We present ground based continuum images in the infrared, from 1.2 to
19 mu m, and an H-2 2.122 mu m line emission image of the post-AGE sta
r AFGL2688, the Cygnus Egg Nebula. We show that the standard model of
this source, comprising a fast wind focussed by a dense, equatorial, d
usty torus into a bipolar flow at position angle 15 degrees and close
to the plane of the sky, cannot explain the combination of kinematic i
nformation from previous studies and morphological information in our
own observations. Nor are the images consistent with a classical bipol
ar flow, since the apex of the two lobes observed in scattered light i
n the visible and near-IR are offset in R.A. with respect to one anoth
er. We suggest a model which is physically similar, but substantially
different geometrically, in which there is a bipolar flow at a positio
n angle closer to 60 degrees, rather than 15 degrees, still collimated
by a dense, equatorial, dusty torus, but the opening angle of the con
es out of which the Fast bipolar flow is directed is closer to 90 degr
ees, rather than 20 degrees or so as previously suggested. The bipolar
flow axis is inclined by about 20-30 degrees, rather than in the plan
e of the sky as in previous models. The dust distribution in the nebul
a has to be extremely clumpy, and there is evidence that large scale m
ass loss from the progenitor AGE star occurred in discrete phases, rec
urring on a timescale of similar to 750 years. This model implies a mu
ch lower velocity for the 'fast' bipolar outflow than does the standar
d model, which is consistent with very recent Nobeyama Millimetre Arra
y images in (CO)-C-13 emission. In support of our new model, we presen
t a full radiative transfer model for the source, in axial symmetry, w
hich reveals that the final phase of heavy mass loss included a superw
ind phase which lasted about two hundred years and removed about 0.7 M
. from the envelope of the progenitor AGE star. Our results imply that
the progenitor star must have been a relatively high mass AGB star. O
ur radiative transfer model also demonstrates convincingly that, in co
ntrast with previous models, the core of the nebula has to be exceptio
nally optically thick, with an optical depth greater than unity even a
t 10 mu m.