New to LRGB- Balancing my luminance to colour exposure ratio?

[pipnina]

Hi

I've read a good starting point is to match time on L with time on RGB, which I presume means 3L + 1R +1G + 1B. I gather that the primary benefit of using luminance is that dark stuff can be brought into view, but in monochrome, while brighter stuff can still have nice colour data. Does this mean that if I am targeting a very deep image (IFN levels deep) I can capture a certain amount of LRGB with the 3:1:1:1 ratio, then stop collecting colour once I can get colour in the darkest parts of the bright objects I'm imaging? I also hear people talk about blue being a noisy band in astro and my initial experiments seem to support that, would it be worth intentionally capturing more B than G and R frames to bring out blues more easily?

Wondering what other people do here. Maybe even advanced techniques like blending RGB highlights into the L highlights to reclaim detail in bright parts of the image or something?

[pharscape]

I'm also interested in this answer. I have not been capturing any Lum and bright details with goods SNR look fine but the fainter stuff looks  a bit rougher than I've seen in other's images. I suspect I'm missing a trick.

[Clarkey]

I'm not sure if there is a hard and fast rule for this but I would say the 3:1:1:1 is a reasonable guide - but experiment. Colour is quite a personal thing, just look at the different processed images there are. No two are the same. Quite often you could also get away with binning colour data to get a better S/N ratio without extending the imaging time. 'All' of the image detail is in the luminance channel.

With regards to the blue channel, it certainly tends to be 'distorted' more but the atmosphere so I would always image this and the luminance as near to the zenith as possible.

[DaveS]

I try to get a healthy amount of RGB using G2v calibration so that the levels are at least approximately the same in the three channels, and in Bin 2. I keep the processing of this relatively "soft" without trying to use deconvolution or unsharp mask. For my ODK / 16200 system I found I needed 1.75x as much blue as red or green, but your system will have its own ratios. I will also attend to any colour balance adjustments and final gradient reductions here. The workflow (In AstroArt 8 will be (For each channel): Stack > Crop alignment edges > Gradient Reduction (Usually Adaptive Subtract) > DDP > Save FITS. During the DDP I will keep all the stacks open with their histograms which I will endeavour to match. Then import the three FITS files and combine in Trichromy. Again crop any alignment edges, and possibly another gradient reduction before sorting out aany wonky colour, usually it's just a matter of touching up slight imbalances. Finally save the RGB FITS.

For the luminance I try to get as much as I can without running into diminishing returns, or until I get fed up, usually at least 6 hours and preferably more, in Bin 1, just in case I have any detail below the 1.2"/px of my bin 2. This will get the full works of deconvolution, denoising, unsharp mask etc.

I will then upscale the RGB by bicubic interpolation before combining with the luminance into the LRGB. I will probably need to raise the saturation after the LRGB combination.

 

Hope this is of some help.

[tomato]

On the dual rig, with mono cameras and filters, I collect LRGB in the ratio 3:1:1:1, sometimes I’ll start a target with both scopes on Lum so this channel gets pushed higher.

It’s probably my inexpert manipulation of the respective software packages but with this ratio of channels I get lots of colour from StarTools, less with APP, but it’s like getting blood out of a stone with Pixinsight.

[pharscape]

19 hours ago, Clarkey said:

I'm not sure if there is a hard and fast rule for this but I would say the 3:1:1:1 is a reasonable guide - but experiment. Colour is quite a personal thing, just look at the different processed images there are. No two are the same. Quite often you could also get away with binning colour data to get a better S/N ratio without extending the imaging time. 'All' of the image detail is in the luminance channel.

With regards to the blue channel, it certainly tends to be 'distorted' more but the atmosphere so I would always image this and the luminance as near to the zenith as possible.

Thanks. The reason I've shied away from using Lum is my (perhaps unfounded) unease that for example a strong red structure picked up in Lum is used to emphasise R, G and B when it should only impact R.  You don't really have that issue/choice in NB as Ha and the others are often very different and L may not see enough data without stars blowing out.