Orion - Flame Nebula, Horse's Head, M42 etc. and lots of hydrogen

[Gina]

Gain is something I'm not sure of.  It would seem that reducing the gain would mean increasing the total integration time to get the same result regarding faint nebulosity but I'm not sure this is true.  I just haven't had the clear night skies to do enough testing.

[wimvb]

Gain affects the dynamic range, ie the span between the read noise floor and the brightest values that still show variation. Low gain = high dynamic range. I don't know how that affects total integration time. Normally nr of subs x exposure time = constant (to reach a certain snr). But at the same time you can recover dynamic range by taking more exposures. What I don't know is if this is two ways of saying the same thing. If it isn't, one could argue that the subs that are "used" to recover dynamic range, can't also be used to decrease noise. (Otherwise there would be a free lunch.) That would imply that if you increase the gain and also reduce the exposure time, you would need a longer total integration time than if you take longer subs at lower gain.

But the benefit of higher dynamic range is why I normally use low gain. Even though my camera has a high read noise at low gain.

All this is very murky to me, to be honest. 

[gorann]

Regarding gain and offset for the ASI1600, I always use unity gain = 139, and offset 50. The main reason being that is seems to be a very common setting among ASI1600 users and it has worked fine for me. My exposure times depend on the brightness and filter. For lum 3 - 5 min, and for NB 10 - 15 min. The reason I do not normally use shorter exposure times is that I rather not stack many hundreds of subs. However, recently at Lizard Island I only had an un-guided Star Adventurer mount so I used 90 s exposures (with an ASI071 color CMOS) to reduce star trailing and then I had to deal with 100 - 200 subs to stack. I think I brough home about 100 Gb of data.

Edited January 12, 2020 by gorann

[Gina]

Did an assessment of settings last night.  Unfortunately, cloud rolled in closely followed by rain before I could take a set of lights.  With several tests I found that a gain of 180 and exposure of 120s resulted in Alnitak just reaching white level.   The diffraction patterns around the brighter stars had gone.

[Adreneline]

On 12/01/2020 at 08:41, gorann said:

Regarding gain and offset for the ASI1600, I always use unity gain = 139, and offset 50.

As do I and it works a treat for me.

I now use an ASIair and there is no facility to alter offset within the software - it is set at 50; you can change the gain but I always use 139.

I have found 120s and 180s exposures work best for me in my Bortle 5 environment. I've gone as far as 300 and not been able to detect any improvement (using IC1396 as my reference). I have found that I waste less subs by using 120s as I live on the flight path to EMA and far too many subs used to get written off when I was regularly using 300s+ exposures with my 428ex ccd; the aircraft are so low when they fly over there is no hope of processing out the light trails - they obliterate half the image!

Sticking to a given gain / offset /exposure also makes life much simpler with calibration frames.

Adrian

[Gina]

Some new data for this from Tuesday, 28th January.  Gain of 180 and exposures of 120s.  In spite of an imaging run of around 4 hours I only got a total of just 30 subs that were at all reasonable - just an hour's worth.  The problem was passing thin clouds.

This is a screenshot of the result of calibrating, aligning and stacking in PixInsight with STF auto-stretch.

Here's a series of different crops and stretches performed in PixInsight of the Orion image.

Edited January 30, 2020 by Gina

[wimvb]

Nice, Gina. 

Here's a tip for keeping the core of m42 under control:

Create a copy of the image, apply stf as permanent stretch. Then apply curves to bring the white point down to 40% (or pixelmath $T*0.4). This will be a mask. 

To the original unstretched image, apply a series of modest histogram stretches (I put midpoint at 0.25). When you start to lose detail in m42's core, apply the mask and do hdr transformation with nr of layers set to 4. Then remove the mask and continue applying histogram stretch. Repeat this alternating procedure until the background looks good.