Ha vs SII star problems??

[Adam J]

All I am quite experianced with Ha / OIII bi-color imaging and thats my normal thing as I dont typically have the time to capture three channels.

However, I recently got a 5nm Astrodon SII filter to complete my set as I fancied a go at SHO.

This is a comparision between my Ha and SII channels.

The HFR reading is comparable between the two channels at just a little over 2 pixels. However the SII stars seem much much brighter and more numerious.

I did wonder if this might be an effect of the strong Ha in the nebula partially blocking the stars? But I guess I am also a little worried that my SII channel is bloated for some reason....not that bloating explains the larger number of SII stars visible..

Ha

SII

I found this example from Sara Wager that seems to show a similar effect?

Cheers,

Adam

Edited July 28, 2020 by Adam J

[MarkAR]

I had a weird effect when doing Sii on the Crescent, one star was 3 times the size as the Ha subs. When processed it was quite a bright yellow star.

I would put it down to more emissions towards the infra red, red dwarfs and giants maybe.

[vlaiv]

I think it is a "stretch" thing more than anything else.

SII signal is bound to be much weaker than Ha and in order to see it more clearly - you need to stretch that channel harder than Ha.

Since star profiles are close to Gaussian shape - this happens:

If you do mild stretch on same Gaussian profile, you will get small diameter saturated circle - here presented as short cross section. On strong stretch you get much wider saturated segment.

This makes stars in the image look "fatter" and also shows more of them or rather they tend to look denser because of their saturation point and number of pixels they occupy.

What you could do as exercise to show if this effect is real or not is to take your Ha data and your SII data still linear and copy half of Ha image (left half for example) onto SII image thus creating sort of "split screen" in linear data.

Then proceed to stretch that image until you are either happy with Ha half - which will leave SII half almost dark due to SII being very faint, or you are happy with SII side - that will in all likelihood overexpose Ha side. In either case, stars should look more or less the same in both halves (there could be some small differences and those would be due to seeing or filter band pass or optical quality).

[carastro]

I don't use Astrodons, but my Baader filters show larger stars in the Oiii and Sii compared to the Ha.  I think that is normal.  

Carole 

[Laurin Dave]

Hi Adam

That looks the same as I see with my Astrodons (all 5nm), as Vlad says its the harder stretch required on Sii and Oiii that's causing more and larger stars..  

Dave

[Adam J]

Not sure about that I will post a equal stretched image later.

[Adam J]

10 hours ago, vlaiv said:

I think it is a "stretch" thing more than anything else.

SII signal is bound to be much weaker than Ha and in order to see it more clearly - you need to stretch that channel harder than Ha.

This is straight out of APP using the same preset stretch for both Ha and SII. No other processing at all except for a crop.

Ha

SII

So much more similar as you say. the question now is what to do about it as I do need to stretch that SII more to get the desired effect. What do you think on the stars now?

I guess Starnet++ stretch and replace the stars using blend mode lighten?

Adam

Edited July 28, 2020 by Adam J

[MarkAR]

Stars seem to match better, nebulosity can be brought out with extra stretching with a mask.

[vlaiv]

22 minutes ago, Adam J said:

So much more similar as you say. the question now is what to do about it as I do need to stretch that SII more to get the desired effect. What do you think on the stars now?

I guess Starnet++ stretch and replace the stars using blend mode lighten?

Well, you can try two different approaches here.

With Starnet++, this is how I would suggest that you try:

- take all channels and stretch slightly and save as 16 bit images. Remove stars with Starnet++

- take Ha channel and subtract Starnet++ version from regular 16bit version. This should give you stars. If you did initial stretch right - this is about as much as you should stretch your stars or maybe just a tad more.

- Now you can either stretch all three channels to your liking and do RGB combine or you could try LRGB approach (I'll explain that one as second approach without Starnet++). Once you have your image - you simply layer stars on top of it and do some blend mode (like lighten or maybe add layer mask with stars and normal mode - whatever puts stars on top of your image).

That way you should have nice nebulosity with good color and white stars (no annoying purple halos and such).

Second approach is probably a bit more complex but it does not involve Starnet++ (which could be advantage in some cases).

You need to do LRGB composition with RGB in linear stage and synthetic L.

Creating synthetic L is rather difficult to do, but in most cases Ha or Ha+OIII will be enough. SII rarely exists on its own and much more frequently it is in the same place as Ha. This lets you use Ha as luminance.

In any case, once you have luminance - stretch it like a mono image until you are happy with what you have.

Now you need to do RGB combine by applying RGB ratio method. But you need to assign sensible weights to each channel. Usually Ha will have some small weight like 1/16 or so of linear (it will still be linear data only scaled to bring it to same value as OIII and SII). Similarly you might need to scale OIII just a bit to bring it down to SII levels.

Here you are not really stretching your data - just making it "color" compatible. Otherwise it will all be red with slight hue variations (because strength of Ha is so dominating).

RGB ratio is rather straight forward:

Resulting_R = Stretched_L * R / max(R,G,B)
Resulting_G = Stretched_L * G / max(R,G,B)
Resulting_B = Stretched_L * B / max(R,G,B)

not sure how you are going to do that in either PI or PS  or whatever software you are using (should be doable in both of those listed but using different approaches - either pixel math in PI or layers in PS).

 

[johneta]

+1 for MarkAR.  Stretch with a star mask.

 

[PadrePeace]

AdamJ

Post stacking:

I’d:

1. Remove the stars in Straton ( zipproth.com/Straton/ ) in all three layers.

2. As you go through Save and Label the associated filter on the three ‘star only’ images and the 3 ‘starless’ images from Straton.

4. Load up (in PS) and work on / stretch each ‘starless’ image as required. 

5. Once happy with your Ha, OIII, and SII nebulosity work, load up the 3x ‘stars only’ images into PS, convert to R, G, B as is your palette preference. 

6. Convert the 3x ‘starless’ images to R, G, B iaw your chosen palette. 

7. Recombine the ‘star only’ images to make one star layer. If you have any unsightly halos due to different star sizes in overlap there is a good adjustment technique using PS ‘Channels’ below which will help you control and resize stars for each filter so they match up better. https://www.bing.com/videos/search?q=My+most+powerful+astrophotography+processing+secrets&docid=608050129387850451&mid=BDE02CD6764A7FB8D2B5BDE02CD6764A7FB8D2B5&view=detail&FORM=VIRE

8. Layer up all 3 ‘starless’ images together with the combined stars image for final none distructive tweaking as necessary. 

Flatten, Job done...

Straton is/was A one off of £10 if Starnet++ is not your thing.
 

Edited July 30, 2020 by PadrePeace

[johneta]

That's pretty interesting 'PadrePeace'

You have obviously had good results with Straton?    Does it work similarly to Starnet++

I haven't had much luck with Starnet++ in my workflow