Can't seem to get the colours right

[admin]

The cool thing with 1.5 is that you don't have to use the SII data for the detail - you can just use it for the chrominance portion and leave it out of the luminance completely - you'll still get a HST palette image. E.g. just set Red Total Exposure to "Not set" while choosing "L + Synthetic L from RGB, RGB" for "Luminance, Color". Or you could go with Ha only if the S-II and O-III isn't up to snuff (which is what I would probably recommend, as the noise floor is just too high otherwise).

You'd get something like this (for example);

NewComposite.jpg (243.39 KiB) Viewed 5172 times

E.g. even with just using the signal from Ha as luminance, your colors aren't impacted.

Be extremely careful when processing channels separately and combining them. You may not be just creating false coloring (which is obviously ok for narrowband) but also introducing false detail.
See here for an explanation.
https://www.cloudynights.com/topic/6722 ... ?p=9588967

Unfortunately, I can indeed spot "detail" in the bi-color rendering that just doesn't exist (for example, the two purple "blobs" next to the bubble).
I cannot stress enough that combining two (or more) individually processed stacks though color channel combines, is virtually never a good idea, especially if your goal is to engage in documentary photography (and not art).

[Rkonrad]

Hi Simon and all,

I've completed 2 videios for you to look at - it's rather long - sorry!

The first one is the general processing and the second works more with stars. I hope they're helpful! I'll shortly be copying a similar message to the tutorial section of the forum. Is it ok if I make the videos public - as its your data?.
In the end I chose not to process them under "compose" LRGB as the Oiii channel was so weak. I needed to boost the Oiii a lot to get it to show up.So I processed the two channels separately and then combined them.

Cheers Richard

https://youtu.be/x4vrk-O83oQ

https://youtu.be/RIv7EQPxQss

final image https://www.flickr.com/gp/rkonrad/wN9816

[szymon]

Very interesting! Thank you for demonstrating. Your advanced use of masking is a good tutorial, and watching you clean up the dirty data in the first video and pull what you can out of it was fun. I noticed in the first video when loading the individual files to process separately you told StarTools that the data was Bayered (second option) -- why was that? The second video was a revelation to me, I've never thought of removing the stars from an image and seeing the nebulas 'naked' was an interesting experience! Also it's a great example of the layer module, which I've never used before. I think it would be interesting to take bicolour data for nebulas and grab real RGB data for stars, to get the star colours correct. Certainly more "possible" for me to do that meaningfully from my light polluted site than it would be to grab RGB nebulosity. Something for me for the future I think!

Of course you're most welcome to make the videos public. I collected the data, sure, and I like the "bubble'n'brain" composition, but it was pointing at the sky which doesn't belong to me so feel free to display it -- especially in such a useful tutorial .

-simon

[Rkonrad]

You're welcome. I'm glad you found the longish tutorial helpful. I reviewed the final image again and if I could do it over again, I would back off on the dark saturation as there is a constant "red" even in the dark background. So just to stress, as is suggested by Ivo, I would be processing the channels LRGB together at the beginning (as linear data) if the Oiii data were stronger.

I chose the second option , " bayered" somewhat randomly. I dont think it matters in the case of narrowband images without the colour bayer matrix. Ivo correct me on this if you can chime in.

Cheers and all the best. Richard

[admin]

Thanks Richard!

I dont think it matters in the case of narrowband images without the colour bayer matrix. Ivo correct me on this if you can chime in.

This does make a difference, as the luminance data gets to be reweighted as 0.25xR, 0.5G and 0.25xB. E.g. it is assumed that the green channel signal is twice as good as the other channels.

Cheers,