Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!apple!sun-barr!newstop!sun!coherent!dplatt
From: dplatt@coherent.com (Dave Platt)
Newsgroups: comp.graphics
Subject: Re: Gray levels and color
Keywords: color gray
Message-ID: <28055@coherent.com>
Date: 14 Jul 89 17:03:55 GMT
References: <20722@hodge.UUCP> <54868@tut.cis.ohio-state.edu>
Reply-To: dplatt@coherent.com (Dave Platt)
Organization: Coherent Thought Inc., Palo Alto CA
Lines: 75

In article <54868@tut.cis.ohio-state.edu> william c ray <ray@cis.ohio-state.edu> writes:

> in a previous article someone says...  'I once knew a fella who could %
> reconstruct color data from a B/W photo, he could read resistor codes on
> b/w photos w/out the color there.  somehow he could actualy see the
> color in the grayscale, because he never guessed and was always right...
> how did he do it?"
>  
> answer... he was a better guesser than you thought...  in any
> color-to-BW scheme, there are multiple hues which will record (generic
> term here, not necc meaning screen graphics) as the same value (shade of
> gray).  ie, if the pigments were right in the bands, several bands could
> have recorded as the exact same value on a print.
> ...
> therefore, as you can see, determining which color is which from the
> grayscale that it records at is an utterly impossible proposition.
> (this means that the fella who could read resistors in B/W was GOOD...)

Well... William's points are on the mark _if_ the problem is one of
identifying arbitrary hues, chosen from the complete RGB colorspace,
from a black&white image of unknown hue->gray mapping behavior.  I'm not
sure that William's arguments rule out the problem as stated by the
original poster, though... _that_ problem is substantially simpler.

I suspect that the color-blind engineer was not simply pulling guesses
out of a hat.  Instead, he was probably depending on a several
simplifying characteristics which greatly reduced the number of possible
choices for each resistor-image.

1) His company had probably standardized on one type of black&white
   film.  Thus, once he had seen a well-populated circuit board, _and_ a
   picture of that same board, he could have memorized the mapping
   between shades-of-gray-as-he-saw-them-on-the-board and
   shades-of-gray-as-he-saw-them-on-the-picture.

2) There are only a limited number of colors used to mark resistors...
   perhaps 15 shades, perhaps as many as 20 (including silver and gold).
   Out of this many shades, there may actually be only a small number of
   "collisions", in which two shades have the same gray-scale
   representation on a specific type of black&white film.

3) The color-coding scheme for resistors is quite general; the commonest
   notation allows for two significant digits of value, a power-of-ten
   stripe, and a precision stripe.  

   However, in practice not all combinations are actually used.  For
   example, you're unlikely to find a 100-ohm resister, a 110-ohm, a
   120-ohm, ... , and a 990-ohm resistor in any real-world component.
   Instead, values that lie within any particular decade (100-1000 ohms,
   1000-10,000 ohms) tend to be drawn from a standard selection of
   "common" values.  I think that there are perhaps 10 or 12
   commonly-used values within any particular decade, rather than 90 or
   so.  This makes a good deal of sense, as the actual precision of
   resistors is often good to only 5% or 10%.  Thus, the likelihood that
   two standard-value color patterns will end up "looking the same" in a
   black&white photograph is not actually all that high.

4) Being an electronics engineer, this man probably had a good
   understanding of common circuit configurations, and might have been
   likely to recognize some circuit configurations simply by looking at
   the component layout on the board.  This might have given him some
   information about the component values that would typically be used
   in that sort of circuit.

So... all told, I think it's well within the bounds of reasonability
that a skilled electronics engineer, who was highly color-blind and thus
already used to recognizing colored objects by their gray-scale values,
could have learned to identify the corresponding photographic gray-scale
values of a relatively small population of resistor-codes under the
conditions that have been described.
-- 
Dave Platt    FIDONET:  Dave Platt on 1:204/444        VOICE: (415) 493-8805
  UUCP: ...!{ames,sun,uunet}!coherent!dplatt     DOMAIN: dplatt@coherent.com
  INTERNET:   coherent!dplatt@ames.arpa,  ...@uunet.uu.net 
  USNAIL: Coherent Thought Inc.  3350 West Bayshore #205  Palo Alto CA 94303