Subject:
|
Re: Weird Logo Issue at L3P Quality Level 3
|
Newsgroups:
|
lugnet.cad.ray
|
Date:
|
Fri, 6 Sep 2002 19:41:41 GMT
|
Viewed:
|
1133 times
|
| |
|
|
In lugnet.cad.ray, Travis Cobbs writes:
> In lugnet.cad.ray, Steve Bliss writes:
> > Considering the usefulness of modeling mirrored parts this way (one file
> > just references the other file) (it saves time and bytes), I'd say this
> > is a problem that needs resolution in software. I *think* it could
> > addressed with a POV-Ray macro, but I haven't seriously played with
> > POV-Ray in so long, I wouldn't know where to start.
>
> I believe that Lars agrees with you regarding the fact that we can't expect
> parts not to use mirroring. However, based on his knowledge of POV, fixing
> this would require a complete change in the way L3P works. It would have to
> internally flatten everything, and do the mirror checking itself. This
> would result in extremely large POV output files, not to mention being
> generally not very "clean".
No, flattening everything wouldn't be a workable solution. We had some of
that (quite awhile ago). It was scary.
> If there is a way to fix the problem using a macro, I'm sure Lars would be
> glad to hear it. However, it is my understanding that the problem with POV
> macros (at least in POV 3.1) is that they are only evaluated once. Since
> only some of the studs will be mirrored, this won't help.
Ah. Yeah, if the macros are only evaluated once, that wouldn't work either.
I'm not sure how L3P performs the translation now; it sounds like it greps
the source file for unique subfiles, repeats for each subfile until it has a
complete list of the files involved.
One complication that could be introduced into this approach would be to
treat the 'mirror of file a' as a distinct object from 'file a'. L3P could
check the transform matrix of each linetype 1 to see if it was mirrored or
not. If it's not mirrored, queue up the subfile as normal. If it is
mirrored, queue the 'mirror of <subfile>' instead. When grepping a mirrored
file, the logic would be reversed - if the transform matrix is normal, then
queue 'mirror of <subfile>', otherwise queue '<subfile>' (well, I think that
would be the needed logic).
At the bottom level, "stud_dot_dat" and "stud_dot_dat_mirror" would be two
different constructs. In the q&d implementation, stud_dot_dat_mirror would
simply be a mirrored reference to stud_dot_dat.
This approach would at worst double the size of output files. But I would
actually expect to see closer to a 25% increase in output file size (that's
a SWAG, don't quote me).
Steve
|
|
Message is in Reply To:
 | | Re: Weird Logo Issue at L3P Quality Level 3
|
| (...) I believe that Lars agrees with you regarding the fact that we can't expect parts not to use mirroring. However, based on his knowledge of POV, fixing this would require a complete change in the way L3P works. It would have to internally (...) (23 years ago, 6-Sep-02, to lugnet.cad.ray)
|
15 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
