LDraw.org Discussion Forums
T-junction or sharp angle? - Printable Version

+- LDraw.org Discussion Forums (https://forums.ldraw.org)
+-- Forum: Models and Parts (https://forums.ldraw.org/forum-18.html)
+--- Forum: Parts Authoring (https://forums.ldraw.org/forum-19.html)
+--- Thread: T-junction or sharp angle? (/thread-17356.html)



T-junction or sharp angle? - Ronald Vallenduuk - 2015-09-23

This question may have been asked before but if it has I couldn't find it.
From the little I know about 3D modelling and rendering both t-junctions and long sharp triangles should be avoided if possible. The question is, which is worse?
See the example below; which of the three solutions is best for this shape?
[Image: 21651372842_e7d8d84481_z.jpg]
T-Junction or Sharp angle by Duq, on Flickr


Re: T-junction or sharp angle? - Damien Roux - 2015-09-24

3rd one is OK. It is not flat enough to give problems.


Re: T-junction or sharp angle? - Philippe Hurbain - 2015-09-24

Agreed. Tolerance spec for sharp angles is 0.025 degrees min.


Re: T-junction or sharp angle? - Roland Melkert - 2015-09-24

I'm not a part author but shouldn't it be using three quads in order to prevent the t-junction like so:


.png   shape.png (Size: 397 bytes / Downloads: 5)


Re: T-junction or sharp angle? - Philippe Hurbain - 2015-09-24

Problem of this solution is that it introduces new vertices -> more triangle equivalents. But sometimes it's the way to go indeed.


Re: T-junction or sharp angle? - Michael Heidemann - 2015-09-24

If you use 2 quads like in the third picture shown there are no problems at all. Why do you want to use an additionally third quad? I do not understand that in this case just as Philo say. The angle in this case is not too sharp.


Re: T-junction or sharp angle? - Roland Melkert - 2015-09-24

It depends, if it were a sticker in this exact shape it wouldn't matter but if it's ether curved or has connecting stuff it could hurt things like normal generation.

And in the case of the triangle quad combination (#3 in the example) seams are almost guaranteed resulting from the rounding differences between the coordinates in the .dat and the ones resulting from the graphic cards rasterization calculations.


Re: T-junction or sharp angle? - Ronald Vallenduuk - 2015-09-24

This doesn't prevent a T-junction, it just moves it: you now have a T-junction along the top ;-)


Re: T-junction or sharp angle? - Max Martin Richter - 2015-09-25

I can't follow you.
In my eyes this solution produces just a superfluous line, nothing else.
If you want to connect another face at the top of yellow bar, then you need to pay attention to it, otherwise you could indeed produce a new t-junction with this design. ;-)

/Max


Re: T-junction or sharp angle? - Roland Melkert - 2015-09-25

I'm confused Smile

A T-Junction, as far I know, is a corner of a triangle or quad which touches another polygon without sharing a point with it.

This can cause problems during rendering resulting from the way the graphic card calculate the pixels to be drawn. Because without a shared point it can't guarantee a closed surface due to rounding differences with the .dat source etc.

To prevent this you need to, imho, split the yellow quad in order to create the point it will share with the red quad. In this example the lines are purely to show the quad outlines, they are not physical lines.

Hope this is more clear, or I'm I missing something here?


Re: T-junction or sharp angle? - Travis Cobbs - 2015-09-25

Roland's proposal introduces a new vertex along the top edge, so whatever this geometry gets placed next to would now have a T-junction without taking that into account. This could easily be fixed by adding a triangle instead of a quad:

[Image: o702QjX.png]

I still believe that the original "sharp triangle" solution is the best.


Re: T-junction or sharp angle? - Max Martin Richter - 2015-09-25

Travis Cobbs Wrote:I still believe that the original "sharp triangle" solution is the best.

+1