Processing using pseudo Luminance from RGB a good idea?

[topboxman]

I have noticed some people capture more RGB images to avoid capturing Luminance images. Then create pseudo Luminance from RGB and process pesudo Luminance before applying pseudo Luminance back to RGB.

Ever since I bought Atik 460EX mono camera, I captured with all 7 filters (LRGB, Ha, Oiii, Sii) and noticed Luminance images had bloated stars. Several people from Cloudy Nights forum have the same issue so they ended up capturing more subs of RGB and not take Luminance anymore and get pretty good results.

My camera has tiny 4.54uM pixels and I image with C-8 EdgeHD at 2000mm focal length. This makes the image scale at 0.47"/pixel. The images look great with six filters except Luminance channel.

So, is processing with pseudo Luminance a good idea? Also, how does processing with pseudo Luminance work?

Thanks,
Peter

[Josh Lake]

I'm not sure if it's a great idea, but I know how to make it work. I use the technique even when I have a real Luminance in order to boost saturation and get some Chrominance Noise Reduction.

If I was going to use a pseudo-L, I'd make my RGB, extract the Luminance channel, then use the LRGBCombination process. I'd only check the L box, point it to the pseudo-L, and then apply the process to the RGB image.

The Saturation slider has to be moved to the left to increase the sat.

I usually play with the saturation settings until I am happy, then do one more run with the Chrominance noise box checked on the defaults.

This technique (from Vicent Peris' tutorial on YouTube, I believe) gave me great success when working with Hubble Data that had just two channels.

[oldwexi]

Hi Peter!
Extracting the L Channel out of RGB gives better results if you set the RGBWS of the RGB before to 1 1 1.(not to sRGB!)
Otherwise the Extracted Luminance Channel is built by the sRGB percentages which have mainly the data from the Green in it and only a fraction
part of the blue and red.

Aloha
Gerald

[topboxman]

Hi Gerald,

Are you saying using Channel Combination of RGB with equally weight of 1:1:1 is best to extract Luminance from RGB? Is that what RGBWS means? I am not sure what sRGB means but does it mean unequal weight of RGB combination?

Forgive me with these questions since I migrated from using OSC camera for three years to mono imaging with filters. So mono imaging is new to me.

When I processed my first RGB image, I simply use LRGB combination with "L" unchecked and not worry about the weights for each RGB channel. Is that the same as RGBWS?


Thanks jlake for the detailed message.

Thanks,
Peter

[topboxman]

Someone from Cloudy Nights forum suggested to capture as much R, G and B images, calibrate them and stack R, G and B separately. Then "sum" the stacked R, G and B. So in other words use Pixel Math to do R + G + B to create a pseudo Luminance. Process  pseudo-L (work on star size) and RGB (push on saturation) separately and then combine processed pseudo-L and RGB together and do some final tweaking.

Would this method work as well?

Thanks,
Peter

[pfile]

that straight sum is what gerald is suggesting... at the root of color space theory there's a lot of psychovisual stuff. i think the weights in the sRGB color space have to do with what wavelengths the human eye are most sensitive to (peaks at green, which is why there are 2 G pixels in a bayer matrix - cleaner G data = higher SNR luminance information in a terrestrial photograph).

i have done at least one pseudo-LRGB project with my 50D, but now i can't remember which one! for the longest time i had forgotten about it, but i have been working on an M33 using the pseudo-LRGB method. i've stopped though since i need to collect more photons first...

[topboxman]

Thanks pfile.

Peter

[oldwexi]

Hi Peter!
Extracting Luminance from RGB image in PI depends very much on the
selected RGBWS Parameter of the RGB image.
Using sRGB parameters in RGBWS and extracting Luminance takes
0.060621 from Blue
0.222491 from Red and
0,716888 from Green.
Means your ReflectionNebulae and Emissionnebulae are very low represented in the Luminance -
and the green nebulae will participate with 71%....

Using instead of sRGB Parameter in RGBWS for Red/Green/Blue/     1/ 1/ 1/
Extracts the Luminance correct.

Aloha
Gerald

[Warhen]

One caveat would be to register them all only once to a single reference image before integrating separately.

Someone from Cloudy Nights forum suggested to capture as much R, G and B images, calibrate them and stack R, G and B separately...
Thanks,
Peter

That being said Peter, the possible benefits of Pseudo-L processing of OSC/CFA images have been debated elsewhere ad nauseum. I've covered it in many tutorials and Jerry Lodriguss at Astropix.com has done a bit of comparison testing which you might want to look at. While you certainly can't get 'something for nothing' (IOW the P-L is just an extraction of the same data achieved by the RGB), it can feel easier to work the RGB for its color contribution, and reserve sharpening and contrast work for the P-L where you are only worrying about one grayscale channel rather than 3.

[topboxman]

Hi Warren,

Yes I am aware about registering before stacking separately.

You are correct about pseudo-L containing same data as RGB. I do not have good Luminance data due to either bad seeing or guiding. Also bad weather has prevented me to give another try at capturing Luminance. Also I live in high light pollution area and capturing Luminance is difficult. I was not really trying to make my life a little easier, I was hoping to make RGB image to look a little better using pseudo-L.

After about three years of imaging with OSC camera, I just migrated to mono camera with filters so I am new to this. I have successfully captured with Ha, Oiii, Sii, R, G, and B. This is my first process of Bubble Nebula with my new Atik 460EX mono camera:

http://peternagy.smugmug.com/Telescopes/Nebulae

Peter

[topboxman]

I thought it was more appropriate to revive this old thread instead of creating a new thread.

So after about the next two years since this thread was last posted, I decided to try pseudo-Luminance + RGB again and compare to real Luminance + RGB. My most recent LRGB capture was NGC6946 or Fireworks galaxy a few weeks ago from heavy light pollution area in Reno, NV. All filters were unbinned 1x1. After finishing processing LRGB, I decided to try extract Luminance from RGB (1:1:1 ratio) and process with RGB the same way and was able to get pretty good results.

Take a look at my images at http://peternagy.smugmug.com/Telescopes/Galaxies . There are two sets of three images. First set is whole FOV and second set is cropped to look closer to the galaxy. Each set shows LRGB, pseudo-L + RGB and RGB respectively. Blink in between images and see if you notice a big, small or no difference.

This reminds me of of an old TV commercial back in the 1970s "Is it live or is it Memorex?" (https://www.google.com/search?q=is+it+live+or+is+it+memorex&tbm=isch&tbo=u&source=univ&sa=X&ei=DFgCVK74E864ogTHgIJY&sqi=2&ved=0CB0QsAQ&biw=1389&bih=901). So this thread should be re-titled "Is it Luminance or is it  pseudo-L?"  8)

I think I will try imaging only unbinned 1x1 RGB and extract RGB to create pseudo-Luminance from now on. This way I can image same DSO for one night instead of multiple nights. About 100 to 120 minutes per filter totaling 5 to 6 hours may be good enough.

Peter

[jkmorse]

Peter,

On top of the thread you quote, Juan offered his thoughts on this and got me to change to only shooting 1x1 RGBs and using a Synthetic Luminance a while back.  Here are two threads worth reading on the subject:

http://pixinsight.com/forum/index.php?topic=6042.msg41037#msg41037

The second thread referenced in the first post is particularly important as it really lays out Juan's explanation.  For convenience here it is again: http://pixinsight.com/forum/index.php?topic=1636.msg9297#msg9297

Best,

Jim

[topboxman]

Hi Jim,

Thanks for the useful links. They are interesting.

Peter

[topboxman]

It's possible that using Luminance filter may be more beneficial from a dark site. My Astrodon E-series RGB filters have light pollution portion blocked in between green and red spectrum and may help create pseudo-Luminance nicely under light pollution area.

http://www.astrodon.com/custom/_2e2a/content/images/AstrodonG2ESeriesScansforWeb.jpg .