Here's what I'm trying to achieve. I've exported some road data from QGIS, which is in the form of road "segments" (sometimes an entire length of road from start to finish, sometimes just a portion of the road). I want to use this data to create a single (or multiple individual) mesh/geometry object for this road network that fits with my exported DEM data.
This is what I've got so far
An example of the pertinent data is here (other fields in the csv exist and some have potential for later use.
"LINESTRING Z (421333.716 465914.083 272.2,421338 465914 272.2,421363 465915 272.4,421371 465915 272.4,421435 465925 272.8,421464 465929 273,421482.398 465931.243 273.1,421504.054 465929.861 273,421694 465913 271.8,421729 465910 271.2,421760 465906 269.9,421929.519 465861.351 263.8,422000.214 465841.341 261.9,422137 465803 264.7)"The number of coordinates in the row can vary between 2 and many. I've got a Table Import node set to look at the file, and then assign the values to attributes. Next I have a python node which has the following code in
import hou import re # Get the current node and geometry node = hou.pwd() geo = node.geometry() # Define the origin for coordinate adjustment origin_x = 424898.693 origin_y = 467956.814 # Function to parse and adjust LINESTRING data def parse_linestring(linestring): # Define regex pattern to extract the LINESTRING data linestring_pattern = re.compile(r'LINESTRING Z \((.*?)\)') match = linestring_pattern.search(linestring) if match: points_str = match.group(1) # Split points into x, y, z coordinates points = points_str.split(',') adjusted_points = [] for point in points: x, y, z = map(float, point.split()) # Adjust coordinates based on the origin adjusted_x = x - origin_x adjusted_y = y - origin_y adjusted_points.append((adjusted_x, adjusted_y, 200)) return adjusted_points return [] # Get a list of all point attributes point_attribs = geo.pointAttribs() # Iterate over each point in the geometry for point in geo.points(): # Extract the LINESTRING data from the first column (assuming it's stored as an attribute) linestring = point.attribValue("WKT") # Extract and parse the LINESTRING data points = parse_linestring(linestring) # Create a new curve in Houdini if points are found if points: # curve = geo.createNURBSCurve(len(points),is_closed=False, order=2) # Create an empty list of point indices for the curve i = 0 for x, y, z in points: #print(z) curve.vertices()[i].point().setPosition((x, y, z)) curve.vertices()[i].point().setAttribValue("roadName", point.attribValue("roadName")) curve.vertices()[i].point().setAttribValue("roadClassificationNumber", point.attribValue("roadClassificationNumber")) curve.vertices()[i].point().setAttribValue("minimumWidth", point.attribValue("minimumWidth")) curve.vertices()[i].point().setAttribValue("averageWidth", point.attribValue("averageWidth")) curve.vertices()[i].point().setAttribValue("formOfWay", point.attribValue("formOfWay")) i = i + 1
Image Not Found
This is creating points and geometry, I can see that things are where they should be. (I am creating them all at 200 rather than the Z component from the file because I was trying to see if I could get a simple mesh to extrude using polyexpand2d node. This is later on though.)
I then fuse to make sure that the various points actually are not being duplicated and that the junctions are fully formed. (When I look at this with the point number and primitive number in the scene though I can see that the number of primitives corresponds to the original number of rows in the csv. I'd like where possible to merge the individually named roads together to reduce the count, and to make visualisation of what's going on easier. I've tried various methods of fuse/merge inside foreach groups, and nothing has merged primitives together at all.)
I'm then looking at the geometry spreadsheet and I notice that point 0 is still in the original csv format, and not one of the points created by my python so I use some VEX to delete it in the wrangle node ( Why is that?)
To cut a long story short, I tried using Sweep with a flattened cube to create a roadbed, which didn't work very well, even with vector UP assignment there were still stretches which refused to orient correctly. I've also tried the polyexpand2d route, that fails to cook. If I reduce the number of roads to just 1 then it will work, but clearly that's not really very helpful at all. What is the best way to approach this? Should I be trying to create 1 road at a time? I've watched countless YT videos and read lots of posts, but I've not really found a usecase that seems similar to mine yet to figure out what is the best approach.
