Narrowband Luminance?

[BlueAstra]

Just starting down the narrowband path. With LRGB I take a Luminance at x1 and a quarter exposure RGB at x2. Occasionally I do LRGB at x1 (Do I need the L?). With narrowband is it normally same exposure Ha,O3,S2 at x1, or is there such a thing as L at x1 and quarter exposure Ha,O3,S2 at x2? I suppose it depends on the target (?).

[steppenwolf]

You don't need the 'L' if you're doing narrowband imaging. However, it might be fun to take a Luminance channel and combine it with an Ha/OIII/OIII or Ha/OIII/Synthesised Hydrogen Beta image where you are trying to get a naturally coloured image.

[ollypenrice]

One of the key attributes of NB imaging is the small stars it produces. If you use an L layer you will be back to the bloat!

However, you could always experiment with one of your channels used as L as well as being mapped to a colour channel. No idea what it would do. I often wonder why more people don't present a multi narrowbad image in monochrome as well. I think it has a certain something. This is the only time I tried it: http://ollypenrice.smugmug.com/Photography/Widefield-images-including/final-deep-narrowbandnohaloes/1012559664_wvgUf-X2.jpg

Olly

[roundycat]

just out of curiosity, why do you take one quarter the exposure at x2 binning?

Dennis

[BlueAstra]

I believe its because when you bin the pixels (2x2) you are creating 1 super pixel that is four times larger than the original single pixel. This super pixel therefore collects four times more light than the original pixel, so you only need a quarter of the exposure to get the same sensitivity as the original single pixel. Of course this increase in sensitivity is at the expense of resolution, so when you increase the size of the colour image to match the original luminance image (at 1x1) it is at much lower resolution. However, the eye uses the L image for resolution and the colour information doesn't need to be at the same resolution.

Having said all that, there is a debate about whether RGB should be at 1x1 (in which case do you need a L image?)

[ollypenrice]

I believe its because when you bin the pixels (2x2) you are creating 1 super pixel that is four times larger than the original single pixel. This super pixel therefore collects four times more light than the original pixel, so you only need a quarter of the exposure to get the same sensitivity as the original single pixel. Of course this increase in sensitivity is at the expense of resolution, so when you increase the size of the colour image to match the original luminance image (at 1x1) it is at much lower resolution. However, the eye uses the L image for resolution and the colour information doesn't need to be at the same resolution.

Having said all that, there is a debate about whether RGB should be at 1x1 (in which case do you need a L image?)

Binning is faster but never in a million years is it four times faster. I don't know the theory but I do know the practice!

Do you need an L layer with unbinned colour? Probably not if you shoot an eternity of colour and process a synthetic luminance layer differently from the RGB layer as you would in LRGB. But I find that in reality Luminance pours in fast and gives you a really strong layer which will take more stretching and more sharpening than an RGB. Right now we have 6 hours colour on a galaxy and tonight we aim to hit it with three hours' Luminance. I am certain that this L layer will greatly outperform the synthetic L layer that we have at the moment. If it doesn't I'll come back and 'fess up! BTW I try to keep the L layer off the stars because it tends to burn the colour.

Binning colour is OK if your sampling rate is reasonable but will be progressively less satisfactory as your FL comes down. I try to avoid it altogether. Stars are never as good.

Olly

[MartinB]

I usually find a luminence really helps pull out the detail when processing a narrow band image. The luminence can be created using Ha data or, if OIII and or SII are adding significant extra detail, you can combine all 3 into one luminence so that the detail is used from all 3 channels.

One big advantage of using a luminence channel is that it allows you to blur the colour channels. This can be extremely useful when doing a hubble palatte image and you are trying to tone down the magenta stars without leaving desaturated halos. It is amazing how the luminence brings back all the detail

[MartinB]

Binning will give you x4 the brightness for a given length of exposure but, unfortunately, at the expense of x4 shot noise! Binning is very powerful because it greatly reduces the read noise. A lower read noise gives 2 benefits - greater dynamic range and allows for shorter subs.

If you are binning colour and want good stars it is essential to keep exposures short otherwise the stars bloat. People often mistake this for poor resolution resulting from binning. Certainly at longer focal lengths binning has the capacity to improve stars because of the greater dynamic range but only if they aren't burnt out! Typically sub 1 minute exposures are needed.

[ollypenrice]

Martin, so a separate set of subs for a star layer when binning colour? This sounds like a promising idea.

Olly

[MartinB]

Possibly Olly. I've never done it. I've always found binned colour ok at long focal lengths but I do normally use pretty short subs (I used longer at Les Granges because, with the dark sky, it takes much longer to get over the read noise compared with my usual location). I'm quite interested in trying binning with my FSQ despite the general consensus being that this is too short f/l. I feel some experimentation coming on!

[BlueAstra]

Binning is faster but never in a million years is it four times faster. I don't know the theory but I do know the practice!

Olly

What would you recommend as the ratio for L(1x1) : RGB(2x2) exposure? I generally image at 1280mm F/6.4 (C8) or 380mm F/4.8 with an SXV-H16 mono CCD.

[steppenwolf]

The luminence can be created using Ha data or, if OIII and or SII are adding significant extra detail, you can combine all 3 into one luminence so that the detail is used from all 3 channels.

Perhaps I misunderstood the OP but I assumed that he meant a 'pure' luminance, i.e. no filter (or perhaps an IR to avoid splatter).

[MartinB]

Perhaps I misunderstood the OP but I assumed that he meant a 'pure' luminance, i.e. no filter (or perhaps an IR to avoid splatter).

It was certainly worth pointing out that that would be a horrible thing to do!

You can do an Ha luminence and bin the OIII and SII which might be ok at long focal length but there is a risk of resolution loss doing this.

Re binning ratios, imaging through filters removes a lot of light so you made need to make your colour exposures longer than 1/4 of the luminence. There are so many variables - local site characteristics, camera, optics, sky back ground and the target I don't think you can have a fixed rule. It's worth experimenting.

[roundycat]

Blue Astra, try measuring the brightness of an unsaturated star that you have photographed binned and un-binned. Give, say, a ten second exposure unbinned and then bin x2 and give another ten second exposure. Compare the two; the binned exposure should show four times the brightness for the measured star?

It won't, more like 1.5 times the brightness as it depends on the chip's read-out pixels, not the super pixel.

Dennis

[BlueAstra]

Thanks for that, I'll try it the next time I put the kit together. Sounds like a good way to characterize the system. I'll do it for each filter.