[FMO] Re: Stardust@home experiences

  • From: "Matson, Robert" <ROBERT.D.MATSON@xxxxxxxx>
  • To: fmo@xxxxxxxxxxxxx
  • Date: Thu, 3 Aug 2006 12:20:36 -0700

Hi All,

Some comments on other members comments:

Marco:

"* I had a number of movies where the sharp surface focus was way at 
the bottom  of the bar. I reported them as "bad focus" but wonder 
whether these were  actually movies with the image sequence the wrong 
way around.

Tavi wrote:

> Marco, I agree. I suspect those movies with the sharp focus at the
bottom
> were actually reversed or inverted sequences.

No, it just means the range of focus was bad.  I've reviewed close to
700
real movies, and all of them had deepest focus at the bottom.  One way
to tell --  when the focus point is above the tile surface, tile
irregularities/dust/etc. can appear white in the center, dark on the
edges.  When focus is below the surface (e.g. further from the
microscope),
the white centers do not occur.  In other words, out-of-focus blur is
assymmetric.  So if the "best focus" appears at the bottom of the
slider,
I'd mark the movie as "bad focus".

One pet peeve on the calibration movies is that after you've reviewed
a hundred or more movies, you're never going to get one of these
wrong.  They're too recognizable since there are so few of them.
And even if they were hard to distinguish from real movies, they TELL
you when you've got a test movie, thus defeating the purpose of
blind testing!  (Most people probably don't notice this, fortunately,
but for anyone with a count above 100 the specificity and sensitivity
numbers no longer mean anything.)  --Rob

 
 
 
-----------------------------------------------------------------------------
** The FMO Mailinglist ** For information on the list, subscribe/unsubscribe 
info etc. go to http://home.wanadoo.nl/marco.langbroek/fmo.html ** Post 
messages to: fmo@xxxxxxxxxxxxx ** list admin: Marco Langbroek, 
fmo-admins@xxxxxxxxxxxxx ** Message archive: 
//www.freelists.org/archives/fmo/ ** 
------------------------------------------------------------------------------

Other related posts: