| |
Rui Martins <Rui.Martins@link.pt> wrote in message
news:Pine.GSU.4.10.10004131721070.339-100000@is-sv...
> > Superimposed polys are allowed, AFAIK. Maybe not the most elegant • solution,
> > but they render fine.
>
> Yes, they have been allowed until now, because as you say, parts with
> superimposed polys render fine, as long as both parts have the same color, • which
> is generally the case.
>
> I'm against superimposed polys, and I would like to propose to discourage • their
> use, because there are some major drawbacks:
>
> -First it's NOT considered good modeling practive (NOT used in B-reps).
No argument here.
> -Second, this type of superposition generates overdraw, which is what we • are
> trying to avoid with the BFC spec, because it slows down the renderer.
But the amount of superpositioning is generally small compared to the overall
size of the part.
> -third, because the polys are superimposed on each other, this usually • means
> that there will be some more points (geometry) to be processed, which also • slows
> down the renderers.
Sometimes more points, but probably more often fewer points. Consider a disc
overlapping the end of a quad which is just over half its diameter. To
prevent the quad from overlapping, you would have to cut the quad into about
7 tris and match them to the perimeter of the disc.
> - last it's NOT difficult to avoid in most of the parts we have.
I doubt most part authors would leave an overlap unless it saved them
significant time and hassle creating a part.
>
> Also in a small part (like the petal) which may be denselly used, and • because
> they also require a lot of polygons to be defined, due to their round form, • this
> part in particular should NOT use superposition of polys.
> (Imagine studs which had overdraw)
But the flower is also a case where overdraw would represent probably less
than 1% of the draw area. I can't believe this would make that much
difference in render time.
>
> I think some guidelines for correct modelling should be made, and I will • help to
> do this if needed, or maybe I can even do it and then show it to the ldraw
> comunity for comments/addition/deletions.
>
> P.S.
> It's a lot more difficult to correct already made parts, especially if • you are
> NOT the author, so we should strive to make them correct, in every sence
> possible, at first try, mainly because the author is available, and because • it's
> easyer than later trying to correct them, when the need arises.
You make some good points, but I would be concerned that we not create so
many guidelines for modeling parts that it discourages people from doing it.
It's one thing to demand that official parts _look_ as correct as possible.
It's another thing to demand that parts give optimal performance. I know
I've made performance suggestions before (usually regarding primitive use),
but I try to provide the corrective code if possible.
-John Van
|
|
Message has 1 Reply:
Message is in Reply To:
73 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
|
|
|
|
