Subject:
|
Re: Primitives to Hi-Res or not...
|
Newsgroups:
|
lugnet.cad.dev
|
Date:
|
Tue, 25 Jan 2000 12:59:31 GMT
|
Viewed:
|
595 times
|
| |
|
|
In lugnet.cad.dev, Imre Papp writes:
> In lugnet.cad.dev, Paul Easter writes:
> > Ok, Can some one tell me where we are heading with these Hi-Res primitives.
> >
> > I welcome all comments,
> >
> > Paul
>
> My opinion is that we do need high res circular primitives (edge, disc, cyli,
> etc) in cases where the part includes curves/surfaces with a significantly
> larger diameter than 'usual'.
>
> The good old STUD.dat (and its familiy) looks OK with 16 sided circular
> primitives. But John Jensen's impressive 2951.dat bucket part proved that we
> need 48 sided primitives to obtain acceptable resolution with larger a
> diameter.
>
> Further we might need even higher resolution primitives (96?, 120?) if we
> start modelling curved train tracks or road patterned baseplates. ( I remember
> that somebody had problems with that.)
>
> My suggestion is to informally accept a rule of thumb (is it a right
> expression?) that one side (or section) of the curved line/surface should not
> be longer than cca. 10 LDU, measured when it is already built in the part
> (i.e. not in the primitive itself). This looks nice when rendered in Ldraw or
> Ldlite.
>
> With this simple rule we could verify that the modelled curve/surface smooth
> enough to yield an acceptable resolution when rendered.
>
>
> Part modellers then will have three options:
>
> 1. Choose from the existing resolutions (16 sided, 48 sided so far)
>
> [According to the above rule 16 sided primitives are good up to cca. 50 LDU in
> diameter, while 48 sided prmitives are good up to cca. 150 LDU in diameter.]
>
> 2. Manually create a private subpart or use inlined curves/surfaces, if he/she
> needs curve/surface with larger diameter or if the rendering performance is
> not acceptable using official primitives.
>
> 3. Start officializing primitives with other resolutions.
>
> This is only an idea, what do you think of it?
>
> Bye
> Ampi
>
>
> ---------------------------------------
> Imre Papp
> Geometria GIS Systems House
> email: ipapp@geometria.hu
> ---------------------------------------
There are two issues here.
a) what resolution is appropriate to realistically render a part,
b) how are the circular features of that a likely to connect/interact with
other parts or feature of the same part.
Issue b) is why we cannot universally substitute the 48-agon primitives for the
16-agon versions. Only the simplest of parts will have used the ndis primitives
to abut with disc or cyli primitives - in many cases the co-ordinates of the
16-agon will have been coded into quads and triangles within the part. This is
especially true of joined cylinders, like the shield handles.
There do need to be some guidelines here to ensure that curves of similar
diameter (particularly on the joining features of a part) are scaled similarly.
For instance a macaroni brick coded using 16-agons will not look good placed on
a 4x4 turntable coded using 48-agons. IIRC the existing 4stud (80LDu) diameter
parts use 16-agons - are these now considered unacceptable? Maybe 80Ldu (not
50Ldu) should be the threshold? (Note that if we do change these, the inner
curve of the macaroni brick still need to use 16-agons to match with the 2x2
round brick.)
I agree we should stick to 16-agons for small (stud-scale) curves, and create
48-agon primitives on an as-needed basis for larger parts which have
significant circular elements. Unless there is a significant amount of reuse, I
see no benefit in option 3 at this time. For example, the required ring
primitives for the curved lines on the decorated baseplates or the train rails
are unlikely, IMHO, to used elsewhere, and they do not interact with any other
parts, so may as well be coded inline at an appropriate resolution by the
author. Same with the monorail track which I would love to find time to do.
Chris
|
|
Message is in Reply To:
 | | Re: Primitives to Hi-Res or not...
|
| (...) My opinion is that we do need high res circular primitives (edge, disc, cyli, etc) in cases where the part includes curves/surfaces with a significantly larger diameter than 'usual'. The good old STUD.dat (and its familiy) looks OK with 16 (...) (25 years ago, 25-Jan-00, to lugnet.cad.dev)
|
7 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
