Whats the green stuff?

[simmo39]

Hi, first off sorry if I have posted this in the wrong place and feel free to move it. Anyway I have been doing a bit of RGB imaging on M 51, this is only my second attempt at rgb and usually do narrow band. Below is a first stretch of the combined data and although I can get rid of most of the green blocks I just wondering what is causing them. Is it frost, internal reflections or something else?  I have manged to get rid of the worst of it on some of my images of M51 and the green didnt show on my first RGB target. Any tips pointers or answers welcome.

[Craney]

Only a guess.... maybe some cloud drifting across  during your Green exposures ??  and it has been included in the stack.

[simmo39]

2 hours ago, Craney said:

Only a guess.... maybe some cloud drifting across  during your Green exposures ??  and it has been included in the stack.

Hi Craney, I had 30 subs of green. I would have thought that the stacking would have got rid of it.

[Craney]

Fair point...

[alan4908]

On 05/05/2019 at 12:35, simmo39 said:

Hi, first off sorry if I have posted this in the wrong place and feel free to move it. Anyway I have been doing a bit of RGB imaging on M 51, this is only my second attempt at rgb and usually do narrow band. Below is a first stretch of the combined data and although I can get rid of most of the green blocks I just wondering what is causing them. Is it frost, internal reflections or something else?  I have manged to get rid of the worst of it on some of my images of M51 and the green didnt show on my first RGB target. Any tips pointers or answers welcome.

I'd guess that the green (and purple) background colours are most likely caused by light pollution.  Do you image from a light polluted area or when you moon was up ?  

Anyway, you can easily remove them in post processing. The best tool to do this is Pixinsight's Dynamic Background Extraction process. If you process in PS then I'd recommend using the Gradient Exterminator plug-in. An alternative to reducing green in images is to use the Pixinsight process SCNR (green) or in PS the plug in HLVG. 

Alan

[simmo39]

3 minutes ago, alan4908 said:

I'd guess that the green (and purple) background colours are most likely caused by light pollution.  Do you image from a light polluted area or when you moon was up ?  

Anyway, you can easily remove them in post processing. The best tool to do this is Pixinsight's Dynamic Background Extraction process. If you process in PS then I'd recommend using the Gradient Exterminator plug-in. An alternative to reducing green in images is to use the Pixinsight process SCNR (green) or in PS the plug in HLVG. 

Alan

Hi Alan. Yes I'm a Pi user, I have used all of the above and for the most part removed them from the final image. Where I shoot from is not to bad for light poullution at the moment as the main Street light that causes me problems is out of action. I was wondering if it was light pollution but I didn't get it on my first RGB image and nothing has really changed between the two images. 

Thanks Andy.

[vlaiv]

I think it's the mix of all the nasty stuff . LP gradients and probably high altitude very faint clouds.

23 hours ago, simmo39 said:

Hi Craney, I had 30 subs of green. I would have thought that the stacking would have got rid of it.

In general, even if you have a single frame with passing cloud it will impact final stack result. Impact will lessen with number of stacked frames, but it will be there, and if you stretch your data enough it will show.

Easiest way to see what is going on is to create little animated gif of very stretched subs for each channel - it will show any transient features like passing clouds. Take each calibrated sub, bin it to get rather small image (something like 300x200px or close to that - you don't want to see detail, you want to examine whole image at once, and this will improve SNR quite a bit). Stretch equally each sub - you can even do simple linear stretch, no need for fancy histogram manipulation, and then create animated gif out of such subs to have "blinking" effect between subs - it's far easier to spot the difference if it "moves".

Probably simplest way to get rid of such background is via control of black point and creating mask for desaturation. If your luminance layer is clean and you don't have gradients on it, you can invert it and use it as layer mask on desaturated image to blend in. Darker background will control color (if color is bright it is easily distinguished between hues, but if it's dark - all colors look the same in the dark :D).

[simmo39]

2 minutes ago, vlaiv said:

I think it's the mix of all the nasty stuff . LP gradients and probably high altitude very faint clouds.

In general, even if you have a single frame with passing cloud it will impact final stack result. Impact will lessen with number of stacked frames, but it will be there, and if you stretch your data enough it will show.

Easiest way to see what is going on is to create little animated gif of very stretched subs for each channel - it will show any transient features like passing clouds. Take each calibrated sub, bin it to get rather small image (something like 300x200px or close to that - you don't want to see detail, you want to examine whole image at once, and this will improve SNR quite a bit). Stretch equally each sub - you can even do simple linear stretch, no need for fancy histogram manipulation, and then create animated gif out of such subs to have "blinking" effect between subs - it's far easier to spot the difference if it "moves".

Probably simplest way to get rid of such background is via control of black point and creating mask for desaturation. If your luminance layer is clean and you don't have gradients on it, you can invert it and use it as layer mask on desaturated image to blend in. Darker background will control color (if color is bright it is easily distinguished between hues, but if it's dark - all colors look the same in the dark :D).

Hi Vlaiv. Thanks for the pointers. I will have to sit and digest all you have said and see if I can do it. The M 51 image is not one of my favourites and I'm was only shooting it to pass time till the nebula return. I'm just hoping that by then I will have sorted these problems. Once again thank you for the hints. 

[alan4908]

56 minutes ago, simmo39 said:

Hi Alan. Yes I'm a Pi user, I have used all of the above and for the most part removed them from the final image. Where I shoot from is not to bad for light poullution at the moment as the main Street light that causes me problems is out of action. I was wondering if it was light pollution but I didn't get it on my first RGB image and nothing has really changed between the two images. 

Thanks Andy.

OK - one thing you could also try to minimize any transient effects is to use the local normalization stacking option in PI.  It's a way of minimizing large scale transient effects provided that they only occur in certain subs. 

To do this you:

1. First create a reference frame for each channel by the normal stacking method. Here, choose only those subs which do not contain any transient effects eg high clouds.

2. Use the Local Normalization process to create local normalization files and use the reference frame that you've just created.

3. Stack all the subs in a particular channel, add the local normalization files and change the normalization stacking option to local normalization.

 Alan 

Edited May 6, 2019 by alan4908

[simmo39]

36 minutes ago, alan4908 said:

OK - one thing you could also try to minimize any transient effects is to use the local normalization stacking option in PI.  It's a way of minimizing large scale transient effects provided that they only occur in certain subs. 

To do this you:

1. First create a reference frame for each channel by the normal stacking method. Here, choose only those subs which do not contain any transient effects eg high clouds.

2. Use the Local Normalization process to create local normalization files and use the reference frame that you've just created.

3. Stack all the subs in a particular channel, add the local normalization files and change the normalization stacking option to local normalization.

 Alan 

Thanks Alan. I will give that ago. 

[don4l]

This isn't much help, but my images are exactly the same.  I think that vlaiv is correct.  It is probably a mix of LP with other stuff.  I did some imaging from a very dark site once, and they didn't have this problem at all.

[simmo39]

2 hours ago, don4l said:

This isn't much help, but my images are exactly the same.  I think that vlaiv is correct.  It is probably a mix of LP with other stuff.  I did some imaging from a very dark site once, and they didn't have this problem at all.

Thanks, hope you are right.

[david_taurus83]

Using Blink in PI is a good way of removing cloud polluted subs. Just make sure you apply a histogram transformation to the subs. When it loads it performs a stretch to the first sub on the list and if you don't change it it will just carry through the same stretch over them all. Press the top button to perform HT. Btw, it's only temp, it doesn't alter your raw data.

[simmo39]

6 hours ago, david_taurus83 said:

Using Blink in PI is a good way of removing cloud polluted subs. Just make sure you apply a histogram transformation to the subs. When it loads it performs a stretch to the first sub on the list and if you don't change it it will just carry through the same stretch over them all. Press the top button to perform HT. Btw, it's only temp, it doesn't alter your raw data.

Thanks for the reply. I will sit and have a play tomorrow, hopefully I will get something sorted from the data.