Contour based formats have been around for a long while now (I have a CLI
format spec from 1993, I think). While they seem to have better accuracy
in terms of multi-curved surfaces (without the need for massive
re-triangulation), there is really no way to change a sliced file once it
has been sliced! You basically have to reconstruct the entire 3-D model
from the slices to do anything to it. Why is this bad? Look at part
reorientation for instance. Using contour based formats generated from a
CAD system, the part designer would have to choose the build orientation
instead of the RP technician. I can't imagine that's always a good idea
(no offense to the designers, of course--I have an ME background too!).
Reconstruction can add inaccuracies to the part, of course.
-Justin
On Tue, 2 Jun 1998 lblasch@opw-fc.com wrote:
>
> To list,
>
> Albert Young said: Basically, we began with the HPGL language
> utilizing such commands as X, Y, Z, pin-up,
> pin-down, and speed, and then we added line
> pattern.
>
> A question for Albert: HPGL1 or HPGL2?
>
> Concerning a new STL format, To my knowledge, every 3D CAD system,
> surface modeler and solid modeling system can output an HPGL or
> PostScript format plot file and most if not all can produce an
> accurate slice or cross section through the object or objects in
> native format. Most of these same programs can support the
> color/texture information used in photorealistic rendering. So...Why
> not use the CAD/Sculpt/Modeler program to create the data set for the
> machines to follow instead of re-inventing the wheel?
>
> As the "layer thickness" of many of the current machines gets smaller,
> the need for extra information in the build file is being eliminated.
> Near flat and flat triangles become meaningless if there is uniformity
> in the fill/hatch and overlap in the boundry trace, so why not just
> define the boundry only.
>
> The data defining the fill/hatch should be a constant (material
> specific table) within the machine that is called for by the boundry
> conditions of that layer. (Different material properties from the same
> resin for each part?)
>
> The photorealistic attributes of the object could then be mapped to
> the slice/section as the slice is generated, and either added during
> the build by a secondary coloration process or by means of resin
> material science (dyes sensitive to threshhold levels of energy or
> frequency could be in the resins).
>
> The result is that the triangle approximation of the contoured surface
> is eliminated and you are working with the model in native format so
> design changes re-use the existing data sets.
>
> What am I missing? The model accuracy (shape, color, texture,...) is
> now a function of the design software. The machines that produce the
> parts are then just like the 2d printers of today, if all you want is
> black and white at 200dpi, that's what you buy. If you want high speed
> realistic color output you pay more and buy a faster/higher resolution
> machine.
>
> Sincerely,
>
> Larry Blasch
> System Administrator for Engineering Services
>
> OPW Fueling Components
> P.O. Box 405003
> Cincinnati, OH 45240-5003 USA
>
> Voice: (513) 870-3356
> Fax: (513) 870-3338
> **********************************************************************
> Disclaimer...The views expressed are personal opinion and not those of
> OPW Fueling Components.
> **********************************************************************
> * "Always remember you're unique, just like everyone else." *
> **********************************************************************
>
> For more information about the rp-ml, see http://ltk.hut.fi/rp-ml/
>
-----------------------------------------------------------------------
Justin Kidder, Research Asst. | Automation and Robotics Laboratory
jrkst34+@pitt.edu | University of Pittsburgh
Home page: http://www.pitt.edu/~jrkst34
-----------------------------------------------------------------------
There are two rules for ultimate success in life:
1. Never tell everything you know.
For more information about the rp-ml, see http://ltk.hut.fi/rp-ml/
This archive was generated by hypermail 2.1.2 : Tue Jun 05 2001 - 22:45:48 EEST