| |
In lugnet.cad.dat.parts.primitives, Travis Cobbs writes:
> I started a discussion on the .dev group regarding sphere tesselation a
> little while back, and it came out that there is some dissatisfaction with
> the current sphere primitive. I have finished the code which generates a
> replacement sphere primitive, and hacked together a little routine to output
> it to a .dat file. Below is the result.
>
> Please note that I have not yet generated linetype 5 lines. I will only go
> to the trouble to do that if there is interest in using my generated version
> as a replacement for the original one. I generated the data to 3 decimal
> places. Is this an appropriate number?
>
> One thing that should be noted is that a triangular tesselation of a sphere
> which contains 16 segments around its equator produces a sphere with
> significantly different break-down of facets than the existing sphere file.
> The existing 1-8sphe.dat file contains 9 quads and 4 triangles. The
> individual triangles are each less than half the surface area of the
> individual quads, producing an unbalanced distribution. In addition, the
> quads are either non-planar, don't fall on the surface of an actual sphere,
> or both. My replacement tesselation is made up of 16 triangles, all of
> which have a similar surface area. In addition, all of the vertices fall on
> the surface of an actual sphere (once you account for rounding).
>
> Here is my file:
>
> 0 1/8 Sphere
> 0 Author: Travis Cobbs
> 0 Incomplete suggested replacement for 1-8sphe.dat primitive.
> 0
> 0 I will produce linetype 5 lines if there is interest in
> 0 using this as an actual replacement for 1-8sphe.dat.
> 0
> 0 This data was generated using the algorithm used by LDView to
> 0 perform sphere primitive replacement. (The sphere primitive
> 0 replacement feature has not yet been released.)
> 0
> 3 16 1.000 0.000 0.000 0.924 0.000 0.383 0.924 0.383 0.000
> 3 16 0.924 0.000 0.383 0.784 0.439 0.439 0.924 0.383 0.000
> 3 16 0.924 0.000 0.383 0.707 0.000 0.707 0.784 0.439 0.439
> 3 16 0.707 0.000 0.707 0.439 0.439 0.784 0.784 0.439 0.439
> 3 16 0.707 0.000 0.707 0.383 0.000 0.924 0.439 0.439 0.784
> 3 16 0.383 0.000 0.924 0.000 0.383 0.924 0.439 0.439 0.784
> 3 16 0.383 0.000 0.924 0.000 0.000 1.000 0.000 0.383 0.924
> 3 16 0.924 0.383 0.000 0.784 0.439 0.439 0.707 0.707 0.000
> 3 16 0.784 0.439 0.439 0.439 0.784 0.439 0.707 0.707 0.000
> 3 16 0.784 0.439 0.439 0.439 0.439 0.784 0.439 0.784 0.439
> 3 16 0.439 0.439 0.784 0.000 0.707 0.707 0.439 0.784 0.439
> 3 16 0.439 0.439 0.784 0.000 0.383 0.924 0.000 0.707 0.707
> 3 16 0.707 0.707 0.000 0.439 0.784 0.439 0.383 0.924 0.000
> 3 16 0.439 0.784 0.439 0.000 0.924 0.383 0.383 0.924 0.000
> 3 16 0.439 0.784 0.439 0.000 0.707 0.707 0.000 0.924 0.383
> 3 16 0.383 0.924 0.000 0.000 0.924 0.383 0.000 1.000 0.000
>
>
> --Travis Cobbs (tcobbs@san.REMOVE.rr.com)
I think these should have 4 decimal places, as with the majority of the
circular primitives. The reason for this is that there is often a need to use
3DPs in parts. Where these include circular primitives and scale them up by a
factor of 10 you lose one decimal place. It will also help to avoid the little
gaps between joined cylinders/disc and spheres, when rendered with programs
that dont do primitive substitution.
The primitive itself looks a big improvement. When I converted all the circular
primitives to 4DPs, I avoided the sphericals because of doubts over the
planarity of the quads.
Which reminds me of something I suggested at the time I did this (but admit
have not been very good at following in my own parts). When joining
quads/triangles to circular primitives, try to avoid calculating the
co-ordinates based on the actual radius (eg. round(6*sin(22.5), 0.001), but
rather round the sin and then multiply (eg. 6*(round( sin(22.5), 0.001). That
way the cordinates you use will match the scaled up primitive. I'll include
this in the authoring FAQ.
Chris
|
|
Message is in Reply To:
 | | Possible replacement for 1-8sphe.dat [DAT]
|
| I started a discussion on the .dev group regarding sphere tesselation a little while back, and it came out that there is some dissatisfaction with the current sphere primitive. I have finished the code which generates a replacement sphere primitive, (...) (24 years ago, 21-Sep-00, to lugnet.cad.dat.parts.primitives)
|
10 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
