P3D Re: Converting 2D into Simulated 3D

[Robert Thorpe <thorpe@xxxxxxxx>]

Dan Vint wrote:
> 
> There are some software packages that do what you are talking about. I have
> no experience with them.

I got the demos of Canoma and Photo Modeler Pro 3.0. This is a little
while ago now. At the time, Canoma did not handle cylinrical or 
spherical primatives. That is, you could only constuct your shapes
from boxes etc. This is a definite drawback if you want to include
something other than buildings.

Photo Modeler would almost work. It had a number of features that did
not go quite far enough though. Mostly it was for constructing
rotatable 3D models that were created from 2 or more source pictures.
It had a feature to construct the model from a single picture, such
as skid marks at an accident scene. This feature however, required
that you have real spacial coordinates for enough objects in the 
picture to establish the geometry. In this process you discovered
a spreadsheet-like screen and you could alter the coordinates in 
the grid (these corresponded to points you identified on the original
photo) but there was no provision for filling it in from scratch.

If this had been the case, you could have outlined the objects in a
single
picture, establishing XY coordinates for the node points. They would 
all then have a Z or depth coordinate of 0. with all the XY coordinates
determined it would be nice to be able to assign each node point a 
depth coordinate, thereby situating it front to back. Once the overall 
scene had been established, you could go back and fiddle the "depth"
of the individual points to simulate "roundness". With the model 
established you could generate two views and create the 3D pair...

so close and yet so far still...

======================
Robert Thorpe
Cedar Rapids, IA
thorpe@xxxxxxxx
http://www.skep.com