[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [APML] Pleiades response to Processing Challenge: LMC



Bert,

Thank you. It's always a pleasure when one works with such a good raw material.

We use two methods for vignetting correction. Both methods are based on a 
quite simple principle: we try to isolate all the background pixels of the 
image to obtain a vignetting correction mask. Background pixels are those 
corresponding exclusively to the sky background. An ideal, vignetting-free 
image would have a perfectly uniform distribution (exception made of local 
variations due to random distribution) of background pixels over the whole 
image. Vignetting correction is difficult (and can be terribly difficult) 
because stellar and nonstellar objects hunt us background pixels. The 
correction techniques try to figure out what the sky background should look 
like in the absence of objects.

Our first method, which is also our preferred one, is what we call "the 
object suppression method". A fundamental property of background
pixels is that they are locally (i.e., at small scale) randomly 
distributed. If this is true, then we can safely copy neighbor, nearly 
adjacent background pixels over stellar and nonstellar images without 
altering the background pixel distribution's properties, including both its 
small-scale (random) as well as global (due to vignetting) properties. In 
Photoshop, we basically create a selection comprising all the stellar 
images (through Select>Color_Range channel by channel) and then use the 
clone stamp tool to copy adjacent background pixels over stars. Then we use 
the vignetting simmetric properties to suppress nonstellar objects, again 
by copying adjacent background pixels. We use a special circular concentric 
grid to help doing all that. This method is extremely accurate, but it 
requires a considerable amount of work and can be very difficult to 
implement when nonstellar objects are covering large areas of the image.

The second method is similar to Jerry Lodriguss' vignetting correction 
method. We take a number of samples, for background pixels, to reconstruct 
the vignetting profile over each diagonal of the image. Then we set up four 
gradients (one for each diagonal), usually with 10 to 20 sampled values, 
and apply them. This method is simpler (also more elegant from an 
algorithmic point of view) but less acurate than object suppression. We use 
it when object suppression is causing us a headache.

We have in mind, since some time ago, the possibility of automating these 
procedures (or at least most of them) with a vignetting correction 
software. Time permitting...

For the Matt's LMC image, we have used object suppression for the entire 
image except its central area, which is too covered by the LMC. For these 
central regions we have used a gradient, and for the very central area we 
have had to guess a constant value, since no background pixels are 
available. As we have used both methods we talk of an "hybrid" technique.

For chromatic balance, we make a number of careful selections of background 
pixels with the lasso tool, then save this selection, and create a curves 
layer. With the Image>Histogram command, having the selection loaded, we 
can see the median values for each color channel. We adjust curves until 
the three median (*not* average!) values for the R, G and B channels are 
identical. This way we achieve a neutral sky background. This doesn't 
guarantee prefect color balance over the whole image, indeed, but is a 
necessary starting point for vignetting correction and image integration 
techniques.

Hope this helps. Regards,
______________________________________
Juan Conejero, Pleiades Astrophoto Lab
skycad@ctv.es
http://www.pleiades-astrophoto.com/en.html

At 10:04 22/05/2002 -0700, you wrote:
>A very impressive job of processing, Juan! Can you give us  more of a
>tutorial on the first set of Photoshop techniques (sample background zones,
>hybrid object suppression, etc)? Those are new terms for me. And
>congratulations to Matt for a fantastic raw image!
>Bert
>
>Bert Katzung
>katzung1@attbi.com
>www.astronomy-images.com
>
>----- Original Message -----
>From: "Juan Conejero" <skycad@ctv.es>
>To: <astro-photo@seds.org>
>Sent: Wednesday, May 22, 2002 9:21 AM
>Subject: Re: [APML] Processing Challenge: LMC
>
>
> > Matt,
> >
> > Here is our attempt on your LMC:
> >
> > For 1024x768 resolution:
> > http://www.pleiades-astrophoto.com/APML/MattBenDaniel-LMC/en.html
> > For 1280x1024 or higher:
> > http://www.pleiades-astrophoto.com/APML/MattBenDaniel-LMC/en.1024.html
> >
> > Congratulations for such a magnificient shot. A really funny and teaching
> > challenge as well. Let's see how do we do with Monte's challenge!
> > Best regards,
> > ______________________________________
> > Juan Conejero, Pleiades Astrophoto Lab
> > skycad@ctv.es
> > http://www.pleiades-astrophoto.com/en.html
> >
> > At 21:39 14/05/2002 -0400, you wrote:
> > >Below is a URL to an unprocessed 8.5MB TIFF of the Large Magellanic
>Cloud.
> > >Also below is a link to a processed version of the image. I chose this
> > >image because I could not make it look as nice as I want. I am hoping
>that
> > >y'all will show me a thing or two with it. That said, this is a difficult
> > >image. It has major vignetting and red fogging (probably from film age
>and
> > >or humidity). If you're new to image processing, this is a tough image to
> > >start with.
> > >
> > >Because there is some nice stuff on the edges, I'm hoping that you won't
> > >crop your version.
> > >
> > >One thing I don't have yet is SGBNR or Grain Surgery. I think that would
> > >help this one a lot.
> > >
> > >You know that if you do a good job with it, the folks on the APML
> > >(especially me) will want to know what you did in gory detail.
> > >
> > >I found that WinZip did not compress the file very much, so I didn't
> > >bother with it. If you're on a 56K modem, it will take at least half an
> > >hour to download. I downloaded the image and opened it in Photoshop to
> > >test its validity.
> > >
> > >http://world.std.com/~mattb/lmc.tif
> > >
> > >http://world.std.com/~mattb/gallery/astro/lmc.html
> > >--
> > >Matt BenDaniel
> > >matt@starmatt.com
> > >http://starmatt.com
> > >
> > >
> > >--  APML Archives at <http://astro.umsystem.edu/apml/>  ---
> > >              Unsubscribe at <majordomo@seds.org>
> >
> >
> >
> > --  APML Archives at <http://astro.umsystem.edu/apml/>  ---
> >              Unsubscribe at <majordomo@seds.org>
> >
>
>
>
>--  APML Archives at <http://astro.umsystem.edu/apml/>  ---
>              Unsubscribe at <majordomo@seds.org>



--  APML Archives at <http://astro.umsystem.edu/apml/>  ---
             Unsubscribe at <majordomo@seds.org>