Horse Example - How many Subs

[PKay (Peter)]

I sorta feel like I am standing on a thin branch at the moment :-)

Anyway I combined two sets of images in PI, each integrated separately.
Did the normal background extraction, noise reduction, histogram stretch.

Set 1: 30 of Gain = 50, exp = 240s
Set 2: 30 of Gain = 70, exp = 120s

And yes, I did the photoshop thing pushing up the saturation a bit just for fun.

Admittedly this is a nice bright ish target.

I can only guess if I added another 30 images at exp = 30s even more colour would reveal.

As Colin pointed out, it might help in this debate to offer some examples.

[cometcatcher (Kevin)]

Originally Posted by PKay

Admittedly this is a nice bright ish target.

Sorry, I had to lol a bit when I read that. This used to be considered a faint target, way back. Nice image anyway.

However for sake of demonstration would it not be better to compare an image of 120 x 60 secs to 240 x 30 secs or 480 x 15 seconds? Same total integration time, different sub length.

[raymo]

Lovely,doesn't need any more colour.
raymo

[Anth10 (Anthony M)]

That's pretty good Pkay,
More subs trumps less subs.

Anth

[ChrisV (Chris)]

Really nice Peter. And agree, it would be nice to see a comparison of sub lengths.

Where are you guys getting the weather to do anything? I've got less than one hour of diddly squat over the past few months. I might have to go mono/narrow band to expand the days of the month I can image to catch any sort of clear skies. I'm really starting to get grumpy (I'm already getting old) ...

[xelasnave]

Great Photo Peter and I must say I have really enjoyed your essays.

All I have worked out is I need a better computer to manage more subs
My lap top has been stacking flat out for near two days...so long I have lost track of time.
So for me it seems I will need to take longer subs in any event because of the limitation placed on me by deep sky stacker.
I only have 8 gig of ram...
I took your advice of varying iso ...anyways trying yet another stack smaller si hopefully I get something that justifies the time I have spent.
Alex

[Stonius (Markus)]

Originally Posted by PKay

I can only guess if I added another 30 images at exp = 30s even more colour would reveal.

Oh, why is that? I wonder why shorter 30s subs are better for colour? Is that to capture colour in the brighter parts?

BTW, Great shot of one of my favourite objects. I can only just make it out in my 16" with a UHC filter.

Cheers

Markus

[N5SE (Sal)]

Beautiful photo

[ChrisV (Chris)]

That's really nice Peter

[PKay (Peter)]

Thanks heaps for the encouragement!

At times we need it to keep going

[xelasnave]

Hi Peter
I plan to try a rather high iso and appropriately short exposures for a bit of fun.
Maybe 6400 and 10 seconds ...whatever the sky glow limits it to.
I now realise a mount not perfectly aligned is really just having built in dithering☺
Do lots of darks see how bad the noise ends up.
Alex

[PKay (Peter)]

that's the go Alex

I think at this stage, time is better spent playing around with different settings, and taking less images (about 30 is ideal).

Taking 100's of subs (at one setting) is more for the experienced who know what they are doing.

Shame to take 250 subs to find your focus was off...

[Zubenel (Wes)]

I’d be be sooo happy with that one !! Top shot PKay

[Atmos (Colin)]

There are some issues that you will run into when stacking frames that are different, whether it be different exposure times or different Gains. When you are stacking frames you need each pixel to be under the same circumstances.

If you stack 30x120s at Gain 76 and then a new stack of 30x60s at Gain 139 you won’t have an issue as they’re seperate. If you stack them all at once you are going to have one stack of 30 that may have an ADU value of 500 and another stack with an ADU value of 750 for the same pixel.

This is exactly the same principle as if you were to stack a set of RGB frames with rejection to create a Synthetic Luminance... it just doesn’t work properly.

My longest integration for a single frame and filter at 10.25 hours (123x300s Ha). This data was collected over 6 nights so it’s not like I captured all 123 of them at once so as to have focus issues. I refocused every 20 or so frames. Now the difference between stacking 30x300s (2.5 hours) and 123x300s is stark. With 10+ hours of Ha there is no need at all to do noise reduction, even on the faintest regions within that frame because even that is smooth. At 2.5 hours some noise reduction is needed for anything other than the bright areas.

Shooting 30x120s, 30x200s, 30x300s & 30x600s isn’t of any real use. The first 30 aren’t long enough to swamp readnoise. The second 30 are getting over but not as much as I’d like. The third 30 are ideal and the forth 30 are going to blow out a lot of the highlights without going any deeper than doing 60x300s.

With the example above, the first 60 frames make your image noisier in the background and your last 30 are overblown in all the brighter regions.

[PKay (Peter)]

Hi Colin

Once again valuable information.
However, the distinction between image integration and image combination must be made clear.


With regards to the 30 image limit I found a quote:
'An oversimplification for the novice, confusing for the experienced and ignored by the professional'.

I did put the question to the PI team and asked who was right? (about 30 images vs say 300).

The answer that I like: 'Both are'.

I am trying to nut out yet another attempt to explain (mainly for myself) the ideal number of subs.

This time I will break it into two areas.
1: The Novice
2: The Experienced.

For the novice, it must be kept simple and easily understood.
For the experienced...they make up there own minds, so it will just be light reading

[Atmos (Colin)]

In a way there is no “ideal number” as it depends what it is you’re wanting to achieve.
30 subs will give you a SNR increase of 5.48
300 subs will give you a SNE increase of 17.32

On a brighter nebula region that has a SNR of 80 on a single image, that would increase from 80 to 438 where as 300 would increase it to 1,386. Both are quite high so 30 frames would be fine.

On a faint Nebula region that has a SNR of 1.125 meaning it is a detected signal buried in noise. While stretching, it would look like noise and not being able to separate from the background. With 30 images you would have 6.165 SNR meaning it would be present but not easily resolved from the background. With 300 images it would be 19.485 SNR which would make it clearly seperatable from the background.

Using the example from my previous post, the 120s and 200s exposures would increase the SNR of the brighter regions while burying the fainter regions in more noise.
The 30x600s exposure would damage some of the brighter regions (saturation) while not giving a better SNR in the dim regions than 60x300s for the same integration time.

[PKay (Peter)]

C

Straight off the bat (topical at the moment)...

Maybe for the novice, best to avoid feint targets.

I tried a few times (Witches Head, Seagull, Running Chicken) and found it too difficult and at times frustrating.

As an aside I did see your mammoth effort of the Chicken (Deep Space) and very well done.
Such a task is totally out of reach for me at this stage.
Maybe in a year or two...