Author Topic: GREYCstoration vs SCNR....  (Read 7534 times)

Offline dhalliday

  • PixInsight Old Hand
  • ****
  • Posts: 307
    • on Flickr
GREYCstoration vs SCNR....
« on: 2009 September 16 02:19:18 »
Another fairly newbie question....
I have been experimenting with this Cstoration thingy...
Aside from having a tendency to make my stars go all screwy,at certain settings...it seems a good NR tool.
Of course the choice of settings is,so far,mostly Greek to me  :laugh:

How does this differ from SCNR...ie when would you use one or the other,or can you use both...
As a Pix novice I am pretty confused...
Could I have one of your (moderators) wonderfully dry/concise explanations of how they work/differ/strengths/weaknesses etc?
Is there a tutorial on the subject of NR somewhere?
Could we get a dialogue on this going here??
As a greenhorn,obviously NR is one of my favorite tools...!!
Just the other day I finally tried "Chromatic"...
What a milestone !!

thanks
Dave
PS here is the image I am currently working on,just to give you an idea of where I am at...it has both processes applied...

http://www.flickr.com/photos/daveh56/3925728886/sizes/l/
« Last Edit: 2009 September 16 03:24:47 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: GREYCstoration vs SCNR....
« Reply #1 on: 2009 September 16 02:20:47 »
« Last Edit: 2009 September 16 03:25:13 by dhalliday »
Dave Halliday
8" Newtonian/Vixen VC200L/ TV 101,etc etc
SSAG/EQ6
CGE Pro
SBIG ST2K,ST10XME

Offline Carlos Milovic

  • PTeam Member
  • PixInsight Jedi Master
  • ******
  • Posts: 2172
  • Join the dark side... we have cookies
    • http://www.astrophoto.cl
Re: GREYCstoration vs SCNR....
« Reply #2 on: 2009 September 16 06:53:49 »
Hi Dave

In fact, a comparison between GREYC and SCNR is "very unfair". The reason is that they target different types of noise.

Let me start with SCNR, witch is simpler. The fundamental idea here is to deal with a very specific kind of noise: green noise. In our astronomical color images is very strange that we have green object. Even further, is is uncommon to have any portion of the image where the color hue indicates that the G channel has a higher value than the others. So, this is the main application: remove the excess of green. Take a look at the documentation for PixInsight LE for a more detailed explanation. The current implementation has a few more options, but is quite the same.

Now, GREYCStoration. This is a filter that has been designed to deal with the noise that is generated in common digital cameras. I don't know exactly how it works internally, but in some sophisticated way, it tries to delete high frequency (small scale) objects that seems to be unrelated (i.e. noise). Unfortunately, small/dim stars, specially if under-sampled, get right into that category. So, when using this filter you have to fine tune the parameters, specially the amplitude, sharpen (protection) and size. Another option, that may be considered paralelally, is to build a inverse luminance mask, or a star mask.

Another weapon in your arsenal, much more similar to GREYC is ACDNR. It is the successor of SGBNR, a famous noise reduction filter that Juan invented. See PI LE's documentation for a description. The current ACDNR has a lot of changes, but the idea behind is the same: blur the image but protect all the borders that are higher than a certain amount, and try to locate and protect the stars. As you may already guess, ACDNR has been designed for astronomical images, so in difficult images it may be right tool.
See here for a tutorial on this tool:
http://pixinsight.com/forum/index.php?topic=347.0

Regards,

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

Offline dhalliday

  • PixInsight Old Hand
  • ****
  • Posts: 307
    • on Flickr
Re: GREYCstoration vs SCNR....
« Reply #3 on: 2009 September 16 09:49:07 »
Carlos
Thanks...
In fact the ACDNR was what I was writing about...sorry..
So is it reasonable to use both? Always?
L masking with both?
are they complimentary ?
And breifly...which setting in the GREYC is most likely making my bright stars go all "screwy" looking...?
As far as  SCNR...I never see a difference after I apply it...!!!   (?color blindness test?)
So I tend to forget it,often...


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

Offline Carlos Milovic

  • PTeam Member
  • PixInsight Jedi Master
  • ******
  • Posts: 2172
  • Join the dark side... we have cookies
    • http://www.astrophoto.cl
Re: GREYCstoration vs SCNR....
« Reply #4 on: 2009 September 16 10:03:32 »
Yes, they may be used in conjuction. This is up to you. In my case, I always run GREYC first, but if the noise is too big, ACDNR comes to help. Rarely I use SCNR, because I prefer to try other algorithms first, so the overal color balance is not affected (I'm supposing that green pixels are noise in the flux, not side effects of electronics, wich should be dealed with bias and darks). But again, this is just my average behavior ;) Each image is a whole new world. At the end, use what works better for you, and the image.

About masking, it is very advisable to do so. If you had rid off the hot pixels, all you have left is flux noise (I'm simplifying a lot this problem). This kind of noise is inversaly proportional to the brightness of the pixels, in terms of the signal/noise ratio. In fact, this ratio grows with the square root of the signal, so it is safe just to say that bright objects have less noise than the background sky or dim objects. This is why a inverted luminance mask works so well. You apply a more aggressive noise reduction where it is needed, and protect those objects that are very unlikelly to contain noise.

Finally, about parameters... cannot say with certainity. Maybe the noise size has been set too large, or the amplitude is too high. Maybe the sharpen is too low... Anyway, to avoid that, use a mask.
Regards,

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

Offline dhalliday

  • PixInsight Old Hand
  • ****
  • Posts: 307
    • on Flickr
Re: GREYCstoration vs SCNR....
« Reply #5 on: 2009 September 16 13:14:12 »
Thanks again Carlos...I guess I was intuitively doing about the same as you suggested !! ;D

But its nice to hear it...
Any word on when Pix may be able to calibrate/debayer/stack etc etc DSLR  images..??
I am trying to understand the various components of my images,dark signal,noise,quality of darks etc...
Obviously I am a guy who is "not quite there yet"....in terms of coming to grips with these,so far,invisible factors..

Does Pix offer any "measuring techniques" for assessing images,or stacking batches of disparate exposure images?  ....
Some way of giving me an idea if certain darks will "match" certain lights? (for example...) I am a DSLR user...
I am using DSS for stacking,and calibration....I had occasion to want to stack some six,AND eight minute subs...
This led to a great discussion about being able to stack them if they were "individually calibrated"....Some said yes/some no,both agreed DSS could not do it...
I realize avoiding exposure mismatch is probably the key... >:D
I understand other programs do have some of these mysterious features...CCD stack,Images Plus etc...

Any comments welcome,and sorry to ask so many questions...but I find it kinda quiet here...
Is there some other place Pix folks hang out and chew this stuff over??

I love your software,no issue with what it CAN do,just wondering...

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

Offline Carlos Milovic

  • PTeam Member
  • PixInsight Jedi Master
  • ******
  • Posts: 2172
  • Join the dark side... we have cookies
    • http://www.astrophoto.cl
Re: GREYCstoration vs SCNR....
« Reply #6 on: 2009 September 16 19:52:42 »
Quote
Any word on when Pix may be able to calibrate/debayer/stack etc etc DSLR  images..??
Not from me :)

I think that DSS reads from the metadata the exposure time... so, at the end it is just a matter of simply rescaling.


The theory behind darks/light matching is quite simple. If the temperature is the same, then it is a linear function of the exposure time. Just to play it safe, take the darks with the longest exposure time you need to match the lights, and then scale down (divide the pixel values). More clever algorithms takes into account some nonlinearity due to temperature, etc., and performs a statistical analysis of the images to give the best scale factor that is possible. BTW, I'm talking about master darks, that have the master bias subtracted. Now, given that you have perfectly calibrated images, it doesn't matter at all the exposure time. Again, it is a linear function, so if the program reads the exposure time, may scale them automatically. There too are some algorithms that perform statistical adjustments... For example, in "our" ImageIntegration process. Also you may disable those adjustments, and manually set a weighting factor, that takes into account the different exposure times.
Another option, for very different times is HDRCombination. Here you don't want just to "average" the result, but get the best of the images to create a larger dynamic range.

My choise, with DSS would be to calibrate each set of exposure times individually. Then, come to PI with those raw (calibrated) images, and integrate them there (let DSS do the registration and integration of each set). Of course, you'll need first to check if you don't need to register them first :). If so, StarAligment or DynamicAligment.

Anyway, all of this is not secret... image calibration is pretty straightforward (at least, forgetting non linearities from temperature variations), and there is nothing more complex that +,-,* and / operations :)


Don't have to apologize! This is this forum for, discussion about image processing.
About other forums... Currently I'm only posting here. The reason: Time is limited :) and I had not been doing active astrophotography for more than a year. I read some other forums, from time to time, but cannot follow or participate there actively (Cloudynights and FotografiaAstronomica, for example). I know that Juan very rarely reads other forums, and do not post anywhere but here, mainly as "business policy". Also this behavior yields a better general mood, and less preoccupations :D :D :D

PS: We are always glad to hear comments/suggestions/critics, etc. So, don't worry and just post here anything :) We are working with PI not for the money, but because this is the thing we like the most.
Regards,

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

Offline Niall Saunders

  • PTeam Member
  • PixInsight Jedi Knight
  • *****
  • Posts: 1456
  • We have cookies? Where ?
Re: GREYCstoration vs SCNR....
« Reply #7 on: 2009 September 23 05:43:19 »
Quote
We are working with PI not for the money, but because this is the thing we like the most.

And that is the single BEST part about being a PI user - if 'you guys' were doing this 'for the money' I reckon PI would never have become what it is now.

You know that we are all very grateful.

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 Simon Hicks

  • PixInsight Old Hand
  • ****
  • Posts: 333
Re: GREYCstoration vs SCNR....
« Reply #8 on: 2009 September 23 06:17:01 »
Seconded  8)

Offline Carlos Milovic

  • PTeam Member
  • PixInsight Jedi Master
  • ******
  • Posts: 2172
  • Join the dark side... we have cookies
    • http://www.astrophoto.cl
Re: GREYCstoration vs SCNR....
« Reply #9 on: 2009 September 23 06:34:05 »
Suddenly I'm feeling like the bass player in a rock band (with Juan the leader guitar and singer :D ).

 ::)


Regards,

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

Offline Niall Saunders

  • PTeam Member
  • PixInsight Jedi Knight
  • *****
  • Posts: 1456
  • We have cookies? Where ?
Re: GREYCstoration vs SCNR....
« Reply #10 on: 2009 September 23 06:48:30 »
Yes Carlos,

But who is having the most fun? You guys on stage, or us guys out front - with the long greasy hair and the jeans that we never wash  O0

At least you know that you would still be making good music - but hearing 'our roar' when you start your drum solo in mid concert - isn't that all the encouragement that you need  :moneyinmouth: :moneyinmouth: :moneyinmouth:

Keep on rockin . . . . . . .

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