Re: direct slicing

From: Yong Suh (Rensselaer  Polytechnic Institute)
To: RP-ML, Ron Jamieson (Cranfield  University)
Date: Friday, September 30, 1994
Subject: Re: direct slicing
 
> I still believe that the way forward is to use adaptive slicing of the
> SOLID model. Engineers are all being forced into using solids based CAD
> and now that we get more 'bangs for our bucks' I dont think adaptive
> slicing routines are an unreasonable overhead. They take longer to produce
> than an SLA file but have the
> advantage of offering more compact files and of course, I remain in control of
> the manufacture. Non of those repair routines we wont be told about,
> fixing a critical part of the design!

 I agree. The year the STL file was proposed was around 1987 and at that time
 full featured solid modeling packages were not widely available and
 not very much powerful. Nowadays, a lot of CAD pkgs are available and 
 they have tools to slice the part accurately without loss of geometric
 information. I don't know why they want to stick to the format that is
 not only inaccurate (because of the facet approximation) but also voluminous.

> Anyway I have voted with my feet as it were and we now produce sliced data in
> the EOS cli(2) format direct from Parasolid, which is processed without
> trouble. Whaty about CV Intergraph etc users. Has anyone else written
> programmes to slice your favourite CAD packages? If so are you happy
> with the results? Do you think this is the way to go forward?
> 
  I've implemented the adaptive slicing with ACIS solid modeling
  kernel and the result was presented at the poster session of
  Solid Freeform Fabrication Symposium '94, Austin, TX. 

  In the software, user specifies an available slicing thickness 
  range (which is determined by the hardware capability) and the allowable 
  deviation from true surface as a tolerance. Then it calculates 
  the exact contours of slices from the exact part model.
  The slicing processing time of most of the example parts (not so complex, 
  but contain curved surfaces, sizes were around 10"x10"x10". One example
  was an engine piston head with inner curved surfaces) was within the order 
  of minuites of CPU time on a Sparc 10 machine. The slice thickness range 
  was around 0.001" - 0.020".

  The advantage of adaptive slicing outshines when you specify more tight
  tolerance. It generates far less number of slices than uniform slicing
  still satisfying the tolerance.  I think this is a must tool if you
  want an accurate part.

  The ACIS is used by many CAD packages(Aries, AutoCAD,etc) as their solid 
  modeling kernel. The slicing routine in ACIS is not optimized for the 
  rapid prototyping slicing process in terms of its algorithm. I think it
  is not difficult to optimize the routine for the process, because the 
  previous slicing contour can give pretty much accurate initial guesses to
  the next slicing to accelerate calculating intersections. Then, you'll
  get even faster processing time.

  Thanks,
  Yong S. Suh
  Graduate Student			email suh@cat.rpi.edu
  Rensselaer Polytechnic Institute
  Troy, NY 12180