LRBG v. RGB again

[Rob Friefeld]

Reading arguments and discussions.
Is this true: If you have time to take RBG binned 1x1, there is no advantage to combining a separate L image, i.e. an advantage beyond using the extra imaging time for more RGB acquisition.

[Geoff]

See here http://www.astro.uni-bonn.de/~mischa/ps/syntheticlum.html
Geoff

[papaf]

I feel as though that page is lacking a case scenario which is very often used when you take a separate L channel: to use less exposure (less time) on RGB than you use one L. The advantage in this case is time saved. That page only considers taking the same amount of time for L and RGB channels, which is kind of useless in my book.

[jkmorse]

I agree with papaf on this one.  The primary benefit of shooting a separate Lum is that I can do 4 or 5 hours of 15 minute Lums and a dozen 5 minute 2x2 binned RGBs each in a single evening.  Compare that to shooting 1x1 binned RGBs and I am down to 3 hours for each, if lucky.  Seems like the LRGB route is the way to go 

Jim 

[Riccardo A. Ballerini]

+1 :-)

here is an extreme example of really brief binned color integration

only 3 120" per rgb channel (binned 2)

http://www.astrobin.com/50744/

[topboxman]

In my case, the only drawback for separate Lum filter is it's terrible under heavy light pollution. I use Astrodon LRGB filters and there is a "hole" in RGB filters to filter out light pollution. I also image with Astrodon Ha filter and blend Ha with Red and make HaRGB images instead of LRGB. I am never successfull imaging with Lum filter under heavy light pollution.

I guess it depends on the environment in your area whether Lum filter works for you or not.

Peter

[starhopper]

Hello Geoff,
do you know a written Pixinsight workflow to do a [LRGB]-RGB as stated by Mischa?

Vicent Peris talked very short at CEDIC that a artificial luminance can be made with ImageIntegration Tool:
Take a set of one R one G and one R in the dialog
Combination=Average
Normalization=Additive+scaling
Weights= Noise eval
No rejection.
...or so

But what is the right way to do it all together?

[RickS]

I do synthetic luminance by integrating the R, G and B subs as usual and then using PixelMath to create a new SynLum image by adding R+G+B.  Then I just include SynLum with my Lum frames when I integrate them.  You can get a sense for how well this works by looking at the weight assigned to the SynLum image by normalization or by toggling it on and off and looking at the integration stats.

I have tried more complex methods (such as building a set of synthetic lum frames from individual combinations of R, G and B frames) but I get the best results from the simple method above.

Cheers,
Rick.

[Rob Friefeld]

Thanks for the interesting responses, guys. The Mischa method surprises me--intuitively the [LRGB]-RGB method seems like getting something for nothing. Much to understand here. And Riccardo's extreme example is an inspiration.

[RickS]

It's not getting something for nothing.  It's making best use of those precious photons    With a traditional LRGB the colour exposures are only used to provide chrominance and the luminance data from those subs is discarded.

[pfile]

this is pretty interesting - i would assume that the SynLum image gets a very high weighting and the rest of the "real" L subs have small weights?

i wonder if II is really "okay" with large disparities between the images it is integrating. obviously it sounds like it works, though.

are you talking about a smallish number of frames in total? like the 1 SynLum + 10 regular L subs?

rob

[Geoff]

Hello Geoff,
do you know a written Pixinsight workflow to do a [LRGB]-RGB as stated by Mischa?

Not really, although these posts by Juan may help http://pixinsight.com/forum/index.php?topic=1636.msg9297#msg9297. 
http://pixinsight.com/forum/index.php?topic=1137.msg5592#msg5592
I've been experimenting with various methods for (a) making a synthetic luminance and (b) combining it with the RGB or LRGB, but have yet to settle on a fixed workflow.
Geoff

[Geoff]

I agree with papaf on this one.  The primary benefit of shooting a separate Lum is that I can do 4 or 5 hours of 15 minute Lums and a dozen 5 minute 2x2 binned RGBs each in a single evening.  Compare that to shooting 1x1 binned RGBs and I am down to 3 hours for each, if lucky.  Seems like the LRGB route is the way to go 

Jim

Well unfortunately there is always this tradeoff---save time, lose quality--spend (waste?) time, gain quality.  See these posts by Juan
http://pixinsight.com/forum/index.php?topic=1137.msg5592#msg5592
http://pixinsight.com/forum/index.php?topic=1636.msg9297#msg9297

[starhopper]

Hello Geoff,

I've been experimenting with various methods for (a) making a synthetic luminance and (b) combining it with the RGB or LRGB, but have yet to settle on a fixed workflow.
Geoff

I tried the following, but the difference between LRGB are marginal.

synL1 = ImageIntegration( red1 green1 blue1 )
synL2 = ImageIntegration( red2 green2 blue2 )
...
synLn = ImageIntegration( red(n) green(n) blue(n) )

finalLuminance = WinsorizedSigmaClipping ( L1, L2...Ln, synL1, synL2..., synLn)    // sigma =4, Normalization = scale zero offset

LRGB = LRGBCombinationTool ( finalLuminance/RGB)   // L=0,5, Sat=0,38-4,5

[RickS]

this is pretty interesting - i would assume that the SynLum image gets a very high weighting and the rest of the "real" L subs have small weights?

i wonder if II is really "okay" with large disparities between the images it is integrating. obviously it sounds like it works, though.

are you talking about a smallish number of frames in total? like the 1 SynLum + 10 regular L subs?

rob

Yes, the SynLum gets a high (though not huge) weighting compared to an ordinary lum sub and the integration including the SynLum shows a measurable improvement in SNR.  I have combined SynLum with smallish and medium numbers of subs successfully, typically between 10 and 40 of them.  The improvement is almost always worthwhile for the data I have tried.

I also tried doing it the way Starhopper describes but that didn't work as well.

If anybody is interested I can probably post some results from a recent processing run.

Cheers,
Rick.