| |
"Manfred Moolhuysen" <moom@nospamchello.nl> wrote in message
news:G1AHn2.3Ir@lugnet.com...
> In lugnet.cad.dat.parts.primitives, Travis Cobbs writes:
> > 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).
>
> Please tell me what the benefit is of your even distribution of triangles
> over the currend distribution whith 9 quads and 4 triangles? To my • opinion,
> this factor alone doesn't seem to result into any severe deformations in
> renderings with Ldraw and L3Lab, and the old primitive uses a smaller • number
> of faces.
The even distribution of points means that the polygons involved stay closer
to the surface of an actual sphere at the point on them which is furthest
away from this surface (the center of the triangles/quads). Whether or not
this produces a more aesthetically pleasing result is a matter of opinion.
I can agree with you that the difference either way is not very great,
although I do personally prefer my tesselation.
Also, the existing primitive isn't really a clear winner from the standpoint
of the fact that it has less faces. Remember that a quad is inherently more
complicated than a triangle, due to the extra point, so this added
complexity has to be factored in. If you arbitrarily state that it has 4/3
the complexity of a triangle, you end up with 9 * 4/3 + 4, which
coincidentally equals 16, the number of triangles in my triangular
tesselation. Obviously, this was an entirely arbitrary calculation on my
part, and I'm not trying to imply that there is a direct correspondance with
rendering speed. However, I feel it is closer than saying that the current
version has 13 faces, and therefore is "better" than the my version with 16
faces from a performance standpoint. I think the actual speed will be very
dependent on the renderer. Some renders render quads as two separate
triangles, which would make the current sphere have 22 triangles, which is
an even worse result. On the other hand a flat quad should only have a
single surface normal, while two triangles should have 2 distinct surface
normals, so that enters into things as well.
So, in short, while I personally favor the tesselation I produced (no huge
surprise there), I don't have a strong opinion about it. All I can say is
that, whether or not it gets used as the official sphere primitive, I will
use it in my LDView program as a replacement sphere primitive when the user
asks for primitive substitution.
> I remind you that for POV renderings, the program L3P substitutes
> this primitive with a real sphere segment.
I am aware of this. In fact, my adding primitive substitution to my
real-time renderer LDView is the reason that I was able to produce this
tesselation in the first place.
--Travis Cobbs (tcobbs@san.REMOVE.rr.com)
|
|
Message has 1 Reply:
Message is in Reply To:
10 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
