topic Re: Creating best matching plane using several(more than 3) points script in Fusion API and Scripts Forum

Creating best matching plane using several(more than 3) points script

caner-korkmaz — Wed, 05 Jul 2023 10:26:48 GMT

Hello,

I'm trying to create a plane using several points.Reason for this is i'm working on surveyers point cloud data and using random 3 points in a surface doesn't always creates the best plane because of the slight surface inperfections or bad readings.Instead what i want is selecting several points in a what i know is a surface and calculating best matching plane.I know resulting plane could not intersect all the point but thats okay.I am not good at python but i tried chatgpt and i think the code it gave me could work.But it gives an error.I think the highligted section in the below code is the problem.If anyone gives me a direction iwill be greatfull.Thank you.

selected_points = []
for point in design.activeProduct.activeSelections:
if isinstance(point.entity, adsk.fusion.SketchPoint):
selected_points.append(point.entity.worldGeometry)

import adsk.core, adsk.fusion def create_best_fit_plane(): # Get the current active document and root component app = adsk.core.Application.get() design = app.activeProduct root_comp = design.rootComponent # Get the selected sketch points selected_points = [] for point in design.activeProduct.activeSelections: if isinstance(point.entity, adsk.fusion.SketchPoint): selected_points.append(point.entity.worldGeometry) num_points = len(selected_points) if num_points < 3: # At least 3 points are required to create a plane return # Calculate the centroid of the selected points centroid = adsk.core.Point3D.create(0, 0, 0) for point in selected_points: centroid += point centroid /= num_points # Calculate the covariance matrix covariance = adsk.core.Matrix3D.create() for point in selected_points: x = point.x - centroid.x y = point.y - centroid.y z = point.z - centroid.z covariance.data[0][0] += x * x covariance.data[0][1] += x * y covariance.data[0][2] += x * z covariance.data[1][0] += y * x covariance.data[1][1] += y * y covariance.data[1][2] += y * z covariance.data[2][0] += z * x covariance.data[2][1] += z * y covariance.data[2][2] += z * z # Find the eigenvector corresponding to the smallest eigenvalue eigenvalues, eigenvectors = covariance.eigenVectors() min_eigenvalue = min(eigenvalues) min_eigenvector = eigenvectors[eigenvalues.index(min_eigenvalue)] # Create the plane through the centroid with the normal defined by the eigenvector plane_input = root_comp.constructionPlanes.createInput() plane_input.setByPlaneAndPoint( adsk.core.Plane.create(centroid, min_eigenvector), centroid ) plane = root_comp.constructionPlanes.add(plane_input) return plane # Run the function to create the best-fit plane create_best_fit_plane()

Betreff: Creating best matching plane using several(more than 3) points script

felix85SA3 — Wed, 05 Jul 2023 16:04:36 GMT

You can find the Object Reference Documentation at (Expand Objects) https://help.autodesk.com/view/fusion360/ENU/?guid=GUID-7B5A90C8-E94C-48DA-B16B-430729B734DC
You need to do at least some minimal research to get that script working. Have a look at the Code Samples as well: https://help.autodesk.com/view/fusion360/ENU/?guid=SampleList

Re: Creating best matching plane using several(more than 3) points script

BrianEkins — Wed, 05 Jul 2023 17:30:15 GMT

I think at the present time, ChatGPT isn't a dependable resource for creating Fusion programs. It will create code, but it will often have little errors that will be difficult for a new programmer to find and fix. It also seems that when it can't determine if Fusion has certain functionality, it just makes up functionality. For example, it calls the eigenVectors function on the Matrix3D object, but it doesn't support that function. It's also treating a Matrix3D object as a Python List, which it is not.

I suspect the algorithm it uses to find the points' average and the plane normal is valid, but it will take some work to convert it into usable code.

Re: Creating best matching plane using several(more than 3) points script

kandennti — Sun, 09 Jul 2023 13:52:05 GMT

Hi @caner-korkmaz -San.

I asked ChatGPT how to do planar fitting without external modules, but they replied that it is difficult.
Therefore, we created a sample using numpy.

# Fusion360API Python script import traceback import adsk.core as core import adsk.fusion as fusion import os import sys import inspect import random script_path = os.path.abspath(inspect.getfile(inspect.currentframe())) script_name = os.path.splitext(os.path.basename(script_path))[0] script_dir = os.path.dirname(script_path) sys.path.append(script_dir + "/site-packages") import numpy as np def run(context): ui: core.UserInterface = None try: app: core.Application = core.Application.get() ui = app.userInterface points = create_sample(200) dump_points(points, "sample data") create_fit_plane(points) ui.messageBox("Done") except: if ui: ui.messageBox('Failed:\n{}'.format(traceback.format_exc())) def create_fit_plane( pointList: list[core.Point3D] ) -> fusion.ConstructionPlane: def fit_plane(pointList: list[core.Point3D]): point_cloud = np.array(pointList) com = np.sum(point_cloud, axis=0) / len(point_cloud) q = point_cloud - com Q = np.dot(q.T, q) la, vectors = np.linalg.eig(Q) plane_v = vectors.T[np.argmin(la)] return plane_v, com # ************* normal, origin = fit_plane( [p.asArray() for p in pointList] ) app: core.Application = core.Application.get() des: fusion.Design = app.activeProduct root: fusion.Component = des.rootComponent plane: core.Plane = core.Plane.create( core.Point3D.create(*origin.tolist()), core.Vector3D.create(*normal.tolist()), ) baseFeat: fusion.BaseFeature = None if des.designType == fusion.DesignTypes.ParametricDesignType: baseFeat = root.features.baseFeatures.add() planes: fusion.ConstructionPlanes = root.constructionPlanes planeIpt: fusion.ConstructionPlaneInput = planes.createInput() constPlane: fusion.ConstructionPlane = None if baseFeat: try: baseFeat.startEdit() planeIpt.setByPlane(plane) constPlane = planes.add(planeIpt) except: pass finally: baseFeat.finishEdit() else: planeIpt.setByPlane(plane) constPlane = planes.add(planeIpt) constPlane.name = "Fit Plane" return constPlane def create_sample( count: int ) -> list[core.Point3D]: return [ core.Point3D.create( random.uniform(-10, 10), random.uniform(-10, 10), random.uniform(-1, 1), ) for _ in range(count) ] def dump_points( points: list, name: str = '', ) -> fusion.Sketch: app: core.Application = core.Application.get() des: fusion.Design = app.activeProduct root: fusion.Component = des.rootComponent skt: fusion.Sketch = root.sketches.add( root.xYConstructionPlane ) if len(name) > 0: skt.name = name sktPoints: fusion.SketchPoints = skt.sketchPoints skt.isComputeDeferred = True [sktPoints.add(p) for p in points] skt.isComputeDeferred = False return skt

200 random sketch points are created and the plane obtained by plane fitting is created.

The fit_plane function that does the plane fitting is a modified version of one I found on the web, so I do not understand the content itself.
I think it is probably a simple least squares method.

The attached file also includes numpy, so perhaps you can run it in your environment.

As a side note, in the CAD software I use for my work (CATIA V5), there is a command to create such a plane.
It would be useful to have a similar command in Fusion360.

Re: Creating best matching plane using several(more than 3) points script

karelQUS5R — Tue, 24 Sep 2024 13:51:56 GMT

i used the code above but nothing happens when I run the code...

dou you heve the same issue?