Quick question. Is an extracted Lum channel from an RGB, a Synthetic Lum layer?

[Tom OD]

Hi There,
I'm playing about with the noise script, and I'm not really sure what a Synthetic layer of an RGB really is. I have the "Extract CIE L* Component" icon on my tool bar, and I use this. Does this constitute a Synth Lum channel.
Thanks Tom

[dayers]

Hi Tom,

I've heard of the CIE L* image referred to as  "pseudo luminance."  I use it for L when I want to create an LRGB image from  RGB.  I can process the L separately from RGB before stretching both images and recombining with LRGBCombination. It also allows for increasing the saturation in the LRGB.

Dave

[pfile]

it's not the same as true luminance, but close i suppose. before you extract L* you should set the RGB weights of the original image to 1:1:1, otherwise the L* is heavy on the green component.

[Tom OD]

Thanks for the replies,
I must see where I set the weights. I'm sure its default, so I hope its already at 1:1:1. I'm asking as a friend suggested that mixing a synth Lum with the natural Lum gives a better noise value.
I extracted the Lum from the RGB as a synth Lum. Its noiser than the natural Lum which makes sense. However when I combined a synth RGB lum with the natural Lum, according to the Noise evalutation script I got better or lower noise values with the Synth 30% +Lum natural 70%, than the 100% Lum natural signal.
After combining, the maths says the mix of synth and natural Lum is better! I was not expecting that I must say.
I'm attaching a screen shot.
Tom.

[pfile]

sorry, i was away from my pixinsight machine when i posted that so i could not check. you are looking for the RGBWorkingSpace process. now i can't remember if you are also supposed to set the gamma to 1 as well - hopefully Juan can remind us.

rob

edit: it looks like by default, RGB images are in sRGB space but with D50 illuminant. here is some old documentation:

http://pixinsight.com/doc/legacy/LE/14_color_spaces/defining_rgbws/defining_rgbws.html

so you'll have to set those coefficients all to 1 before extracting L*