When is enough really enough?

[Jessun]

Did numerous searches on this but didn't stumble on a good answer. 'Silly question' however came up with 500 hits. Feel free to sort this one under 'Silly':)

My question: When do you have enough subs?

Reason for my question: Having added a few extra hours to M81/M82 I found that no additional features could be fished out nor was there any real increase in quality, i.e. noise remained similar.

The subs were all of the same length and I do suffer from light pollution.

My question expanded: Will you at some point with any image run into a point where adding more just leaves you threading water. Since we are using average stacking or slightly more advanced versions of it I can imagine reaching a point when averaging in more data just keeps confirming what has already been captured. The average of 1 and 3 remains 2, just again and again.

I wonder if it would be worth finding this point, and then swapping to longer subs to possibly increase the return on the faintest ROI - perhaps as a separate layer.

In stacking, as the amount of data builds up, do you change the stacking settings, like the Kappa value - which I believe is a median rejection number - confident that that you have enough data to filter out what you want?

Likewise for noise, and let's call the background noise if indeed it's really just black. Will adding more subs bring the noise down only to find a point where adding more subs actually starts building up increased overall noise again?

Spin-off: Do you sometimes find yourself saying after a massive imaging run that you go 'I should have gone for 10 instead of 5 minute subs on all this..."

All else is as far as I can tell pretty much nailed down doing this hobby - apart perhaps for the f-ratio conundrum and reflectors vs refractors - but the sub lengths and the number of them are IMHO not nailed entirely but resides in the 'it depends' category.

I don't see myself belonging to the category of experienced imagers that I ask here at all, I'm merely thinking beyond the capability of me and my kit, curious on pretty much the theory and maths involved.

[ollypenrice]

I'm not a mathematical imager, as you know, but from what I can remember the real flattening of the beneifit curve in S/N ratio comes in the forties of subs.

I'm very 'hands on' and because of my work I often end up imaging the same thing twice or several times. A few yeas ago I did two 5.5 hour images of M42, a bright object, one in an OSC and one in a mono with filters, both at F5.3. This was for a magazine. For myself, I naturally combined them and the combined image was even nicer to work on than the singles. I could stretch harder and found little regions of dusty nebulosity. The noise was reduced. It was a peach of an image to process. This year I added 8 hours of Ha and a solid dose of LRGB at F3.9. It is still going and ain't over yet!

I've just done the same thing with the Leo Triplet. I had about 17x15 luminance images from last year and added the same again this year, all at F7. The double stack is certainly better, and instantly recognized, once you get your hands on it in Ps. The soft outer haloes are larger and smoother. The tidal tail of the Hamburger galaxy is stronger and a tidal extension to its upper right has doubled in size. The amount of sharpening it will take in the brightest parts has not changed significantly, so there the benefit is 'maxed out' I think. What will another four or five hours add? I'll tell you next time a guest wants to do the Triplet!

That's a 'hands on ' contribution to the answer. Those of a mathematical bent will come in with their own perspective.

Regarding LP, the trouble is that the camera is going to see that as real signal, which it is, so whether longer and more subs will help the imager separate wanted signal from not wanted signal strikes me as unlikely. I have no experience of LP and don't want any!

Olly

[Jessun]

Hands on works a treat Olly, that's a fact! And now you routinely haul in 30 min subs...

I believe I hit the S/N roof fairly quickly with this pesky LP around, BUT the first addition about 100 minutes elevated what I started at yours (in full Moon) by a bit. (This re-stacked and re-processed image is in another thread that doesn't really deserve bumping)

I see how around 40 subs flattens the curve. I reckon that matches darks / flats etc in terms of diminishing return.

I'm starting to feel that the sub length is much more important that I initially gave it cedit for - being somewhat limited in LP. You pretty much commit in a sense that in chosing a sub lenght for a run the theoretical upper 'limit' for what you'll extract is set in stone.

This means that my 81/82 for the environment and equipment I am working with has already maxed out.