I am trying to slice a closed mesh and get the new surface area and volume.
HEMesh has a mesh.getArea() in the Ref_HEM_Soapfilm example, but the function does not exist in the library and I can’t find documentation for the HEMesh. Am I missing something?
Is there another library I should try for mesh splitting and analysis? I only see a way to split mesh by planes. Is there a way to split a mesh with another mesh?
Thanks for the help in advance. - J
Can you say more about what you mean? Like, given a sphere mesh and cube mesh, split the sphere into between one and seven parts, depending on how it intersects with the cube…?
After doing more research, I found CSG.
I found a java library (JCSG) that I tried to use for mesh boolean operations, but I didn’t have any luck.
This is correct. I want to take two solids and either remove one from the other (difference), add one to the other (union), or get the overlap of the two (intersect). These are the typical boolean operations that I have in most 3d modeling environments.
Hi @jNew jNew,
I’m afraid JCSG doesn’t work well with Processing 3.
I gave it a try a few months ago and came to the conclusion that the library is not compatible. If I remember correctly, you can do some boolean operations with the bundled primitives (Cube().toCSG, Sphere().toCSG, …) and convert them to PShapes (as kindly described by GeorgeProfenza here) but cannot do the opposite: converting a custom PShape to CSG format for transformation.
# Convert a CSG primitive to PShape (based on George Profenza example code)
def CSGToPShape(mesh, scl):
pshape = createShape(GROUP)
for p in mesh.getPolygons():
face = createShape()
face.beginShape()
for v in p.vertices:
face.vertex(v.pos.getX() * scl, v.pos.getY() * scl, v.pos.getZ() * scl)
face.endShape()
pshape.addChild(face)
return pshape
# !! Doesn't work: tries to convert a PShape to CSG format for boolean operations
def PShapeToCSG(shp):
polys = []
for ic in xrange(shp.getChildCount()):
c = shp.getChild(ic)
verts = []
for iv in xrange(c.getVertexCount()):
v = c.getVertex(iv)
verts.append(Vector3d(v.x, v.y, v.z))
polys.append(Polygon.fromPoints(verts))
return CSG.fromPolygons(polys)
Regarding Hemesh, Frederik Vanhoutte (the author) told me once that his library doesn’t support boolean operations.
Thank you solub, this is what I experienced as well. I also found that JCSG uses a newer version of Java that Processing3 does not support, so I couldn’t use newer releases.
Perhaps the pre-Processing 4 private release, which is Java 11 compatible, would work for those recent JCSG releases…
@jNew a couple of questions:
It’s probably unlikely but in case you work in 2D, the Geomerative library has a RPolygon class that allows boolean operations like union() or diff().
If you have 3D meshes and your workflow requires to stick to Processing, there is also the possibility to communicate with a third-party software directly from your sketch. From what I understand you are into Computational Design so you might be familiar with Maya or Grasshopper. If so, you could:
Here below an example sketch using the OscP5 library to communicate with Grasshopper through the gHowl component.
Please note that:
add_library('peasycam')
add_library('oscP5')
def setup():
perspective(60 * DEG_TO_RAD, width/float(height), 2, 6000)
size(1400, 800, P3D)
smooth(8)
global osc, shp3
#Instantiate the PeasyCam/OscP5 libraries
cam = PeasyCam(this, 80)
osc = OscP5(this, 12000)
#the two shapes to be used for Boolean Operation
shp1 = #YourFirstMesh
shp2 = #YourSecondMesh
#computes the 'difference' mesh between shp1 and shp2
mesh = transform([shp1, shp2])
#convert the resulting mesh to a PShape object
shp3 = meshToPShape(mesh)
def draw():
background('#FFFFFF')
shape(shp3)
def transform(shpList):
'''Sends the meshes vertices to Grasshopper via UDP using the OscP5 library & gHowl component.
The Grasshopper definition will then reconstruct the 2 meshes and perform the Boolean Operation of your choice (here "Mesh Difference").
The resulting mesh is then converted to an .obj file through a C# script and exported to the location of your choice.
Finally, Processing will import that mesh from the chosen location and return it for PShape conversion. '''
net1 = NetAddress("127.0.0.1", 12001) #remote location for the 1st mesh
net2 = NetAddress("127.0.0.2", 12002) #remote location for the 2nd mesh
for i, net in enumerate([net1, net2]):
message = OscMessage("/gHowlTest")
for child in shpList[i].getChildren():
for iv in xrange(child.getVertexCount()):
vert = child.getVertex(iv)
message.add(str(vert.x))
message.add(str(vert.y))
message.add(str(vert.z))
osc.send(message, net)
#Pausing the sketch for a moment because the '.obj' conversion in Grasshopper may take some time
delay(1000)
return loadShape("C:/Select/Your/Path/TheFileName.obj")
def meshToPShape(m):
'''Traverse the selected mesh and construct a PShape object from its faces/vertices'''
shp = createShape(GROUP)
for c in m.getChildren():
face = createShape()
face.beginShape(TRIANGLE)
face.fill(int(random(30, 50)), int(random(180, 205)), int(random(170, 190)))
for iv in xrange(c.getVertexCount()):
v = c.getVertex(iv)
face.vertex(v.x, v.y, v.z)
face.endShape()
shp.addChild(face)
return shp
Edit: Giving it a second thought, it would make more sense to send the location of your meshes (as OBJ) instead of transmitting their vertices one by one. This way we can overcome the UDP transmission limitation and have a cleaner workflow.
The example sketch below will:
\data directory with the HE_Mesh exporterdifference or union or intersection)PShape
__
import os
add_library('peasycam')
add_library('hemesh')
add_library('oscP5')
def setup():
perspective(60 * DEG_TO_RAD, width/float(height), 2, 6000)
size(1600, 1000, P3D)
smooth(8)
global shp3
cam = PeasyCam(this, 30)
osc = OscP5(this, 6001)
shp1 = #Mesh1
shp2 = #Mesh2
#Location of the future OBJ files
directory = sketchPath("data")
#Convert meshes to OBJ and place them in directory
toObj(shp1, "objName1", directory)
toObj(shp2, "objName2", directory)
#Select Boolean Operation of your choice + send instruction via UDP
operation = "difference" # -----> 'difference' or 'union' or 'intersection'
filename = "meshDifference"
osc.flush(directory + '\ ' + operation + ' ' + filename, "127.0.0.1", 6002)
#Wait for Rhino/Grasshopper to compute the transformation
delay(8000)
#Load output + convert to PShape (not mandatory)
shp = loadShape(sketchPath(filename + '.obj'))
shp3 = toPShape(shp)
def draw():
background('#FFFFFF')
shape(shp3)
def toObj(shp, filename, directory):
'''Traverse the selected mesh and construct a HE_Mesh from its faces.
The output will be saved as an OBJ in the chosen directory. '''
#Create 'data' directory if not existent
if not os.path.isdir(directory):
os.mkdir(directory)
#Detect face type (triangular vs quandragular faces)
type = 'triangular' if shp.getChild(0).getVertexCount() == 3 else 'quadrangular'
#Reconstruct mesh from faces
temp = []
for child in shp.getChildren():
face = []
for i in xrange(child.getVertexCount()):
v = child.getVertex(i)
face.append(WB_Point(v.x, v.y, v.z))
if type == 'triangular':
a, b, c = face
tri = WB_Triangle(a, b, c)
temp.append(tri)
else:
a, b, c, d = face
qua = WB_Quad(a, b, c, d)
temp.append(qua)
#Convert to HE_Mesh mesh format
if type == 'triangular':
mesh = HE_Mesh(HEC_FromTriangles().setTriangles(temp))
else:
mesh = HE_Mesh(HEC_FromQuads().setQuads(temp))
#Export to OBJ in 'data' directory
HET_Export.saveToOBJ(mesh, directory, filename)
def toPShape(s):
'''Traverse the selected mesh and construct a PShape object from its faces/vertices'''
shp = createShape(GROUP)
for c in s.getChildren():
face = createShape()
face.beginShape(TRIANGLE)
face.fill(int(random(30, 50)), int(random(180, 205)), int(random(170, 190)))
for iv in xrange(c.getVertexCount()):
v = c.getVertex(iv)
face.vertex(v.x, v.y, v.z)
face.endShape()
shp.addChild(face)
return shp
¯¯
On the Grasshopper side, as soon as the Boolean Operation instruction is received, the GhPython component will automatically:
difference or union or intersection)__
import os
import Rhino.Geometry as rg
directory, operation, filename = iMessage.split()
meshes = [rg.Mesh(), rg.Mesh()]
def toMesh(path, i):
with open(path) as f:
for l in f:
if l.startswith('v '):
x, y, z = map(float, l.split()[1:])
meshes[i].Vertices.Add(rg.Point3d(x, y, z))
elif l.startswith('f '):
face = [int(e.split('/')[0])-1 for e in l.split()[1:]]
if len(face) == 3:
meshes[i].Faces.AddFace(rg.MeshFace(face[0], face[1], face[2]))
else:
meshes[i].Faces.AddFace(rg.MeshFace(face[0], face[1], face[2], face[3]))
meshes[i].Normals.ComputeNormals()
meshes[i].Compact()
rg.Collections.MeshFaceList.ConvertQuadsToTriangles(meshes[i].Faces)
def toObj(mesh):
newDirectory = directory.split('data')[0]
print newDirectory
if not os.path.isdir(newDirectory):
os.mkdir(newDirectory)
filepath = newDirectory + "%s" % filename + ".obj"
with open(filepath, 'w') as f:
f.write("# OBJ file\n")
for vert in mesh.Vertices:
f.write('v ' + "%s %s %s\n" % tuple([vert.X, vert.Y, vert.Z]))
for face in mesh.Faces:
if face.IsTriangle:
ids = map(lambda x: str(x+1) + '//' + str(x+1), [face.A, face.B, face.C])
f.write('f ' + ("%s %s %s\n") % tuple(ids))
elif face.IsQuad:
ids = map(lambda x: str(x+1) + '//' + str(x+1), [face.A, face.B, face.C, face.D])
f.write('f ' + ("%s %s %s %s\n") % tuple(ids))
def run():
global oMesh
objs = [file for file in os.listdir(directory) if file.endswith('.obj')]
for i, obj in enumerate(objs):
path = directory + obj
toMesh(path, i)
if operation == 'difference':
oMesh = rg.Mesh.CreateBooleanDifference([meshes[0]], [meshes[1]])
elif operation == 'intersection':
oMesh = rg.Mesh.CreateBooleanIntersection([meshes[0]], [meshes[1]])
elif operation == 'union':
oMesh = rg.Mesh.CreateBooleanUnion(meshes)
toObj(oMesh[0])
run()
¯¯
Definition + Gh file
Below an example with 2 meshes (woman face + roman bust) converted to OBJ in Processing and retrieved by Grasshopper after UDP transmission of their locations. The instruction sent for boolean operation is: difference
The output loaded back in Processing:
I appreciate all the input, but I am trying to make a stand alone app. If I was making the app for myself, I would just develop it as a plugin for Rhino.
It looks like I will not be using processing for the development anymore… back to C++.
I am considering C++ because it has VTK. Upon further investigation, VTK actually has a java wrapper. I might try using this with processing. VTK has some really nice 3D tools.
I see. Don’t know if that’s relevant in your case but it seems the CGAL library also has a SWIG interface for Java.
