[MF3D.FORUM:325] Re: 1/30, er, no.

[Tom Deering <tmd@xxxxxxxxxxx>]

Oleg said:
>I ran my typical shots through the general formula in
>John Bercovitz’ paper, http://werple.net.au/~kiewavly/bases.html.

snip

>The rule tracks remarkably well.

I quote the words above because I don't understand them.  The 
Bercovitz paper is cited, but then apparently dismissed, since the 
Bercovitz paper nicely shames the 1/30 rule.

You'll see what I mean. Here's a graph from that paper. Please view 
in a fixed-width font such as Monaco or Courier.

                                FAR POINT = 22 meters
                                FOCAL LENGTH = 70 mm

       100  !--------------------------------------------------------!
            !                  .                  .                  !
            !                  .                  .                  !
            !                  .          m = MAOFD solution         !
            !                  .          g = 1/30 "rule"            !
            !                  .                  .                  !
        80  ! .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .!
            !                  .                  .                  !
            !     mm           .                  .                  !
            !      mmm         .                  .                  !
            !        mmmmm     .                  .                  !
an     60  ! .  .  .  .  mmm.  .  .  .  .  .  .  .  .  .  .  .  .  .!
--         !                 mmmmmm              .                  !
b0         !                  .    mmmmmm        .                  !
            !                  .         mmm      .                  !
            !                  .           mmm    .                  !
            !                  .              mm  .                  !
        40  ! .  .  .  .  .  .  .  .  .  .  .  mm .  .  .  .  .  .  .!
            !                  .                ,mm                  !
            !     ggggggggggggggggggggggggggggggggmmggggg            !
            !                  .                  .mm                !
            !                  .                  . mm               !
        20  ! .  .  .  .  .  .  .  .  .  .  .  .  .  m  .  .  .  .  .!
            !                  .                  .  mm              !
            !                  .                  .   mm             !
            !                  .                  .    m             !
            !                  .                  .    mm            !
         0  !--------------------------------------------------------!

            0.1                1                 10                100
 
                         NEAR POINT DISTANCE (meters)

As you can see, there is one spot where the two curves intersect.  So 
it would be easy enough to "prove" the validity of the 1/30 rule by 
choosing parameters where the two lines happen to be nearby.

Oleg offers several scenarios where the 1/30 rule seems to work well:

    >Near detail:
    > Near point: 2 feet
    > Far point: 15 feet
    > True stereo base: 0.88 inches
    > 1/30 rule's base: 0.80 inches

Okay, but now change one parameter: Near point
Near point: 4 feet:
True stereo base: 2.10 inches
1/30 rule's base: 1.60 inches

Wow, that's 25% off.  And that scenario is more likely, in my 
opinion.  Or with a far point of 8 feet, all other things the same, 
the 1/30 rule is off by 20%.  It would be trivial to create a list of 
inaccurate 1/30 solutions.

As you can see from the chart above, the geometrical solution looks 
like a curve.  But the 1/30 rule always produces a straight line. 
Where they overlap, the 1/30 rule seems to work, but often it does 
not. This simple ratio is by no means a "rule."

Cordially,

Tom

http://www.deering.org