Second dumb question-about LRBG blending

[Niall Saunders]

Now I am confused.

Everything I have ever understood about deconvolution (which could be summarised behind this little dot . ) required that the data had only undergone 'linear' transformations, otherwise it would be almost impossible to construct the necessary 'deconvolution' matrix needed to (try to) recover the original data.

OK, so a star in the final image is still 'a star', and deconvolution would be attempting to reduce all the 'noise effects' that will have 'spread out' the original (theoretical) 'pin-point' source. But, unless you only deconvolve STARS in your final image, then the deconvolution matrix will be applied to the WHOLE image, including areas that have perhaps undergone significant 'extra convolution' (by way of HDRW, etc, etc).

Surely this is then guaranteed to introduce a whole load of unpleasant processing artifacts?

Now I am confused!

[Jack Harvey]

Yes I agree it is supposed to work as you said.  But I tried it once and found 2-3 iterations of selective deconvolution using a mask often improved the image???  So I often use it at the end of the processing session.  Not always, and you've to check carefully for artifact, and often mask areas that get too affected.  I am just reporting my results, sorry cannot help with the theory of everything on deconvolution<G>.

[Harry page]

Hi Dave

I took your Lum M51 image and it is slightly deeper than the RGB with some of the faint stuff just coming through and a bit more signal in the core so I could

extract a little bit more detail 

I perhaps have over stretched it a bit but you get the Idea

i do not know how long you have been doing this , but it far outdoes my early efforts and some later ones

Keep the faith

Harry

[dhalliday]

Harry
Thanks again...
Looking at the two...with my untrained eye its hard to see a HUGE difference...but I think there is...
Nevertheless I have seen enough "L" for a while...
I have had a week or so to "think it over"....
And have SUCCEEDED in doing BOTH combinations...
So..goals met..
I will try just LONGER RGB subs...hopefully tonite,on M1...!!
(IF your alarm goes off at 2 AM,its minus 12 out and you STILL get up and go out...you are motivated

I don't know about your sky,but I can only expose about 4-5 mins in Lum...maybe 10 in RGB..
My only worry is that the NOISE in RGB subs is going to be worse,...
Exp times are longer,and signal is lower...(!!)

Harry,again,I owe you...!

Dave

[Harry page]

Hi Dave

I have a very bad sky ! and I can manage 10 min plus subs

Do you use something like a IADS filter, I highley recomend one for lum work and I use it all the time with my OSC


More to think about


Harry

[dhalliday]

Harry
Well..I have them for the DSLR... (Astronomik CLS)and an Astronomik UHC...
No one seems to like that UHC filter...but I love it in my DSLR...
(It gives me about 2.5 times the exposure,and decent color...(after correction)
May not work in the CCD..(even if I could "stick it in there" somehow...)

Having just shelled out around 500$$ for the NB set...the bank is getting low...
Again I think I may well wind up at 1800mm using RGB binned data...!
Luckily I am a HUGE fan of planetaries...
Nice to have choices..!!

If I get some decent M1 data I will post it here...
What do you think about the "post processing deconvolution"...?
I am apparently not the only one finding it can work...
THAT mystery remains largely unsolved...

Dave

[Harry page]

Hi

Not tried the post method , I will have a little play and see


Harry

[dhalliday]

Let me try and catch us some fresh (BIG/unbinned RGB) crab to try it on...
Because M1 is something I want to get well...!!!

TWO HOURS LATER...

Hmmm cannot catch much when the moon is out !!