92946 vs. 15672

[Willy Tschager]

Hi,

I'd like to discuss the issue raised in:

https://www.ldraw.org/cgi-bin/ptdetail.cgi?s=92946

and explained in deep in:

https://rebrickable.com/blog/107/mysterious-slope/

I therefore modified 92946 as well as 15672 by adding a 0.5 LDU gap to the slopes as well as rounding the edges.

   

Have also a look at the parts:

Slope Plate 45  2 x  1
  15672.dat (Size: 2.86 KB / Downloads: 3)
Slope Plate 45.73  2 x  1
  92946.dat (Size: 2.81 KB / Downloads: 4)

If you're fine with the solution I'm gonna upload the files as well as obsolete the already official:

https://www.ldraw.org/parts/official-par...s/92946s01

as all prints are achieved by stickers.

w.

[Philippe Hurbain]

Seems good to me. I just suggest to change sides triangulation to

Code:

4 16 10 0 -.5 10 8 -20 10 8 20 10 0 20
4 16 10 4 -20 10 0 -.5 10 -14 -.5 10 -14 -2.45
3 16 10 4 -20 10 8 -20 10 0 -.5

as it provides better smoothing.

[Magnus Forsberg]

I haven't dug into your designs yet, but how do you deal with the fact that all ldraw 45 sloped parts are wrong?
Pic14-16 in that article show that the tip shoud have a height of 5 ldu, but all of our 45 slopes have only a 4 ldu height.
Moving the wall forward an lovering the top is maybe not the correct solution here.

[Willy Tschager]

I haven't dug into your designs yet, but how do you deal with the fact that all ldraw 45 sloped parts are wrong?
Pic14-16 in that article show that the tip shoud have a height of 5 ldu, but all of our 45 slopes have only a 4 ldu height.
Moving the wall forward an lovering the top is maybe not the correct solution here.

I'm well aware that all slopes are wrong and that despite reality all LDraw have no seam. This was the most natual thing to do. Feel free to suggest a better solution. I could obviously move the wall back, which would move the height of 92946 back to normal and cover the gap in 15672 with a small quad by leaving the top edge of the slope where it is.

w.

[N. W. Perry]

This type of situation is a perfect example of why I'd suggest that true part measurements be added as a type of shadow (or meta-) data. This way the information could be added as situations arise, but would not require the re-working of whole categories of parts. An LDraw editor could then provide a user option to render parts using these more accurate dimensions, such as when building assemblies like those shown in the article.

Under such a system, these two parts would probably be identical except for this shadow data. But until then, it seems to me the above solution is good.

[Willy Tschager]

Here's the variant without the gap:


  15672a.dat (Size: 2.7 KB / Downloads: 3)

  92946a.dat (Size: 2.66 KB / Downloads: 3)

Please let me know which one I should submit.

w.

[Magnus Forsberg]

Please let me know which one I should submit.

w.

I prefer these, with the back wall in the right place. No "artificial" gap.