Since nobody else will, I'll be the first to admit I have no idea how to tie that.
Z, my knot-forming photo shows the knot (in cordage,
with a small red piece meant as *highlighter*)
in some tying steps. If you get the uppermost
of these three stages, you'll understand.
And the uppermost (and most *open*/exploded)
shows the rope as coming from the right side
making a
slip knot --i.e., there is a bight of the
left-side used to complete the
overhand (for the right),
and the bight is given a half-turn (legs crossed)
and laid upon the left-side end.
SEE THIS.
Now, the completion of the directional eye knot
(it can be loaded either direction, but loading such
that the S.Part is the end pulling into twin collars
should not jam the knot --should not collapse the
collar going around the opposite, unloaded end),
comes by taking this
slip knot's bight (yeah,
squeezing it together for grip!) as it's laid
down under the left-side end (which it was laid
upon) --i.e., turning around it...-- and up through
the central nipping area.
(In these sorts of knots there can be different ways
to orient or even position the tucking of the eye-bight.)
There are many such constructs, using the
overhandor other (
fig.8) base.
And I think I've presented in one of the IGKT forums one
with "twin eyes", where the point is not only the pure
symmetry of the knot but the defeating of collapsing
a collar (that is around the unloaded end) by having
"twin" eyes --i.e., eyes each of which collars a unique
*end*, and the one that
would collapse ... , cannot,
for --"twin eyes" being loaded together qua one-- the
other eye resists the pull & denies the extension of
that eye which would result from collar material being
pulled through the nip into it.
--dl*
====