What does the Luminance channel contain with OSC cameras?

[chris_todd]

So I recently obtained a modified DSLR, and was thinking about using narrow band filters with it (so I can shoot from my suburban home).

If I shoot with, say, an Ha filter, then all the signal will be in the red channel, so I should be able to throw out the G and B channels.  Except, I don't know how to do that, so I was wondering if I just extracted the Luminance, would that effectively encode all the signal I collected in the red channel?

Or more to the point, what, exactly, is encoded by the Luminance channel in a shot from a OSC color camera like a DSLR?  Is it simply the intensity of the signal at each pixel, regardless of the color filter in front of that pixel?

[pfile]

well, i'm not sure there is a luminance channel per-se. in any OSC the color filter array is just that - R, G, B filters. they are pretty wideband so i suppose given color theory you can 'reconstruct' the luminance, but in your specific case (Ha filter) this is a bad idea.

it's easy to discard the G and B channels in pixinsight. first, before alignment, choose a debayering method which does not mix channel data (so bilinear or superpixel is good). then integrate your subs, then do Image > Extract > Red Channel, and there's a greyscale image containing your red-channel data.

it's a bad idea to create the luminance from an Ha-filtered image because there is little to no signal in the G and B channels, which also participate in the construction of the luminance data. by necessity since the signal is low, the SNR is terrible and your Ha data will be completely corrupted by the G and B channels.

Ha is possible with an OSC of course but you are leaving a lot of signal on the table. you end up stimulating only 1 of the 4 pixels in each bayer matrix. OIII is a little more forgiving with an OSC, or at least a canon DSLR, since the OIII wavelength seems to stimulate the two G pixels and the B pixel equally. This of course depends on your filter bandwidth but it does seem to be the case with my 12nm astronomik OIII filter. i have experimented both with extracting the 2 G channels and the B channel from my OIII images and stacking them together; the result seems pretty much the same as simply debayering normally and integrating the G and B channels together. the SNR in the debayered G channel is much better than the B channel, but PI's imageintegration process can weight the subs by SNR so this is okay.

[chris_todd]

Thanks for the clarification, pfile.  Not sure how I missed Image->Extract Red Channel.  Doh!  Your explanation makes a lot of sense, though I obviously have some learning to do on de-bayering, but learning something new is half the fun with astrophotography!