Subject:
|
Some Words To BFC
|
Newsgroups:
|
lugnet.cad.dev
|
Date:
|
Tue, 4 Apr 2000 09:33:03 GMT
|
Highlighted:
|
(details)
|
Viewed:
|
1689 times
|
| |
|
|
Hy,
I finaly got internet-access here, and could follow the discussions about BFC.
Some words I would like to say first:
I think it doesn't make sence to blame Steve about BFC stuff, since I asked
him if the spec
is in a state that we could try it. Since it was (and I know that it's still
not complete) I tried
implementing it into MLCad.
Steve and I did several tests, found missing things (matrix inversions) and
corrected them.
So if someone likes to blame Steve, he should blame me as well since I did the
coding and therefor
made the standard realistic.
I believe we should all play together, instead of writing crazy letters.
So ... no for the real stuff!
I warn everbody to put to much additional into BFC since I found out, that the
BFC overhead is not to
underastimate. To much additional checks would cost drawing speed and therefor
make BFCed file-drawing slower.
The performance gain in MLCad using BFC varies from 0% up to 30%. This is
influenced by e.g. the actual
viewing angle of the model and the sice of the areas removed by a BFC check.
The bigger they are the more time
you win ... So in realy big models it might even happen that you need more
time drawing the model, than without
BFC statements!
My suggestion is to keep the BFC statements as simple as possible. Meaning we
should rather correct wrong defined
parts than letting the software correct wrong definitions!!!!
As an example I take the matrix-repair statement:
Checking the matrix for bad definitions is easy, but then you have to find
out, which of the values is wrong, and correct it.
The additional code to do this correction is not much, but will cost
additional calculation time.
Someone might say thats not much, but an average model has arround 500000!
triangles and quads. So we have to multiply this
small time by 500000 and then we are somewhere in the second range!
I think the BFC standard (I know there is no real) is realy good already, no
need to make additional effort and we should
try to make it even faster (not slower).
So my recomendation is: Just enjoy it, make use of it, make good parts and
everything is perfect!
Michael
|
|
Message has 5 Replies:
 | | Re: Some Words To BFC
|
| If you 're able to correct wrong things in a model so BFC can be used, is it than not possible to make a program that goes over al your parts once and corrects all the parts so that all your models will be automatically correct? Michael Lachmann (...) (25 years ago, 4-Apr-00, to lugnet.cad.dev)
| | | Re: Some Words To BFC
|
| (...) That's why I droped out of the BFC discussion a long time ago, we are making this a big issue with proposals for a new extension when we could make things much simpler. If you know that a part is wrong, instead of adding a "0 INVERSE" (...) (25 years ago, 4-Apr-00, to lugnet.cad.dev)
| | | Re: Some Words To BFC
|
| Reflecting on (my impressions of) what Steve, Mike, and Rui have been saying, it seems we have two possible directions to go. One direction, which Steve has developed, assumes we will have some files which are BFC compliant, and some which are not. (...) (25 years ago, 4-Apr-00, to lugnet.cad.dev) !
| | | Re: Some Words To BFC
|
| I've got a little question. Why are you working so hard on BFC. The rendering in Ldraw, L3lab, MLcad,.... is already very fast. Is the speed improvement gained with BFC supposed to provide the extra time, so you can turn an model in realtime, with a (...) (25 years ago, 5-Apr-00, to lugnet.cad.dev)
| | | Re: Some Words To BFC
|
| (...) What would you suggest simplifying or removing? (...) How do you mean? What is a 'bad definition' in a matrix? Having a negative determinant? And what do you have to do to correct it? My understanding was that the rendering program could (...) (25 years ago, 6-Apr-00, to lugnet.cad.dev)
|
61 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
