[MF3D.FORUM:1142] Re: Deviation and Viewers

["Michael K. Davis" <zilch0@xxxxxxxxxxxx>]

David,


At 10:57 PM 8/22/00 -0700, you wrote:
[snip]
>If I were shooting with just one camera I probably wouldn't
>bother with calculations at all and would simply shoot 3 or 4 exposures for
>each image, varying the positions so that there would be 3 to 6 different
>separations to choose from. When you figure out the deviation you tend to
>like, then you can work on making the formula fit with your perceptions, or
>forget about it entirely. The variation of the formula that I talked about
>at my workshop at the convention was based on simplifying the math so that
>one could easily get results that were within an acceptable range. Steve
>Berezin will have my handout on his website (it may already be there, but I
>don't think it is).

Thanks for the reply - your comments sound VERY practical.  I hadn't really
thought about "bracketing" the separation.

>From my other reading, it can tell I won't be able to reach a
one-size-fits-all figure for separation.  I suspect I should start out as
you suggest, but do so for each of various near-far distance ranges.  In
other words, a choice I make from among several different separations taken
of a scene that had Infinity in it, wouldn't be applicable to a scene that
runs from 10 feet to 20 feet.  Using this bracketing approach to
determining separation, how many distance ranges should I gather data for
(assuming I'm not interested in macro - can't focus closer than 1 meter)?
Would three or four ranges be enough?  

I think I remember your article had a formula for each of three ranges...
I can't remember the breakpoints at the moment, but having played with them
at length, I do remember feeling uncomfortable with the disparity between
separations calculated with your "long" formula and your abbreviated
formulae - especially at the cusps of the distance ranges you designed them
for.  

If an abbreviated formula is shorter (and easier to remember!) because of
things reducing to 1 when certain assumptions are made, for example, it
bothers me to find a large difference between the results had with the
abbreviated version and its big brother.  It makes me want to just use the
longer version at all times.  I do understand the value had in taking a
compromise that affects accuracy (or perhaps I should just say "that
affects consistency, or predictablilty"), but I tend to be most comfortable
getting all the pegs into their holes.  

Ultimately, I'm still willing to venture that your suggestion of shooting
multiple separations is an unbeatable - I can redirect my obsessiveness
into doing so for LOTS of near-far ranges.  

I look forward to your upcoming publication.

Thanks!

Mike