How little the RGB channels contribute. Terrestrial example

[Stonius (Markus)]

I've been playing with my ZWO-canon adapter and took this inspiring test shot of my bookshelf last night (primarily because it had information in the RG &B channels and it was cloudy).

I was playing with the colour information independent of the luminance channel to see what a difference it made to the final result, given that many people have spoken about how you can get away with shooting RGB at higher gain and just using noise reduction and whatnot since most of the detail the eye sees is in the luminance channel.

Boy, were they ever right. It would be good to repeat these tests with, you know, actual stars, but I thought it was interesting to see it on familiar objects.

There are four images. The top two are both LRGB. Underneath them is the RGB data the colour in the top image was derived from. On the bottom right, the RGB channels have been gaussian blurred by 50 pixels and there is still only a subtle difference between the LRGB derived from that (top Right) and the LRGB with the full res RGB channels (top left). But my favourite in terms of richness of tone would have to be the full res RGB (Bottom Left). The reds seem to lose out when the luminance is added to the mix, but this is quite possibly because I am very new to post processing and doing something horribly wrong :-).

Hope this is as interesting to other people as it was to me :-)

ASI 1600 MM, ZWO Filters, Canon 70-200mm 2.8 zoom. Photoshop.

Cheers

Markus

[kosborn (Kevin)]

So, you like to travel...


Sorry, that was facetious. That is actually really good information and good advice for DSO processing techniques.


Kevin

[Atmos (Colin)]

When burring RGB for stars, you're usually not blurring them that much

This does go to show how much damage you can do to the RGB when you have good luminance. It isn't quite that simple though, although with a sharp luminance you do keep the detail what you do lose is accurate colour contrast. This colour contrast is noticed in a few important areas, most noticeably, around stars. If the RGB is too blurry you end up with coloured rings around stars and that does make for a pretty ugly look. This is something I've had to deal with when trying to merge MUCH sharper Ha with RGB.

[Stonius (Markus)]

Originally Posted by Atmos

When burring RGB for stars, you're usually not blurring them that much

This does go to show how much damage you can do to the RGB when you have good luminance. It isn't quite that simple though, although with a sharp luminance you do keep the detail what you do lose is accurate colour contrast. This colour contrast is noticed in a few important areas, most noticeably, around stars. If the RGB is too blurry you end up with coloured rings around stars and that does make for a pretty ugly look. This is something I've had to deal with when trying to merge MUCH sharper Ha with RGB.

Yes, I think you're probably right. It will be interesting to try, and see just how brutal I can be with the colour channels for a real shot when I get to that stage.
:-)

[Lognic04 (Logan)]

VERY interesting! No difference in the blurred RGB apart from lower saturation, which can easily be fixed!

[Stonius (Markus)]

As Colin noted above, on fine details, you'll notice the colour has been blurred out of existence. Take, for example the rainbow strip on the spine of the 'Lonely Planet guide to the World'. In the Blurry RGB version the colour has been completely obliterated!

[Lognic04 (Logan)]

I do think your blurring was rather extreme though! Just goes to show though that you can still get good colour however!