St. wrote:
>   Gulp! You've done another one Jim. Will you please tone it down a little? 
> It's making me reach for my Wings shortcut.  ;)
> 
>    Jokes aside, that's absolutely fantastic! You know I like my modelling 
> Jim, and well, that image just beats the pants off me for modelling, 
> artistic, and probably technical too.
> 
>       How did you do the stitching? One model duplicated several hundred(?) 
> times?
> 
>       (Before sending this, I just looked at it again, and I'll repeat the 
> gulp).
> 
> 

Y'all are embarrassing me.  But the compliments are much, much 
appreciated.  It looks like a simple enough model of a fairly modest 
shoe but it still took a remarkable amount of work.  'Took three tries 
of three separate strategies before I could get a version of the toe 
webbing that was even close!  The joins at the sole and the joins where 
the stitching crosses were reworked in several ways.

The stitching is actually approximated with little torii, scaled to look 
like little flat loops. They are oriented along splines which track the 
bottom of the creases for the stitching, which line the edges of the 
straps, and are physically modelled into the mesh.  The splines are 
generated by calculating the midpoints of the polys which form the floor 
of the crease prior to final smoothing.  These polys can be exported 
from Wings, as you know, ("export selected,") but as you may also 
realize, they cannot be exported into any set order along a path. 
Without the points in a coherent order along a path they cannot be used 
to generate a useful spline.  At first I tried sorting the polys along 
one of the dimensional components, x, y, or z.  This could work as long 
as the paths were exported in segments which did not turn back on 
themselves in the dimension.  Then I hit apon a method which seems 
obvious enough once you  think of it.  If I exported the polys in 
contiguous bands, each poly sharing vertices with the next, it was easy 
enough to reorder them like a string of adjoining dominos by searching 
for shared vertices.  All I needed to do was mark one ending poly with a 
separate texture.  So that is what I did.  I do the reordering with a 
little script in Python. The method is obviously tedious, of course, 
when it comes to manually selecting and exporting some 53 different 
paths from Wings.  (And I am still not finished, the interior stitching 
of the toe webbing is yet to be done.) Also the path of the spline does 
not adhere exactly to the surface of the smoothed mesh and there is the 
perenial problem with the roll of objects when they are oriented to a 
twisting spline.  I believe there are ways to correct this, though, when 
I get up the energy for a second pass.

The best in all of this was the ease with which I could get presentable 
renders of what I have, so far, with little set up effort using Jaime's 
includes and lighting system.  That would have been the fatal straw 
otherwise.  I hope to try some of the light dome systems also.

Post a reply to this message