Special:
|
[DAT] (requires LDraw-compatible viewer)
|
Subject:
|
Re: Part number needed & New primitive submission
|
Newsgroups:
|
lugnet.cad.dev
|
Date:
|
Sun, 1 Aug 1999 18:05:40 GMT
|
Viewed:
|
2133 times
|
| |
|
|
[discussion cut to lugnet.cad.dev]
On Fri, 30 Jul 1999 23:06:22 GMT, "Adam Howard" <abhoward10@hotmail.com>
wrote:
> Steve Bliss <blisses@worldnet.att.net> wrote in message
> news:379efb57.87946213@lugnet.com...
> > A few things I noticed/suggestions:
> >
> > 1. The way you use box4nl, you are drawing a bunch of quads which will
> > never be seen. Drop the end-quads from box4nl.
> > 2. On the outside frame, butted-construction is not the best way to put
> > graphics primitives together. My experience is when areas are 'beveled' • to
> > fit together at the corner-points, the results can better handle
> > rotation/scaling without visible defects. YMMV.
>
> I have rethought my original decision on both of these.
> I realize the end quads will never be seen, but is that a reason to create a
> new primitive. Even If I did create a new primitive for the lattices, some
> of them would still be embedded in the window frame and never be seen.
Yes, I feel it is a reason to change the primitive (by removing two
opposing end panels). You can easily add the needed panels to the frame
area.
> I also disagree with 'butted-construction is not the best way to put
> graphics primitives together'. I think it is an excellent method and I
> haven't seen any defects created by it. In my opinion it is exactly the
> same as writing quad lines to cover the same surfaces (there is no overlap
> in this part). Scaling and orientation will be applied equally regardless
> of how those surfaces are generated, whether they are called from a
> primitive file or written directly in the part file.
But when graphics primitives (in this case, I mean lines, quads and
triangles, not files in the ldraw\p\ directory) are rotated and scaled,
points which were previously lined up, sometimes end up out-of-line. This
is a minor, but real, problem. The way around this is to use common points
whenever possible. When drawing a 'frame' surface:
+--------+
| |
| +----+ |
| | | |
| | | |
| +----+ |
| |
+--------+
It's arguably better to make the surface from four trapezoids:
+--------+
|\ /|
| +----+ |
| | | |
| | | |
| +----+ |
|/ \|
+--------+
Rather than four rectangles:
+-+----+-+
| | | |
| +----+ |
| | | |
| | | |
| +----+ |
| | | |
+-+----+-+
Because adjoining quads use common points, which will 'stay together'
during the process of rendering the scene.
Also, put the following line in a file, and view it in LDLite (with shading
turned on):
1 14 0 0 0 1 0 0 0 1 0 0 0 1 4315.dat
Notice a bright line along the side of the end hinge? That happened
because the part was written with a single, solid quad for the surfaces
between the hinge-fingers. That quad goes from X=-40 to X=40, the entire
width of the part. That leaves a bit of extra surface on the quad, behind
the hinge-finger. Now, that surface is hidden by the hinge-finger, and
will never be displayed, so it's no big deal, right?
> I think this part meets and exceeds the quality level required for ldraw
> parts (for examples see 30044 (render and notice how stud and arch do not
> meet around the edges), look these from a blown up over view: 3899, 4528,
> and 4529).
I agree, your part meets the quality level needed, but it does have a
visible defect.
> I know you have a lot of experience writing parts, and I'm sure you have
> seen the defects you described before, but computers have come a long way
> and maybe now this is a better way to design parts.
Huh? 'Computers have come a long way'? What does that mean?
It's not a better way to design parts, it's just (slightly) easier for the
part-author.
> You realize this issue goes far beyond this part. And I think that we as a
> community need to discuss this and make a firm decision on whether this is a
> quality method for creating parts. Obviously you and I have different views
> on this issue and I would like to hear from other part authors.
I agree completely.
Steve
|
|
Message is in Reply To:
31 Messages in This Thread:
  
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
