9 replies to this topic

[pgseye]

Hi All,

 

Can someone help me to understand the basics of how channel extraction and mapping in Pixinsight (and I presume Siril), works? If you're shooting in mono, then clearly you load up each filter's images and assign them to their respective RBG channel. But what happens if you do channel extraction on OSC broadband data (apologies if this should be obvious to me, but it's not)? Does the software automatically assign wavelengths (via pixel values) to a channel? e.g. reddish pixels representing wavelengths in the 600-700 nm range to red, greenish pixels in the 500-600 nm range to green, etc.

 

What about dual (or tri) OSC narrow band data? There should then only be 2 (or 3) narrow spikes representing Ha, Oiii (and Sii) , which if you split into RGB channels gets assigned how?

 

I'm trying to understand because I really like some of the different colour palettes (especially Hubble) and would like to be able to emulate this (I know it's not colour-accurate as in shooting in mono might provide) with a colour camera. I see that you can almost get to true SHO (that a mono setup would provide) with some of the new filter sets using OSC (e.g. Askar D1/D2) - albeit faux-SHO.

 

If you've got any good resources to point me towards on colour palettes in astrophotography, I'd be appreciative.

 

Thanks,

 

Paul

 

 

[kathyastro]

Channel extraction does not know or care about wavelengths.  It simple gives you the R, G, and B channels as three separate monochrome files.  You can then do with them what you want.

[prookyon]

 

 

What about dual (or tri) OSC narrow band data? There should then only be 2 (or 3) narrow spikes representing Ha, Oiii (and Sii) , which if you split into RGB channels gets assigned how?

 

If you have a dual or tri narrowband filter in front of OSC camera then those narrow spikes will register only on pixels where the OSC built-in bayer pattern filter also lets that same wavelength through.

The sensor does not know anything about the wavelength - so for example H-beta and OIII that are both bluish-greenish and very close in wavelength will be indistinguishable for the sensor. There is no way to tell how much of the signal came from one and how much from the other.

So practically speaking the only filters that make sense in front of OSC are those where wavelengths are so far apart that realistically only one color (either R, G or B) will pick up each wavelength.

Usually that means Ha / OIII and S / OIII filters since both of those pairs are such that one wavelength is only registered by R pixels and the other by G and B pixels.

[Zambiadarkskies]

If you have a dual or tri narrowband filter in front of OSC camera then those narrow spikes will register only on pixels where the OSC built-in bayer pattern filter also lets that same wavelength through.
The sensor does not know anything about the wavelength - so for example H-beta and OIII that are both bluish-greenish and very close in wavelength will be indistinguishable for the sensor. There is no way to tell how much of the signal came from one and how much from the other.
So practically speaking the only filters that make sense in front of OSC are those where wavelengths are so far apart that realistically only one color (either R, G or B) will pick up each wavelength.
Usually that means Ha / OIII and S / OIII filters since both of those pairs are such that one wavelength is only registered by R pixels and the other by G and B pixels.

Exactly. This is why there is no such thing as an SHO triband filter for osc that can split H, S,O as hydrogen and sulphur 5ii are both "red" and cannot be distinguished in one filter.

Edited by Zambiadarkskies, 19 May 2025 - 06:45 AM.

[Zambiadarkskies]

Also, sho captured with narrowband filters with OSC are not somehow faux. So long as the discrete wavelengths are captured, it is no different to mono except perhaps in efficiency and integration time.

[BQ Octantis]

Kathy is right…it is the sRGB color space that has mappings to dominant wavelength and excitation purity. But sRGB has no idea what the bands were that you used for the data or how you decided to dump it into its channels…

 

BQ

[kathyastro]

Ha is red, so it gets dumped into the R channel.  Sii is also red, so it, too, gets dumped into the R channel.  Likewise a stop sign is also red, so it gets dumped into the R channel.

 

The R channel just has brightness values.  It doesn't know where they came from.  As far as the R channel is concerned, Ha, Sii and the stop sign are all the same.

 

Likewise for Hb, Oiii, and grass in the G channel.

 

"False colour" is only false because we override the default channel mapping that is done by the Bayer matrix in OSC cameras.  You can manually put the Ha data into the G channel, and the software won't complain.  This is routinely done with SHO (Hubble palette) mapping.

Edited by kathyastro, 19 May 2025 - 07:57 AM.

[pgseye]

Ha is red, so it gets dumped into the R channel.  Sii is also red, so it, too, gets dumped into the R channel.  Likewise a stop sign is also red, so it gets dumped into the R channel.

 

The R channel just has brightness values.  It doesn't know where they came from.  As far as the R channel is concerned, Ha, Sii and the stop sign are all the same.

 

Likewise for Hb, Oiii, and grass in the G channel.

 

"False colour" is only false because we override the default channel mapping that is done by the Bayer matrix in OSC cameras.  You can manually put the Ha data into the G channel, and the software won't complain.  This is routinely done with SHO (Hubble palette) mapping.

Thanks Kathy (and all) - this has helped me a lot. So this automatically happens in channel extraction with dual narrow-band filters. Say you had the two filters - Ha/OIII and SII/OIII: for the first, Ha will always go to red and OIII to green, and for the second SII will always go to red and OIII again to green. Nothing ever gets assigned to blue then. Then we just manipulate the colours by re-assigning channel mappings and pixelmath? (in some ways it makes working with mono seem less convoluted)

[kathyastro]

Nothing ever gets assigned to blue then.

I over-simplified for clarity.  In fact, the Bayer matrix filters will let some Oiii and probably Hb (though I have no experience with it) into the Blue channel.

[pgseye]

I over-simplified for clarity.  In fact, the Bayer matrix filters will let some Oiii and probably Hb (though I have no experience with it) into the Blue channel.

Fair enough - thank you for clarifying.