 | | Performance Improvement
|
|
Since the BFC discussion has everything to do with performance improvement, I would like to propose the following changes to the primitives: - 4-4disc.dat - 2-4disc.dat Remove the point which lies in position (0,0,0), because it's NOT required and (...) (25 years ago, 5-Apr-00, to lugnet.cad.dev)
|
| |
| | Re: Performance Improvement
|
|
Oops ! wrong links. (...) Should be HTTP://is-sv.link.pt.../nstud.png (...) Should be HTTP://is-sv.link.pt.../xstud.png (...) Also I forget to mention that this reduces the number of triangles by 2 (for each stud) isn't that great ? See ya Rui (...) (25 years ago, 5-Apr-00, to lugnet.cad.dev)
|
| |
| | Re: Performance Improvement
|
|
Rui Martins <Rui.Martins@link.pt> wrote in message news:Pine.GSU.4.10.1...0@is-sv... (...) and (...) point (...) I don't see a problem with doing this for the 4-4disc, but it would be a problem for the 2-4disc. The 2-4disc is often used in (...) (25 years ago, 5-Apr-00, to lugnet.cad.dev)
|
| |
| | Re: Performance Improvement
|
|
(...) In this case you should be using 3-4disc.dat (rotate it if necessary). (...) Maybe for this case there should be a "5-8disc.dat" But that would probably mean changes to already made parts. (...) This is what is generally called a T junction, (...) (25 years ago, 5-Apr-00, to lugnet.cad.dev)
|
| |
| | Re: Performance Improvement
|
|
Rui, I think that it's better to keep the 4-4disc and 2-4disc files the way they are. The current configuration is more flexible, requires fewer special-size primitives, and is less likely to be mis-applied. The new proposal would only provide a (...) (25 years ago, 5-Apr-00, to lugnet.cad.dev)
|
| |
| | Re: Performance Improvement
|
|
Steve: (...) I can't see how Rui's suggested change gives any problems at all and I don't mind a few percent speed increase. (...) And? Does the "diameter line" not cross the radius? If there is a significant risk of a misalignment, then we (...) (25 years ago, 7-Apr-00, to lugnet.cad.dev)
|
| |
| | Re: Performance Improvement [DAT]
|
|
(...) As I said before, in this case it's more correct to use 3-4disc primitive, to avoid the T type juntions. (...) Yes, cracks may appear ! Same problem like in a T type junction But has someone said, nothing like real/concrete testing, so here (...) (25 years ago, 7-Apr-00, to lugnet.cad.dev)
|
| |
| | Re: Performance Improvement
|
|
Hi Well, i think it's also possible to substitue some quads to the triangles. But i've made some tests with l3lab and there is no differences on rendering time between the triangle version and quad version of my disc. Does someone know how quads are (...) (25 years ago, 15-Apr-00, to lugnet.cad.dev)
|
| |
| | Re: Performance Improvement
|
|
(...) Depends on the program. LDraw renders triangles by duplicating the final point, and rendering it as a quad. LDLite (I believe) splits quads into triangles. Steve (25 years ago, 16-Apr-00, to lugnet.cad.dev)
|
| |
| | Re: Performance Improvement
|
|
(...) Usually software Renderers internally use only triangles, because it simplifies the rendering algorythm, but in an hardware renderer, you can't be sure, it depends much on the hardware architecture. There a few hardware renderers which also (...) (25 years ago, 17-Apr-00, to lugnet.cad.dev)
|
| |
| | Re: Performance Improvement
|
|
Damien Duquennoy wrote... (...) L3Lab draws a quad as two triangles. You can see this yourself, when you turn off dither-transparency and use true transparency. Because the two triangles share a side, the quad-division line gets drawn twice and is (...) (25 years ago, 19-Apr-00, to lugnet.cad.dev)
|
