Re: [APML] 7331 (again!)

["Joseph Dartsum" <jdartsum@hotmail.com>]

>From: Joe Pedit <pedit@email.unc.edu>
>Reply-To: astro-photo@seds.org
>To: astro-photo@seds.org
>Subject: Re: [APML] 7331 (again!)
>Date: Sat, 22 Sep 2001 08:31:27 -0400
>
>Joseph Dartsum wrote:
> >
> > Can you describe your definition of signal to noise ratio, and what
> > operations in Picture Window contribute to increasing it?
> >
>
>Joe D.,
>I would be interested in hearing how you would define signal and noise
>in scanned film images?  What methods would you use to increase the
>signal to noise ratio in scanned film images?  Would you use methods
>that would alter the spatial aspect of noise (i.e., alter the range of
>the semivariogram)?
>Joe P.

It's pretty obvious..

Let's consider CCD, where each pixel has a linear response with Poisson 
counting statistics.  It's sqrt(S), where S=signal counts.  The greater the 
signal, the greater the S/N ratio.

Let's consider film, where we consider a "point" (mathematically speaking).  
It's a well known phenomenon, that you can go 1 to 2 stops overexposure and 
get a decent printable negative AND you get better (finer) grain.  (due to 
the non-uniform distribution of film grains)  This is the same situation as 
above: more exposure gives you less grain (throughout the entire negative), 
i.e. less noise characterstics in shadow..nominal.. highlight areas.   Film 
has the unfortunate additional problem of "scattering", which adds a 
Gaussian type blur to its response, which complicates the signal to noise 
issue even further.  (Take shot of an object using TP2415 and color 
film..the TP will show guiding/seeing errors, the color film will show nice, 
round stars.  The latter is film-grain based Guassian blur, which often 
misleads amateurs into thinking they are getting beautifully guided, high 
resolution images. NOT!).

The answer to your question is very simple, use a single long exposure 
negative, well guided, under optimal seeing conditions.  It's the same 
conclusion in Dr. Mike Newberry's CCD article (see CCD Astronomy article), 
where he went thru some long and detailed algebraic calculations.  Like he 
says, "it's nothing more than we already knew".  The catch, is that long 
exposures "trap" you for averaging effect of seeing, which is a killer: this 
will destory high-resolution imagery.  So, you inevitably run into a 
paradox, or what you would know as an engineering tradeoff.  Short exposures 
to "freeze" seeing (fractions of seconds), or long exposures to increase S/N 
ratio.

Fundamental result: S/N ratio increases according to S, the bigger S is the 
better.

The amateur community is misquoting a "result" on combination of images to 
improve S/N ratio, which indicates there non-understanding of BASIC 
probability theory.  I'm sure you are aware of this.  S/N ratio implies the 
*knowledge* of a probability-density function, i.e. the signal has been 
infinitely sampled to determine this (which is impossible), from which you 
can calculate the mean/std-dev = S/N ratio.  From 2 to 3 (or whatever number 
of film images you've taken), this is obviously way UNDER sampled..you have 
no idea about the measurement statistics, with so few samples.  Amateurs are 
using another definition of "signal to noise" ratio (not the strict 
definition, as used by scientists and mathematicians), which is simply their 
false non-scientific impression..pseudo-science.  It's been widely quoted in 
print, web articles by amateurs.

When Tony Hallas states the following:

Hello everyone,

    Last night I used TWO HOUR hypered Kodak Royal Gold 200 @ 90 minute
exposures of 7331 (great test object) and recorded a LOT more than the 
previous
negatives. Tracking was very good, as was the focus this time. I "polluted"
these better negatives with the original Picture Window result and came up 
with
a new image and a close up.. you can see H II regions and galaxies 
everywhere...
in the close up, the companion galaxy to the upper left even has two faint H 
II
regions showing. As the signal to noise ratio is increased through PW more
detail and color is showing up... almost CCD! :-)
*******************************************

Look at what he says.  He's got 2 new negatives (better S/N ratio) combined 
with a previous result (poorer S/N ratio).  This makes no sense.  Of course, 
the resultant image will be no better (in terms of S/N ratio), than the 
poorer image.  (See Dr. Mike Newberry's 3 part article in CCD Astronomy, on 
signal to noise ratio, it's a simple algebraic calculation at a junior high 
school level).  You don't even have to do math to understand this, anyone 
knows: "the system response is no stronger than the weakest link".  The 
"weak link", is the negative with the poorer S/N ratio.  You NEVER want to 
combine images with varying S/N ratio, period.  This is widely done in the 
amateur community, which is laughable.

If you look at his use of the word "polluted" (destroying high resolution), 
you can tell even he can tell that the resultant image is poorer than what 
he started out with.  The whole evaluation process of astrophotography is so 
SUBJECTIVE (judging images by human eye & how "aesthetic" they are), any 
critical analysis is by definition..undefined.  Tony Hallas never bothered 
to answer my question about signal to noise ratio, so that's very telling..  
His Sky & Telescope article has some glaring errors in this respect.  It's 
also a poor reflection on S&T editorial, which claims to be the "Essential 
Resource in Astronomy".  Not.

Amateurs should stay within their domain, "pretty pictures".  Do NOT go into 
an area, where you do not have any expertise, like mathematical statements 
on signal to noise ratio (or any science).  You guys are leaving a 
paper-trail, which will come back to haunt you.  It's a simple matter of 
quoting these posts, web-sites...instant loss of credibility.  At the last 
AAS (American Astronomical Society) meeting, the word from professionl 
astronomers was that collaboration between amateurs and professionals 
was...DEAD.  The automated tracking systems (meteors, comet trackers/LINEAR, 
supernova) have completely eliminated the existing form of amateur 
contributions.  The foulup of "signal to noise ratio", as given by the 
above, is further evidence that amateurs have nothing meaningful (in a 
technical sense) to contribute.  If these statements were presented in a 
technical forum (ASP, Astronomical Society of the Pacific, meeting), they 
would LAUGH AT YOU.  (the above was some simple junior high school level 
algebraic calculations.  Or, just basic systems-level concepts 
understandable by the avg layman).  There were stories floating around about 
ludicrous statements made by amateurs, which drawing a great deal of 
ridicule and laughter.  I'm afraid the Internet has highlighted the problem 
with amateurs, and put the nail-in-the-coffin, for any cooperative pro-am 
work.  Amateurs have become irrelevant.  Not only that, they are 
incriminating themselves for being a bunch of pseudo-scientists, like 
Face-On-Mars crackpots.

The "pretty picture" that was posted by Tony Hallas is just THAT.  A "pretty 
picture".  This is a nice way of saying, it has no technical merit.  (all 
the rave comments about h-alpha regions, small galaxies, are simply 
gibberish).  Try taking the same picture with a 14" Cass from space (without 
the corruptive effects of seeing, destroying high resolution).  The 
resultant image will blow away the land-based one easily.  The only effect I 
see from Tony's hypering (which was totally hit-and-miss, no quantitative 
way of graphing out the effects of hypering), was compensating for the 
Schwarzchild exposure (threshold effect) of film.  You will record fainter 
stuff.  His claims of better signal to noise, is simply gibberish.

FILM:
1. threshold effect
2. "Gaussian Blur" induced by grain scattering
3. non-linear response, "S" shaped curve
4. other effects

The only thing I saw from the posted images, was attritutable to 1.  The 
effects of 2) will always be there, & CCD will always have the edge there.

"the worst part of the telescope is the atmosphere"
..as quoted by Thierry Legault (one of the group of French high-resolution 
astro imagers)

Look at the cumulative effect of seeing, of several hrs of exposure (3 
negatives)..this plus the "Gaussian Blur" effect of film-grain scattering, 
destroys any concept of "high-resolution"  in Tony's images.  (just admiring 
the "pretty picture" is deceptive..it's not as good, as what COULD be 
obtained.  Look at Legault's Saturn & Jupiter pics, where he took image upon 
image, throwing away the corrupted images due to "seeing".  They look like 
space-borne telescope images).

Tony Hallas wrote:
< 2 second STV exposures>
< .5 sec exposures!! approaching 1/3 arc-sec resolution>

Dude, don't make me laugh, ok?  2 second exposures??  You're AVERAGING the 
cumulative effects of seeing.. While your  STV is merrily _sitting there_ 
exposing for 2 secs, your film is _measuring_ the "bouncing, osculating" 
wavefront on film.  That's image BLUR.  Does your site have 1/3 arc-sec 
resolution, or does the recent fires at your site somewhow "increase" seeing 
conditions?  Or, have you found a new site, which "blows away" professional 
sites like Las Palmas, Keck, etc.  Let's leave the hyperbolae, while 
drinking beers at the bar, OK?

I would be careful with your claims, as this is further loss of credibility 
on your part.



Joe:
to answer your question, I would simply stay away from film.  I spoke with a 
Harvard astrophysicist, who was contemplating using film for the last solar 
eclipse.  The nonlinear response, made calibration far too messy.  All 
astronomy has gone to CCD (linear response), for ease of analysis.  I think 
any attempts to do any quanitative analysis for film, is a waste of time.  
CCD has far greater dynamic range, negative film is limited at ~ 7 f-stops.  
  Film is for "pretty pictures".  As a PhD, you should not be messing with 
this non-scientific medium.  It's for the birds..bird-brain amateurs that 
is.

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


_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com/intl.asp


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