Pixinsight - BackgroundNeutralization

[Simon Pepper]

Hi all,

So I am trialling the free license with PI to see how I get on and I am struggling with the BackgroundNeutralization step. Watching some YT tutorials I have stacked and integrated and performed the nuke stretch using the ScreenTransferFunction then people are saying to use BackgroundNeutralization followed by colour calibration and SCNR as the first few steps. However when I complete the BackgroundNeutralization it just wipes out a lot of good data and turns from the narmal green (as I see in tutorials) to a really hevily contrasted image. Am I missing a step here? I am sure the data is good as I have processed in PS and got a decent result. Pretty sure its 4.5 hours of lights plus calibration frames. Any help from those with a PI degree or not will be appreciated

See snips of before and after BackgroundNeutralization

 

 

After the colourcalibration and SNCR it gets even darker and loss of more signal. 

 

Thanks

[geeklee]

Hi Simon - typically after a BackgrounNeutralization, you may need to apply ScreenTransferFunction again.  Give this a go first and report back 👍

Remember, in PI the STF is non destructive, just an aggressive view of what the data could look like stretched.

Edited April 26, 2021 by geeklee

[scotty38]

If you deselect the "locked" channels in STF I think that will cure your problem...

 

Edit oh sorry that wasn't your question, ignore me lol

Edited April 26, 2021 by scotty38

[geeklee]

1 minute ago, scotty38 said:

If you deselect the "locked" channels in STF I think that will cure your problem...

That's a good way to view without that green cast at the beginning with STF though 👍

[Simon Pepper]

I have a lot to learn here... Ok so I did that and the image looks a lot better a few gradients but I am sure I can find a yt tutorial on that. Quick question then when using the STF tool do you need to press the nuke button after each step through other screens like colourcalbration, backgroundneutraliasation I thought all that was needed was to use the triangle and drag and drop onto the image?  

 

Thanks

 

[geeklee]

13 minutes ago, Simon Pepper said:

Ok so I did that and the image looks a lot better a few gradients but I am sure I can find a yt tutorial on that. Quick question then when using the STF tool do you need to press the nuke button after each step through other screens like colourcalbration, backgroundneutraliasation I thought all that was needed was to use the triangle and drag and drop onto the image?  

Have you utilised one of the background extraction tools before the BackgroundNeutralization & ColourCalibration?  This might be AutomaticBackgroundExtractor (ABE) to start with or DynamicBackgroundExtraction (DBE) if you're comfortable with that.  

The STF is acting on the pixel values at the time of application so anytime these change - for example a BackgroundNeutralization, ColourCalibration or Background Extraction (ABE/DBE) then STF can be reapplied.  Really, anything being changed while the image is still linear will affect that original STF.  An obvious one will be noise reduction (NR) - you'll apply NR, it'll look fine, then you'll apply STF again and wonder what's happened - always remember the STF is very aggressive and just applying a later stage STF stretch isn't always the best way forward (although great to test with when you're learning - you can still adjust it while its applied in HistogramTransformation).

The triangle you mention applies the process in question (e.g. ColourCalibration).  At this point the pixel values have likely been changed and the STF applied is based on older values.

Hope this helps.

Edited April 26, 2021 by geeklee

[Simon Pepper]

Hi Lee, 

Nope all I did was nuke it using STF then dove straight into the BackgroundNeutralization. I will have a look around to see if I see the background extraction and try running that first. 

I think all of your comments make sense  definitely a lot to take it and learn, but will definitely stick with it.

Thanks 

[geeklee]

7 minutes ago, Simon Pepper said:

I think all of your comments make sense  definitely a lot to take it and learn, but will definitely stick with it.

If you'd like any clarification, just come back - there are plenty of PI users on SGL 👍

Not sure if you have any of these, but books like "Inside PixInsight" and "Mastering PixInsight" both contain great "starter workflows".  These are well described, straight forward work flows that let you utilise the basic tools to get to a point where you're ready to tackle more.  Websites like Light Vortex Astronomy also contain things like this.

The background extraction came to mind as you mentioned you still had some gradients.

Your image above is looking good BTW, just with the bits you've done already 👍 🙂

Edited April 26, 2021 by geeklee

[Simon Pepper]

Perfect will do will check out those books too

[scotty38]

Just to note that I am massively new to PI too and Lee has helped me out before too!

If you look at the STF tool just to the right of your screenshot there's a "PC" type icon I think that "enables/disables" the STF settings so you can always  see what your real image looks like if that makes sense.

Don't forget if you have it enabled and then apply any histogram settings don't panic when it suddenly becomes washed out as you need to then disable STF as you're now stretching twice. Obvious when you realise but it caught me out 🙂

 

Oh I have those books too as I'm intending buying PI when my trial runs out.

Edited April 26, 2021 by scotty38

[Adreneline]

20 minutes ago, Simon Pepper said:

definitely a lot to take it and learn, but will definitely stick with it.

It is soooooo worth sticking with it Simon - it can seem daunting and so unlike any other photo-processing / manipulation software but IMHO no other piece of software has the ability to extract as much information from your hard won data as PI.

Good luck.

Adrian

[Simon Pepper]

Perfect will do will check out those books too  Is there a toggle in PI that shows once you have completed a step lets say TGVDenoise to toggle on and off the change so you can see what its applied? In PS you can select that layer and then deselect to quickly see if there is an improvement or not?

 

Thanks 

[scotty38]

Image/Undo Image/Redo should do what you want, it's how I do it although there may well be other  ways 

[Simon Pepper]

10 hours ago, scotty38 said:

Image/Undo Image/Redo should do what you want, it's how I do it although there may well be other  ways 

Ah yes ctrl z and ctrl y... Not ideal a button would be quicker but that does work thank you

[TerryMcK]

Just a minor one don’t apply SCNR to the linear image. Do it post histogram transformation ie when the image is permanently stretched or non linear.

[geeklee]

27 minutes ago, TerryMcK said:

Just a minor one don’t apply SCNR to the linear image. Do it post histogram transformation ie when the image is permanently stretched or non linear.

Why do you not do it in the linear stage Terry?  I've seen tutorials that utilise it in linear and others non-linear - depending on data.  Most I've seen say it's a step after colour calibration (i.e. in the linear stage).  

[TerryMcK]

Hi Lee. Although it can be done at any stage I've had much better results leaving it very late in the processing just to remove any extraneous green pixels. When I've done it pre-stretch it has actually removed fainter data.  

[geeklee]

1 minute ago, TerryMcK said:

Hi Lee. Although it can be done at any stage I've had much better results leaving it very late in the processing just to remove any extraneous green pixels. When I've done it pre-stretch it has actually removed fainter data. 

Thanks Terry.  I guess it's also easy to dump the default process on an image whereas I usually tend to experiment with different amount and not just "1.00".  Usually I end up with much less, but still sometimes end up taking some green out later anyway 🙄 

It's always best to experiment with these things especially as SCNR is typically linked to colour calibration.

[TerryMcK]

Here are the two settings I use:

This one works well on overall greenish cast images. It also works a treat if you invert a starmask created with Starnet to remove green around stars. When you invert the image back to normal you see that the red halos have gone.

 

Whereas this one eliminates green pixels/cast in the nonlinear state while preserving the teal colour of OIII in planetary nebulae and supernovae remnants.

 

Play around with the amount in either instance.

[Simon Pepper]

On 27/04/2021 at 20:57, scotty38 said:

Image/Undo Image/Redo should do what you want, it's how I do it although there may well be other  ways 

Loving it already the splitting out RGB channels and playing with PixelMath and LRGB tool is awesome although its easy to get carried away with those exotic pallet combinations! Will definitely be ordering the license though. Not sure the Mrs is overly pleased though she doesnt see me anymore  

[Simon Pepper]

Thanks for all the tips!

[Simon Pepper]

Hi Guys,

So this is what I came up with IC405 from stacking to edit all in PI. I followed some yt videos and took your advice above! I think its come out ok I can see a hot pixel snuck through and I did have issues with gradients but not sure if thats the dust of the nebula showing up. Its definitely better than the edit I did in PS so thats a win for PI! Ideally after some feedback please how you think it has come out where it can be improved? Not sure if you can view in .xisf, so added jpg and tif just in case. 

Thanks 

IC405PI.xisf IC405PI.tif

[Simon Pepper]

Just now, Simon Pepper said:

Hi Guys,

So this is what I came up with IC405 from stacking to edit all in PI. I followed some yt videos and took your advice above! I think its come out ok I can see a hot pixel snuck through and I did have issues with gradients but not sure if thats the dust of the nebula showing up. Its definitely better than the edit I did in PS so thats a win for PI! Ideally after some feedback please how you think it has come out where it can be improved? Not sure if you can view in .xisf, so added jpg and tif just in case. 

Thanks 

IC405PI.xisf 134.15 MB · 0 downloads IC405PI.tif 133.86 MB · 0 downloads

Looking at it as jpg and at this scale its a little blue and possibly some horizontal banding?

[scotty38]

Looks great that does and a huge improvement!

[geeklee]

On 05/05/2021 at 19:58, Simon Pepper said:

Ideally after some feedback please how you think it has come out where it can be improved?

Definitely a big step forward compared to the images in the initial post - great job.  If you hadn't posted the PI screenshot in post 5 I'd have left it there..... but.... 😁

I'm not sure what stage you were at in PI when you posted the screenshot - colour calibration by the process open? (early anyway).  I think the stars and highlights have started to become blown out.  If I compare the largest star in the flame nebula, the larger ones around and the highlights in IC410 - larger stars, whiter stars and the highlights starting to go white.  The earlier image has a more balanced stretch and colour with smaller stars + a more neutral if dark background and still loads of faint Ha in the background.  I think there may have been finer structure in the nebula as well (maybe a little too much saturation & noise reduction?) but the small size in the screenshot might be tricking me.

The only way to check is keep an eye on the early image - you can use "History Explorer" to go back in time or just keep a cloned image off to one side maybe (just drag and drop the image identifier onto a blank part of the workspace).

Hopefully I'm not taking anything away from what you've done - you should be really pleased with where you've taken the image from the initial post!

Edited May 6, 2021 by geeklee