Argh! Pathtruder doesn't like me! :(

[Santeri Piippo]

I'm trying to finish the kolhii stick part 44811 but I just cannot get Pathtruder to generate the proper shape.

My dev files: https://dl.dropboxusercontent.com/u/6605...rcurve.zip
My commandline:

Code:

pathtruder -p1 p2.dat -p2 p1.dat -s1 s1.dat -s2 s2.dat -ts 1 -tn 4 -a out.dat

The shape is supposed to fill the area like a partially flattened torus cone. Can someone point out what did I do wrong here?

[Santeri Piippo]

Okay NVM for real this time, managed to get an acceptable result somehow by just running pathtruder with -s1 s1.dat -p1 p1.dat -p2 p2.dat

I don't know what happened.

[Magnus Forsberg]

Is that avatar from Chip's Challenge?

[Santeri Piippo]

Yeah. It's the teeth monster that chases you around.

[Philippe Hurbain]

Like your avatar, Pathtruder likes to bite (sometimes even its creator...).
Problem you have here is that your second shape file (s2) has line segments that are backwards compared to s1, so each shape section gets twisted while transitioning from s1 to s2. As stated in documentation, "Orientation of segments in shape file is important (a segment should end where the next starts): orientation of generated facets depends of that. MLCad "Swap" button (in line editing window) can help correcting that."...
For the record, path files gets sorted by Pathtruder on input, but shapes files are not, for better flexibility.
Attached the corrected s2 file, and result obtained from your command line (using -ts 10 instead of -ts 1 to get a smoother shape transition). Note that it would be probably simpler to generate only one quarter of the shape and use mirroring (assuming shaper is really symmetrical, I have not verified). Maybe resulting mesh resolution is a bit too high?

[Magnus Forsberg]

PathTruder is a nasty tool to use...

How do I create this result, using LETGUI?

Why is the gui in LETGUI not as good as other tools in LETGUI?
It doesn't help me understand the different settings I should use.

All the descriptions on the PathTruder page use the different commands. Like -ts, -a or -tn etc.
But LETGUI doesn't use those. It only have tick-boxes like "transition center" or "rotation angle"

And what settings are "legal" in which box? Could we have something like this: "Rotation angle (-r) 0-360[sup]o[/sup]"


I did however manage to get a similar result as Philo, but my surface is bfc-inverse. Did you flip it?
Or is that done by one of the settings I don't find?

[Max Martin Richter]

Is you swap-file BFC'ed in the right way?

A couple of months ago I used PT for generating a simplified UIC60 rail curve for a friend, who made a "custom" rail for the LEGO trains. And I can remember that I tried this and that until I understand what to do. :-)

/Max

[Philippe Hurbain]

Why is the gui in LETGUI not as good as other tools in LETGUI?
It doesn't help me understand the different settings I should use.

Agreed, help bubbles could be better...

I did however manage to get a similar result as Philo, but my surface is bfc-inverse. Did you flip it?

No I didn't... but have you noticed that in Santeri example p1.dat goes to path 2? Doing so, the shape file is explored in opposite direction, resulting in BFC inversion. I generally don't mind about BFC provided it's homogeneous, and invert it if needed when I finally get the right shape
Nonetheless, you get the same shape BFC inverted by swapping p1 and p2 and using the invert curvature (-i) option.

  pathtruder.JPG (Size: 55.41 KB / Downloads: 5)

[Michael Heidemann]

Please do not blame LETGUI, its the author of LETGUI that might not understood completely what to do to get good results (its me).
If you have wishes for help bubbles, please let me know, so I can integrate that in the next release.

[Willy Tschager]

I gave PathTruder a chance with my own paths, but I going nowhere. I attach the files I used and the end result it should look like. Any help is appreciated.

w.

[Philippe Hurbain]

Indeed, with default parameters, the shape you get is "inside out". You have 3 ways to correct that:
- swap p1 and p2
- swap the endpoint vertices of direction vector in files s3 and s4. Remember that the 1st enpoint of direction vector is placed on p1, the 2nd is placed on p2. Both methods above are thus completely equivalent
- Use the -i option (Invert curvature) but this will also invert BFW of resulting file. No big deal...

[Willy Tschager]

- swap the endpoint vertices of direction vector in files s3 and s4. Remember that the 1st enpoint of direction vector is placed on p1,

This and the -e option (plus adding the teils saved my day). thx,

w.

[Philippe Hurbain]

Indeed, -e is useful to get top and bottom perfectly horizontal. But generally I don't use this option too much now, I would just snap the shape to top and bottom (making them subparts) using unificator. Simpler to figure, and guaranteed rounding error free result.