Author Topic: Managment of "synthetic luminance" and RGB questions (NOW MOSTLY DBE !!)  (Read 20073 times)

Offline dhalliday

  • PixInsight Old Hand
  • ****
  • Posts: 307
    • on Flickr
I am struggling with weak data...(M101)
I have 13 hours RGB...and 4 of lum.
If I extract the ? 3 hours of synthetic lum from the RGB...is there any way to/any value in, trying to "add" the two lum results together...
How,pixel math ? (yuk :yell:)
Any feed back appreciated,as I am struggling for detail,and need a strong lum mask to play with the color data and bad sky background also..
Here is the RGB (so far)
It is pathetic...but maybe Lum can save it...
http://www.flickr.com/photos/daveh56/4516821001/
PS...if you scroll down in "comments"...the RGB tif file links are there.........

Dave H
(post #10789)
« Last Edit: 2010 April 17 18:28:20 by dhalliday »
Dave Halliday
8" Newtonian/Vixen VC200L/ TV 101,etc etc
SSAG/EQ6
CGE Pro
SBIG ST2K,ST10XME

Offline Niall Saunders

  • PTeam Member
  • PixInsight Jedi Knight
  • *****
  • Posts: 1456
  • We have cookies? Where ?
Re: Managment of "synthetic luminance"
« Reply #1 on: 2010 April 13 15:08:03 »
Hi Dave,

Yes - if you have 'extracted' the Lu from the RGB image - let's call this image "Lum1", and you have your second image, "Lum2" then, in PixelMath, all you need to do is to type in Lum1+Lum2, and that will 'add' the two luminance images together.

Enter the formula above in the <RGB/K> field **
Drop down the <Destination> panel, and untick the <Replace target image> checkbox, by ticking the <Create new image> checkbox
Click the <blue square> (which is the "Apply" button, or hit key <F5>)

And your new 'averaged' image should appear

**Alternatively, click the <Edit> button at the far end of the <RGB/K> field and, on the window that appears, click in the "--> nn images available <--" box, which will provide a drop-down selector area that will allow you to select each image - one at a time - between which you can insert a '+' sign

Don't be too afraid of PixelMath - it is just a tool like everything else in PI - and so it is a tool that will take some getting used to, but it will hopefully become a tool the you can eventually rely on.

HTH

Cheers,
Cheers,
Niall Saunders
Clinterty Observatories
Aberdeen, UK

Altair Astro GSO 10" f/8 Ritchey Chrétien CF OTA on EQ8 mount with homebrew 3D Balance and Pier
Moonfish ED80 APO & Celestron Omni XLT 120
QHY10 CCD & QHY5L-II Colour
9mm TS-OAG and Meade DSI-IIC

Offline vicent_peris

  • PTeam Member
  • PixInsight Padawan
  • ****
  • Posts: 988
    • http://www.astrofoto.es/
Re: Managment of "synthetic luminance"
« Reply #2 on: 2010 April 13 15:22:58 »
It's very easy:

First, do an average of the three channels BEFORE doing any color calibration. We have a color image with the three RGB images in each channel. In PixelMath, put this expression: ($T[0]+$T[1]+$T[2])/3. For the output, select a new greyscale image. This new image will have the identifier synLum.

Then we load the real lum image; it will be identified as Lum. You must combine this image with the synthetic luminance. This part requires a bit of work...

We create a preview of the whole Lum image and will apply a PixelMath calculation on it. The pixelmath formula is: $T + synLum*k. Activate the rescale checkbox.

Once you have applied the formula, you will calculate the noise of the resulting image. This can be done with the NoiseEvaluation script. By varying the weight (k) of the synthetic luminance, the strenght of the noise in the resulting image with also vary. You must find where is minimum.  :) Try to put different k values and see where you have the minimum noise contribution.


Regards,
Vicent.

Offline mmirot

  • PixInsight Padawan
  • ****
  • Posts: 881
Re: Managment of "synthetic luminance"
« Reply #3 on: 2010 April 13 18:03:48 »
Why can't you integrate the two images by noise weighting  without normalization?

Max

Offline Carlos Milovic

  • PTeam Member
  • PixInsight Jedi Master
  • ******
  • Posts: 2172
  • Join the dark side... we have cookies
    • http://www.astrophoto.cl
Re: Managment of "synthetic luminance"
« Reply #4 on: 2010 April 13 18:47:06 »
Just a thought... couldn't it be better to use the true Luminance as high frecuency information, and synthetic L for large scale? I'm thinking in cases like this, where a synthetic  L may be deeper, but not sharper. :)
Regards,

Carlos Milovic F.
--------------------------------
PixInsight Project Developer
http://www.pixinsight.com

Offline vicent_peris

  • PTeam Member
  • PixInsight Padawan
  • ****
  • Posts: 988
    • http://www.astrofoto.es/
Re: Managment of "synthetic luminance"
« Reply #5 on: 2010 April 13 23:50:49 »
Why can't you integrate the two images by noise weighting  without normalization?

Max

Yes.  :P I was wrongly thinking that you need three images. But that's wrong because because you're not going to do any pixel rejection.


V.

Offline dhalliday

  • PixInsight Old Hand
  • ****
  • Posts: 307
    • on Flickr
Re: Managment of "synthetic luminance"
« Reply #6 on: 2010 April 14 04:15:00 »
But isn't the "real" lum data so much stronger than the "synthetic"..?
We are not talking about "averaging" them are we..??
Do people do this combo as a routine..?
What about Juan's position on RGB vs LRGB,etc...???

I am going to digest all this advice...
What I am realizing is I need darker skies,more aperture...and a more sensitive camera...(!)

SO...to fix all this I bought a better mount...(!!)

Starting backwards...that's always my style. >:D
cheers

Dave
Dave Halliday
8" Newtonian/Vixen VC200L/ TV 101,etc etc
SSAG/EQ6
CGE Pro
SBIG ST2K,ST10XME

Offline mmirot

  • PixInsight Padawan
  • ****
  • Posts: 881
Re: Managment of "synthetic luminance"
« Reply #7 on: 2010 April 14 05:26:29 »
But isn't the "real" lum data so much stronger than the "synthetic"..?
We are not talking about "averaging" them are we?


Dave

The integration setting is on average but it is really a weighted combine based on noise levels.  The stronger S/N image would be assigned a heavier weight.


What about Juan's position on RGB vs LRGB,etc...???




That's a separate issue we all struggle with. You asked how.
 How well it works is a different question.

Hopefully the L contributes a deeper image.
Therefore you often get faint stuff without good color. 
Now what happen if you are adding extra L with syth L?
You will get a even deeper image L image without S/N in color to match. 
So now your L RGB  may have even less color in the faint areas.

That's down side to LRGB. If your RGBs are deep enough then the faint areas will  be ok.
 However, don't be surprised if you took as much acquisition time to get the RGB deep enough.
 Noise matching in the faint areas are not a issue with RGB only

Max.

Offline Niall Saunders

  • PTeam Member
  • PixInsight Jedi Knight
  • *****
  • Posts: 1456
  • We have cookies? Where ?
Re: Managment of "synthetic luminance"
« Reply #8 on: 2010 April 14 08:55:20 »
Quote
SO...to fix all this I bought a better mount...(!!)

More hardware is ALWAYS better - apart from the fact that the cloudy skies and high winds used to package every new piece of equipment at least guarantee you more time to play with PI  ::)

Cheers,
Cheers,
Niall Saunders
Clinterty Observatories
Aberdeen, UK

Altair Astro GSO 10" f/8 Ritchey Chrétien CF OTA on EQ8 mount with homebrew 3D Balance and Pier
Moonfish ED80 APO & Celestron Omni XLT 120
QHY10 CCD & QHY5L-II Colour
9mm TS-OAG and Meade DSI-IIC

Offline dhalliday

  • PixInsight Old Hand
  • ****
  • Posts: 307
    • on Flickr
Re: Managment of "synthetic luminance"
« Reply #9 on: 2010 April 14 09:45:10 »
Max
OK...its weighted...
What is the "deal" with adding lum to RGB...?
For example in the "LRGB" combo program we can adjust the contribution of "lum"...

I notice it affects the results a lot...
What is it all about...?...does lum REDUCE the color signal some how..??
Also...my RGB lacks the depth of lum data...
Maybe RGB subs should be DEEPER than lum..??
What is a "good" blend..?
Is it all by "eyeball",..??
Any further wisdom appreciated...its all a bit of a mystery..but a fascinating one.. :D

Niel;
waiting for full moon to set new mount up... >:D

Dave
Dave Halliday
8" Newtonian/Vixen VC200L/ TV 101,etc etc
SSAG/EQ6
CGE Pro
SBIG ST2K,ST10XME

Offline dhalliday

  • PixInsight Old Hand
  • ****
  • Posts: 307
    • on Flickr
Re: Managment of "synthetic luminance"
« Reply #10 on: 2010 April 14 14:46:46 »
PS
Here is what I have...(so far)
http://www.flickr.com/photos/daveh56/4521846538/sizes/o/

It is a RIDICULOUSLY long integration time,due to my LP issues.
The red data seems weak...so I am going to focus on adding MORE red data...
I am making this up as I go along...
Need feedback/advice !
I note that there are TWO places in LRGB combo to adjust lum input...
Could someone explain the difference...??

OK...I think I am figuring out....
another false alarm..!!
SEVERAL versions on my site now...but feedback here or by PM would be good...
This is 24 hours of data....
http://www.flickr.com/photos/daveh56/
« Last Edit: 2010 April 14 15:29:44 by dhalliday »
Dave Halliday
8" Newtonian/Vixen VC200L/ TV 101,etc etc
SSAG/EQ6
CGE Pro
SBIG ST2K,ST10XME

Offline Niall Saunders

  • PTeam Member
  • PixInsight Jedi Knight
  • *****
  • Posts: 1456
  • We have cookies? Where ?
Re: Managment of "synthetic luminance"
« Reply #11 on: 2010 April 14 15:44:12 »
[Off Topic]

Just had a look at your new 'baby' Dave - now I am really jealous - so, NO MORE HELP FROM ME  >:D >:D >:D
Cheers,
Niall Saunders
Clinterty Observatories
Aberdeen, UK

Altair Astro GSO 10" f/8 Ritchey Chrétien CF OTA on EQ8 mount with homebrew 3D Balance and Pier
Moonfish ED80 APO & Celestron Omni XLT 120
QHY10 CCD & QHY5L-II Colour
9mm TS-OAG and Meade DSI-IIC

Offline dhalliday

  • PixInsight Old Hand
  • ****
  • Posts: 307
    • on Flickr
Re: Managment of "synthetic luminance"
« Reply #12 on: 2010 April 14 20:14:12 »
Niall
Regarding my new gear...
They have a 30 % chance of going "blotto" once you try to use them... >:D
and so far it has only slewed around my living room...

So be nice...
Listen,I was trying to work with just the RGB data (and shooting more-best sky I have had in MONTHS !!)

When I scroll over the three different calibrated/stacked sets...I get a "k" number showing up...
What is this...?...two of them are .08 and the blue is .04...
But the blue frames have much darker sky...
Help!!
I am confused...
This is SUCH a good whack of data....hate to "cook it wrong"...
Would you like the files..?? (RGB)

I suspect some of the data (?green) is weaker) despite all being 23 frames (at 12 minutes..)
How do I decide if red (for example) needs more subs..?
BECAUSE when I do thje RGB combine...this is somehow "wieghted"...no?
The worst data ("weakest link") drags down the bunch...is this so..??
regards

humbly,penniless O:)

Dave
PS latest RGB is better;
http://www.flickr.com/photos/daveh56/4521977493/
« Last Edit: 2010 April 14 22:25:54 by dhalliday »
Dave Halliday
8" Newtonian/Vixen VC200L/ TV 101,etc etc
SSAG/EQ6
CGE Pro
SBIG ST2K,ST10XME

Offline dhalliday

  • PixInsight Old Hand
  • ****
  • Posts: 307
    • on Flickr
Re: Managment of "synthetic luminance" and RGB questions
« Reply #13 on: 2010 April 16 05:23:12 »
Niall
Have you truly abandoned me over that new gear...? :'(

I guess,judging by the lack of response to my RGB question...I am just going to go "by the (Pixinsight)  book.
Take all the data,as stacked,RGB...and combine it.No adjstment of histo,no individual DBE..etc.
 
THEN do DBE etc etc.on the whole ball of wax.
Maybe I am overthinking it all...
Would still like to know what "k" stands for... >:D

Dave
Dave Halliday
8" Newtonian/Vixen VC200L/ TV 101,etc etc
SSAG/EQ6
CGE Pro
SBIG ST2K,ST10XME

Offline Niall Saunders

  • PTeam Member
  • PixInsight Jedi Knight
  • *****
  • Posts: 1456
  • We have cookies? Where ?
Re: Managment of "synthetic luminance" and RGB questions
« Reply #14 on: 2010 April 16 06:44:13 »
Yes Dave - you are on your own (unless I receive an 'imaging' invite over to your place - private jet included [non-volcanic-ash-sensitive, of course] ;))

But, of course, you did ask a question that has not been answered - - - - -

When you 'hover' over the image and get the 'pixel inspector' window, the <RGB/K> values that you should see will be the 'ADU' values of the pixel at that pixel position in the image, with the following information being implied :-

You will only get all four values for a tri-channel imgae (i.e. RGB, even though the channels, as you know, need NOT actually contain the 'colours' that their name might imply)
The <K> reading is actually the 'on the fly extracted' "Luminance" information from that tri-colour image - based on a 'behind the scenes' <L> image from the L*ab extraction process
The <K> reading (if I recall correctly) is subject to the 'working colour space' restictions imposed on the image - which is why the 'colour weighting' should be set to 1:1:1
All readings are also subject to the 'pixel readout' options - and may be the 'average' (etc.) of more than just the 'single pixel' pointed to by the cursor
The readings, in 'PI Native' format, will be in the range [0.0, 1.0] - however, there are readout options available to allow you to re-scale this range to one that more closely suits your needs

So, basically, you can usually interpret 'K' in an 'RGB/K' context to be the 'extracted, or synthesised, Lum channel' associated with an RGB image, or just the value of the channel itself in a pure 'mono' image.

Does that help you? Can I come round and play?

Cheers
Cheers,
Niall Saunders
Clinterty Observatories
Aberdeen, UK

Altair Astro GSO 10" f/8 Ritchey Chrétien CF OTA on EQ8 mount with homebrew 3D Balance and Pier
Moonfish ED80 APO & Celestron Omni XLT 120
QHY10 CCD & QHY5L-II Colour
9mm TS-OAG and Meade DSI-IIC