Background sky noise reduction, an idea.

[ollypenrice]

I thought of this fairly recently and have found it quite useful. Of course everyone else may have been doing it for years so ignore me if this is so!

It concerns background sky noise found in hard stretches. Here's an example at pixel scale. The lighter parts of the background are recording 25 in Curves as you can see. 'Noise' is that which is below 25.

If you carefully pin the Curve above 25 and then give it a lift below 25 you brighten the darker pixels which, in this case, are the noise. Some, which were close to 25, will be lifted to 25 giving a flat look, but some low level variation will be left, giving a reasonably natural look. The nature of the 'lift' you give the low values controls the extent of noise reduction and it can be further controlled using Edit-Fade.

This is a quick demo example, not an attempt to find the exactly perfect lift. It might be too much here once zoomed out to 100% but you get the idea. It can give very unintrusive NR, in my view.

Olly

 

[Tim]

I like the remaining low level variation, as you say a certain amount of background interest is required for the natural look.

Do you use the cosmetic correction routine during stacking in Pixinsight at all? It achieves a similar result but probably requires the same amount of user intervention.

Skies here so poor recently that we're in danger of forgetting how to capture an image, let alone process one

[ollypenrice]

23 minutes ago, Tim said:

I like the remaining low level variation, as you say a certain amount of background interest is required for the natural look.

Do you use the cosmetic correction routine during stacking in Pixinsight at all? It achieves a similar result but probably requires the same amount of user intervention.

Skies here so poor recently that we're in danger of forgetting how to capture an image, let alone process one

What I like about this is precisely that it does conserve some of the noise and in the same proportions as is existed before the intervention. I don't stack in PI but in AstroArt. (I'm frightened of PI!) I use AA's hot pixel filtration and column repair while stacking but it doesn't offer any NR at that stage.

I gather the UK skies have been a pain. Ours have not been great, though we had a fully clear week around new year and have had a steady flow of clear ones since then. A flow rather than a torrent, that is! The days have been frequently sunny, though, and un-naturally warm. Rather nice.

Olly

[Petergoodhew]

Thanks Olly.  I recall you showing me something along these lines at Les Granges and I've been struggling to recall exactly how you did it.

I will definitely be using it in future.

Yes permacloud has set in over West London.  Will have to clear the cobwebs from the scope if it ever clears.  I'm desperate for some clear skies so heading to Tenerife in a couple of weeks to seek some out.

[wimvb]

I've used this in some images a while ago when doing final curves transformations (PI). If you make the curve horizontal under the histogram peak, you can get a very flat background, and virtually "eliminate" noise. But you will also flatten any structural detail in, say, a dark nebula or a dusty region of a galaxy. So you would need to use it with very careful masking to get it to work. That's where I dropped the idea at the time.

To add to your proposed technique, you could leave the slope of the original curve the same, but pin the curve at the lower side if the histogram. Then pull down the curve at the higher side. This should leave the "dark side" (no pun intended) unaltered, and create a very mild increase in contrast on the "bright side". I don't know if possible in PS, but in PI, you can introduce an offset in histogram / curves transformation, to lift the overall background of the image.

If noise reduction were this easy, why invent "total generalized variation noise reduction" when you can just squeeze the histogram?

[Tim]

Next time it is cloudy, take a gander at Kayron's PI tutorial for stacking. The tute is over helpful if anything, but pretty comprehensive, and the closest I have come anywhere near understanding PI, and I have had the full version since the beginning!

http://www.lightvortexastronomy.com/tutorials.html

[ollypenrice]

5 hours ago, wimvb said:

 

If noise reduction were this easy, why invent "total generalized variation noise reduction" when you can just squeeze the histogram?

This isn't a global noise reduction technique, it's a technique applicable to background skies in certain circumtances and controllable by the user. The whole point of it is that it does not look like NR becaue it is image specific. If you prefer 'total generalized variation noise reduction' then use it! You are right about genuine low level signal (due to dust, etc.) needing masking. As for the use of language in PI, why call Screen Stretch 'Screen Transfer Function?' Fifty years ago we had a laugh in my class about NASA speak. We decided that the lunar lander had an 'Extra Vehicular Lunar Surface Descent Module' on the outside. A ladder.

If I ever get to work for PI I'm going to invent 'Target Orientated Screen Format Composition Resolution Optimization,' currently lingering under the confusing term of 'framing.' 

Olly

[Tim]

Somebody posted a link to some Nikon Photoshop tools recently, free ones. Some of the NR in there is useful sometimes.

[wimvb]

@ollypenrice: I forgot the  after the last paragraph in my previous post; or rather, I implied it.

As for the PI jargon, there is a discussion on that burried somewhere in the PI user forum. I just stick to the acronyms, and use the tools to the best of my ability.

52 minutes ago, ollypenrice said:

 

If I ever get to work for PI I'm going to invent 'Target Orientated Screen Format Composition Resolution Optimization,' currently lingering under the confusing term of 'framing.' 

Olly

I thought that function was just called 'Target Oriented Composition Resolution Optimization Procedure', or TO-CROP for short. ()

Cheers,

[Gina]

The nomenclature is the worst thing about PI IMO.  It's as if they deliberately want to put new users off   Or is it just elitism?!

[brrttpaul]

I found that to be the biggest problem also in PI, cant understand why the different names across the board for basically the same thing, getting there now but it takes a while

[ollypenrice]

10 hours ago, wimvb said:

@ollypenrice: I forgot the  after the last paragraph in my previous post; or rather, I implied it.

As for the PI jargon, there is a discussion on that burried somewhere in the PI user forum. I just stick to the acronyms, and use the tools to the best of my ability.

I thought that function was just called 'Target Oriented Composition Resolution Optimization Procedure', or TO-CROP for short. ()

Cheers,

No wories, we English enjoy our ironies! The smiley was understood.

Olly

[iapa]

22 hours ago, ollypenrice said:

This isn't a global noise reduction technique, it's a technique applicable to background skies in certain circumtances and controllable by the user. The whole point of it is that it does not look like NR becaue it is image specific. If you prefer 'total generalized variation noise reduction' then use it! You are right about genuine low level signal (due to dust, etc.) needing masking. As for the use of language in PI, why call Screen Stretch 'Screen Transfer Function?' Fifty years ago we had a laugh in my class about NASA speak. We decided that the lunar lander had an 'Extra Vehicular Lunar Surface Descent Module' on the outside. A ladder.

If I ever get to work for PI I'm going to invent 'Target Orientated Screen Format Composition Resolution Optimization,' currently lingering under the confusing term of 'framing.' 

Olly

I thought it was a Lunar Accessible Directional Descent Equipment - Reusable

I remember Messrs Armstrong and Aldrin's [accept possible incorrect position of apostrophe] usage such a device. 

[ollypenrice]

11 hours ago, iapa said:

I thought it was a Lunar Accessible Directional Descent Equipment - Reusable

I remember Messrs Armstrong and Aldrin's [accept possible incorrect position of apostrophe] usage such a device. 

Correct apstrophe. Go to the top of the class!