Hi everyone. It's been a long time (years) since I had messed with houdini at all. But all of a sudden i decided to give it another go.

So i just grabbed this wallpaper from my tablet and I was trying to recreate something like this in Houdini.



As I was messing with Houdini to create procedural track pieces based on an input curve I came up with two weird artifacts.

First one I believe to be a bug of sorts. I have clean (matching) geometry on both pieces and when I try to boolean subtract one from the other the boolean SOP is creating random points out of nothing that have nothing to do with neither input geometry.



On a second note maybe I got used to how other programs deal extrusion along a curve. But out of a profile I am trying to extrude a profile along a curve and when I use "curved curves" the extrusion orientation doesn't match the curve at all.
What am I doing wrong exactly? (Note: I'm using arclen to make sure the extrusion depth is the same as the curve length)




Would you advise me on ditching this approach and use a nurbs extrusion and only then converting my mesh to polygons or my approach is worthy of getting to accomplish my goal?

I'm attaching the Hip(NC) file so you can take a look.

P.S. Using the latest release build.

Hope you guys can shed some light on this lost soul! Also if you have any tips or comments regarding the project overall let me know.

A procedural thank you all!

Use Sweep SOP instead of PolyExtrude and resample the curve before feeding it into backbone curve input.

Also, you don't need to use boolean at all. Create a cross-section shape with curves (or however you like), convert it to polygons, and feed it into the second input of the Sweep SOP.

Using the sweep node is in fact a great tip! I am still using the boolean but just as a way to cut out the circle from the rectangle. The only issue is that if I use the resample on the path curve or if I convert the geometry after the sweep node then the end cap faces get unaligned. I even tried to convert the sweep shape into polygons and leave out the endcaps group out, but without the resample the result is uneven!

NOTE: THE PROBLEM THAT I AM MENTIONING ABOVE ONLY APPLIES TO THE CURVED CURVE (SITUATIONS WHERE THE RESAMPLED TWO FIRST POINTS OR THE TWO LAST POINTS ARE NOT ALIGNED, OBVIOUSLY)

After some tips from other people and reading my own comment, i ended up figuring the solution. Extending the curve on the last couple points creates a set of points with a tangent facing the appropriate direction. That solves the issue!