a touch more blue?

[Stephane]

Originally Posted by Addos

Hey Stephane, sorry about that - should have included that in my first post. And thanks! My first time blending Ha in to an image like this, it certainly is tricky finding the right blending point to keep that lovely blue fringe.

…

the 294 though has ~7.3e read noise at nil gain in bin2 mode

That’s quite interesting. So 7.3e is clearly not insignificant at 60 seconds. Thanks for sharing that experience.

As for blending in the Ha, I am currently imaging NGC 55 without a filter (OSC) and plan to do an Ha boost with the red channel of the L-Extreme. I’ve never tried this before, but looking at the quality of your image makes me think it’s well worth that extra effort.

Another great point about this image is how sharp the closer edge of the galaxy is in contrast to the other end. I’m not sure if this was done intentionally or not but it creates a powerful 3D effect!

[Addos (Adam)]

thanks alot guys! very gracious comments on my mediocre image


Next new moon I'm going to have another go at this one, and hopefully with the read noise sorted I'll get alot more of those eliptical and spiral galaxies in the background to come out!


Martin - I'm 100% with you on that one! the 2600mm is my next purchase too the dynamic range, low read noise across the board and no flat darks are highly exciting features!
the 294 is great, but i really want to take advantage of the esprit's huge flat field, and im never going to be able to afford a 6200 setup


Dave - the thing I like most about processing 25gb of data for broadband targets is when you get to the end, and realise you screwed something up in registration/local norm and have to do it all again. lol. think i ran about 6 hrs processing time on this lot... x 2.


Ryan, Rod, Pete - thank you! Pete, totally agree, but cite aforementioned Ha blending troubles. The amount of Ha I got in my raw sub was quite impressive - there were traces through all of the spiral arms, not just the star forming regions! Subtracting all of that without losing the star forming region bits was what I was struggling with in the blend! More practice definitely required!


Stephane - no not significant, but remember i lost all that background detail because the mean adu had to be so high. might be more viable at a dark site, but in a bortle 6 or worse its my opinion imaging at the high read noise level isnt worth it. again, this is highly specific to the 294. you boys with 2600's prob wouldnt have the same experience, as the read noise at nil gain on those chips is less than half that of the 294


i love that about this target too (clearly defined close edge). think the universe did that one though, not me