Do I need Luminance?

[BlueAstra]

Do I need Luminance?

Another thread mentioned using RGB only rather than combined with L. A suggested method for LRGB images is to shoot L with 1x1 binning and RGB with 2x2 binning. Usually the RGB exposure time is less than the L exposure because of the effective increase in pixel size. If the RGB is half the L exposure, a total exposure time for LRGB set would be 1+0.5+0.5+0.5=2.5 . If I were to ditch the L exposure, I would have to use 1x1 binning with the RGB, so my total RGB set time would be 1+1+1=3 . So the RGB set at 1x1 would take longer than the traditional mixed LRGB. However, half exposure time for RGB is a little optimistic, so an increase to perhaps 1+0.66+0.66+0.66 may seem reasonable. This would make the exposure time for mixed binning LRGB set the same as an RGB set at 1x1.

So my question is, if I shoot RGB at 1x1 do I need Luminance? Or have I not considered something?

[Whippy]

I would imagine it depends on how you work I guess, I always used to plan that it'll take one or two sessions to do the luminence first, then do the RGB on another evening or two depending on how the evening pans out. That way I'd have some data to play even if it's just mono with and try and work out how much more data I'd need. Quite often even binning colour, I'd need more than one evening so using that method would have been quite frustrating for me. Another thing to take into account is atmospheric scattering, which in real terms is how much more Blue data you'd need.

FWIW, I used to bin my colour but I didn't in my more recent images, I just did an hour of each colour, that seemed to work out alright. TBH I didn't notice a massive jump in quality from doing binned to not binning, but then colour processing isn't one of my strong points .

Tony..

[RobH]

What you've missed is the fact that you will be getting less light onto your RGB filters than through a clear luminance filter, so to get the same amount of light onto the chip, you'd need to increase your exposure times.

That said, I usually shoot my RGB unbinned, and use 10 minute subs. When I come to create an RGB image, without luminance it's normally looking good, but adding deep luminance really brings it out.

It very much depends on the target too.....if I decide to have a crack at M42 this winter, I'll shoot just RGB, as it's a bright target, but with faint galaxies, the extra depth of a big stack of luminance allows the faint arms to be brought out without running into colour noise.

I almost never use luminance layers for stars though....RGB is much better for maintaining star colour IMO.

HTH.

Rob

[roundycat]

You infer that x2 binning will give twice the sensitivity although admitting this is maybe a bit optimistic. I'd say very optimistic. My tests have made it about 1.4 times as sensitive.

Take a picture of a star for, say, five seconds at x2 binning and then repeat without binning. Measure the brightness of the unsaturated star for both cases. One divided by the other will give you your sensitivity increase.

The gain increase depends on the configuration of the read out pixel, not the binning.

If you look closely at binned info and compare it with un-binned you will see a definite decrease in resolution with binning.

Dennis

[BlueAstra]

With L=1 that would mean the time for an LRGB set would be 3.14, and for RGB it would be 3, so in this (very) simplistic assessment RGB at 1x1 would actually be quicker than L+RGB(2x2).

In my non-permanent setup I generally go LRGB, LRGB,... so I don't revisit for RGB.

Regarding the brightness of the L over the RGB images the spectra of the RGB filters, when added together, cover more or less the same area as the L filter. So wouldn't the combined RGB image be more or less the same brightness and resolution as the L image (assuming L=1, R=1, G=1, B=1 exposures, all 1x1)?

[Uranium235]

My tests have made it about 1.4 times as sensitive.

Thats exactly what I wanted to know, thanks Dennis. What camera were you using? It would be nice to know the approximate value for the Atik cameras.

[ollypenrice]

I've read lots of theory and taken lots of pictures. The pictures don't always agree with the theory so I pay more heed to what they say since they are what I post on SGL! A case in point would be the one made by Dennis. You do get a sensitivity gain from binning but not the 4X that some would have you believe. I'm happy to accept Dennis's value.

I have picked up from Rob the idea of avoiding L for the stars and agree that it works.

I also agree with his other remarks. Some objects just need the tight, sharp, contrasty detail that seems etched into a deep L layer. Some don't. I shot no L for the dark LDN objects I was working on recently and may shoot none on the present Iris project because in order to have enough data for the faint dusty stuff (hours and hours and hours!) I feel a synthetic L layer will probably give me all the hard edged detail I need for the bright part. We'll see.

So I just don't think every image needs the same formula. I don't like binned colour (not such good stars) so try to avoid it. But, of course, I'm not short of time.

Olly

[roundycat]

Why are so many people shy about their name?

Blue Astra (I once had a white one). It is easy to work out the relative exposure times for binned versus unbinned and with L thrown in if you wish. Don't forget to factor in the new set of darks for an L master and the flats for the L filter.

Uranium, (another nom de plume, Olly). My figures are much the same for the ST10XME and the STL11000. It only takes a few minutes to do the entire test. I'll be interested to see what you get for your Atik. It will almost certainly vary by model.

Dennis

[BlueAstra]

Why are so many people shy about their name?

The internet is a dangerous place and I'm sure if I used my real name you would soon have all my bank details, and have complete power over me . I had a blue and a red Astra, but now have a black Honda, so perhaps I should update my user name!

I'll try the LRGB exposure comparison on a bright star next time out. With the little time I have, what with weather and setup, I'm always reluctant to try 'experimental' measurements rather than image gathering. But its perhaps a false economy since a little investment in knowing the capabilities of your equipment should lead to improvement overall.

Graham

[roundycat]

Noted! Furthermore I am not after your money!

I hate 'wasting' potential imaging time on experiments but some of them simply must be done. A lot of people imagine (and are actually 'told' by the ignorant) that x2 binning will give them four pixels for the price of one, hence four times the gain. When their short exposures do not deliver the goods they look elsewhere for the problem.

There is no substitute for knowing about binning with your camera as well as its linearity and chip sensitivity variations. (why we take flats).

Dennis

[Uranium235]

Next time im out (hopefully Saturday) I will test the 314L+ and see what figures I get. Maybe the 285 chip will fare slightly better?

Rob

[adamsp123]

Interesting topic, I upgraded my mount recently and as I could now can run much longer subs than previously I ditched the binning for some trials and the results so far look much better with unbinned colour than binned.

[Uranium235]

Ok, last night there was clear sky for a couple of hours so I ran the test on the 314L+.

I measured 5 non-saturated stars in a 5 second sub (L filter) at 1x1, then same again at 2x2. The average sensitivity gain for my camera at 2x2 bin was x1.85, though this would need to be varified by other 314L+ owners.

[ollypenrice]

Ok, last night there was clear sky for a couple of hours so I ran the test on the 314L+.

I measured 5 non-saturated stars in a 5 second sub (L filter) at 1x1, then same again at 2x2. The average sensitivity gain for my camera at 2x2 bin was x1.85, though this would need to be varified by other 314L+ owners.

OK, so certainly nothing like the x4 that gets bandied about. I never believed that, just on a subjective look at what I was getting.

BTW, with 12 hours RGB/OSC I didn't shoot any L for the Iris in the end but since I'm a curious so and so I may just give it a try all the same, or see if I can cadge some from a friend who has just done a long run on it.

Olly

[opticalpath]

Interesting thread this! I hadn't really thought about it before, but I think there may be a misconception here about the effect of binning. The principal benefit of binning is to increase the signal to noise ratio in the image. A 2x2 binned image will not make stars 4 times brighter, but it will make faint image data more visible due to improved signal to noise (SN) ratio.

The improved SN ratio arises because the binning is done 'on-chip', before the sensor pixels are read and digitised. A 2x2 binned sensor will have 4 times fewer read-events than an unbinned sensor, so you get 4 times less read noise and therefore better SN ratio in your image.

If you measure the flux of a star in an image using an aperture photometry tool such as in Maxim or AIP4WIN, you will get near identical measurements in a binned and unbinned image of the same exposure. That's because the circular aperture used to measure the star surrounds the whole PSF of the star, and in both cases measures the total flux that arrived at the sensor from the star (minus the sky background). It's the same total flux coming from the star whether you bin or not, so the measured total brightness should be the same. When doing photometry, you would not expect to get a different answer if you bin the sensor ..... as long as you don't saturate any pixels in both cases.

Everything is further complicated because stars' light is not uniform on the sensor: it's spread unevenly over several pixels. When you bin 2x2, say, you collect exactly the same number of photons as you would have unbinned, but they are just counted and *displayed* over a smaller number of pixels, so stars look smaller and sharper ...... but not 4x brighter.

Imaging faint stuff, SN ratio is king and that's what binning gives you, at the cost of resolution.

Adrian