The human eye sees detail better than it sees colour, so you can take colour shots at a lower resolution and in less time and combine them with a high resolution Ha or luminosity shot and still get a nice looking result.
Geoff
furthering to what has already been explained (which is the main reason in anycase)
When imaging in Narrow band your capturing a very very small area of light and hence it separates it from the other colours and gives you a very distinctive line. Here is the difference between OSC broad band and OSC narrow band of the same image with the same equipment/time/location
Like i said the details in the full colour (broad band) image are lacking but in the blended and cropped version they are very prominent Do also remember this is a OSC DSLR all images are 10 min ISO 800 average of about 16 images with flats and darks subtracted.
hope that gives you a clearer idea of how it all goes
furthering to what has already been explained (which is the main reason in anycase)
When imaging in Narrow band your capturing a very very small area of light and hence it separates it from the other colours and gives you a very distinctive line. Here is the difference between OSC broad band and OSC narrow band of the same image with the same equipment/time/location
Like i said the details in the full colour (broad band) image are lacking but in the blended and cropped version they are very prominent Do also remember this is a OSC DSLR all images are 10 min ISO 800 average of about 16 images with flats and darks subtracted.
hope that gives you a clearer idea of how it all goes
I couldn't easily see the differences and decided to have a little play with your data. I just aligned the Ha data and the RGB and used the Ha data as luminousity in PS and here is the result. I hope you don't mind.
I don't mind at all when somebody brings in a new line in.
My one is worked from a blended HaR GB so that i can keep natural broad band colours for the stars. Its a tricky standoff when introducing the narrow band flavours.
This is a topical question that gets raised every now and then and there are usually lots of differing opinions about it.
I have done both and have formed my own guidelines about it.
Basically the theory is changes of hue in an image are usually gradual whereas contrast of luminance/detail are sharp. Hence 1x1 for Luminance and 2x2 for RGB.
There is however, a technique where you can use the desaturated RGB 1x1 and add it to the luminance or as an additional luminance layer. This can accentuate and liven an image and add extra detail, more signal, take more sharpening etc.
When do you do it and when will you get a gain and when won't you?
Well, firstly the seeing has to be good. Otherwise you are wasting your time and you are better off with 2x2 for RGB. Secondly I would add if your imaging system is fast you probably will get a significant gain by doing RGB 1x1 and doing the above.
The above is modified by your camera's pixel size and the resolution of your system. Faster systems have a larger number of arc secs per pixel so when you bin 2x2 the star sizes can enlarge. 1x1 can show stars significantly smaller. Also if you are undersampled ie. your arc seconds per pixel is greater than the ideal .66 given 2-3 arc second seeing as typical and using the Nyquist theorem of 3X for an adequate sample (ie. 3 x .66 = 3). You aren't getting enough resolution in the first place and binning will worsen that.
So if slow F ratio or small pixel size or poor seeing then do 2x2 binning. If fast F ratio or larger pixel size and good seeing then RGB 1x1.
Also for galaxies or globular clusters where you want maximum resolution then you would ideally use 1x1 binning for the LRGB or HaRGB. I normally use 1x1 binning for Ha but occassionally use 2x2 when it is a dim object or the seeing isn't great or weather is a bit cloudy and I am trying to get max signal in minimum time.
I first noticed the difference as being very large when using an FSQ106ED with F3.7 reducer. The difference in star sizes between 2x2 and 1x1 for RGB was huge. I have since used 1x1 binning for RGB with my TEC180 F7 refractor and noticed a significant increase in resolution in the images.
However if I were using my TEC180 at F12 (which I have done) I would use 2x2 for RGB otherwise the image signal is too dim and it takes too long.
i call it synthetic lum layer i sometimes pick out a specific channel or take the image as a whole to do the same thing too many process's in my toolbox of joy i thinks i forget them every now and then.
One of the reasons for Binning is to improve the SNR (Signal to Noise Ratio) for the same exposure time
Another is that you can significantly reduce your RGB exposure times by a factor of up to 4 for 2x2 and up to 9 for 3x3 binning (depends on chips).
The eye detects the luminance detail much better than the chrominance detail.
Combined you can get more image for less time or a deeper image for the same time.
What you lose in resolution while binning say RGB is gained by utilising the fine 1x1 luminance detail (unbinned).
Ha could be put into two different cases because you can use Ha as a luminance channel as well as a supplemental Red channel, so you may want the extra detail in the Ha and suffer a longer exposure time due to it being narrowband, or on the other hand due to the normally long exposures required, you might want to bin Ha to save imaging time and use L for luminance or start playing with combined Luminance/Ha layers - a whole nother black art !
It really depends on what you are trying to do and how much you want to fiddle to extract the type of detail you are looking for - in the time frame you have.
The CCD's effective Well Depth (a single pixel's full capacity as a measure of the amount of electrons it can hold) can be increased with 2x2 binning, but since most of the noise is read noise and this remains the same as before - so in a simple case of twice the well capacity you are doubling the SNR.
It is not always a simple as this because some cameras have antiblooming gates and so each pixel is still being limited to its maximum and the binning mode the CCD supports may not be linear or allow the full combined charge.
So it can be camera dependent.
Greg's example of the FSQ is possibly a slightly different problem, since his image scale (assuming 9um pixels) was possibly around 10 arc secs per pixel - meaning the average small star which should be less than one pixel (3-5 arc secs) when binned 1x1 without the reducer is now at 2x2 one large square pixel with no detail.
So any fine detail in RGB is limited to blocks of 10 arc secs.
and in Luminance to 5 arc secs.
So the focal length of your telescope (and camera pixel size) can also introduce some constraints on what works best in your particular situation.
It is generally better to be oversampled for average seeing conditions to start with. (agan depends on what you are wanting to capture - the big vista or fine detail)
Generally you would bin to achieve an image scale of around 2-3 arc secs per pixel (for average seeing conditions).
Pick a good night for your luminance imaging ! and you might not be able to tell the difference.
Binning is also useful for focussing because it halves the exposure time and in some cases the download time as well - which can mean for some cameras a saving of 10-15 seconds per focus exposure !
Just saves a lot of otherwise wasted time.
Just dont forget to switch the camera back to unbinned when you have finished.
Note - that some bin modes are done on chip and others are done in software externally so the download times for some bin modes may not be as fast downloading as they should be.
Thanks all - some great info in the this thread. There's obviously a lot of variables re scope, camera, chip, seeing etc. Just when I thought I had my head around all this, some of this new info makes we want to have a rethink - might go back and do a few more experiments with my rig to check my assumptions.
I chatted to a friend today regarding the SNR gains that can be had and some of the issues of full well depth
It gets much more complicated than I thought !
Its very chip specific.
Apparently somebody did some measurements and concluded that the new 16mp Kodak chip actually doubles its noise when binned - so you get the gain of shorter exposure times or a deeper exposure for the same time but at the expense of noise.
Most other chips peak out at maximum binned 'revised' full well depth of around double, some dont even get to double.
I think I understand that some antiblooming chips are limited by the bleed circuitry ?
The technology and circuitry inside the CCD are what counts and not all of this info is readily available in an easy form for us to just look it up.
So more heresay than science so far from me Brendan !
If you look up SBIGs data tables on their website - some of them list the full well depth binned and unbinned - so thats a useful start since that is specific to the chip as opposed to the camera.
I'd certainly be interested in any QHY9 experiences people have had at around the 1 arcsec/pixel range.
Most of my (limited) RGB work has been done at great dark sky locations, so lots more experience with city conditions required. Always good to have plenty of projects on the go....
Do also remember this is a OSC DSLR all images are 10 min ISO 800 average of about 16 images with flats and darks subtracted.
i sometimes pick out a specific channel or take the image as a whole to do the same thing 
