How to use the MR::cutMesh function correctly?

[czc98]

        std::vector<MR::MeshTriPoint> triPoints;
        for (size_t i = 0; i < loopPoints.size(); i += 3)
        {
            MR::Vector3f v3(loopPoints[i], loopPoints[i + 1], loopPoints[i + 2]);
            triPoints.push_back(MR::findProjection(v3, this->mesh).mtp);
        }

        // Makes closed continuous contour by mesh tri points, note that first and last meshTriPoint should not be same.
        tl::expected<MR::OneMeshContour, MR::PathError> result = MR::convertMeshTriPointsToClosedContour(this->mesh, triPoints, 
        MR::SearchPathSettings({MR::GeodesicPathApprox::DijkstraBiDir, 100}));
        std::vector<unsigned int> edgeIds;
        if (result.has_value())
        {
            std::vector<MR::OneMeshContour> contours;
            contours.push_back(std::move(result.value()));
            std::vector<MR::EdgePath> edgePaths = MR::cutMesh(this->mesh, contours).resultCut;
            MR::EdgePath edgePath = edgePaths[0];
            for (MR::EdgeId& edgeId : edgePath)
            {
                edgeIds.push_back(edgeId.get());
            }
        }

I find MR::cutMesh function in documentation, and i try to use it, but i didn't get the desired effect.
I made a simple attempt to take three points on the mesh surface as Meshtripoints, and print the number of points and faces before and after execution that has changed.
I used the following code to extract the information from the points and surfaces to reconstruct the Mesh, but it did not change. After executing MR::cutMesh function, do I need to execute another function to achieve the clipping effect?

        std::vector<float> points;
        MR::VertCoords &_points = this->mesh.points;
        for (auto &point : _points)
        {
            points.push_back(point.x);
            points.push_back(point.y);
            points.push_back(point.z);
        }
        std::vector<unsigned int> faces;
        MR::MeshTopology &topology = this->mesh.topology;
        MR::Triangulation tris = topology.getTriangulation();
        for (auto &face : tris)
        {
            faces.push_back(face[0]);
            faces.push_back(face[1]);
            faces.push_back(face[2]);
        }

[Grantim]

Hello! The code looks valid, could you please share data (mesh and loopPoints) so we could investigate the issue

[Grantim]

Ok, I see:
please have a look at these two issues:
#1374
#1512

auto rescut = MR::cutMesh(this->mesh, contours).resultCut;
auto onePart = MR::fillContourLeft( this->mesh.topology, rescut );
auto otherPart = this->mesh.topology.getValidFaces() - onePart;

//then you can do this
Mesh onePartMesh, otherPartMesh;
onePartMesh.addPartByMask( this->mesh, onePart );
otherPartMesh.addPartByMask( this->mesh, otherPart );

[czc98]

Ok, I see: please have a look at these two issues: #1374 #1512

auto rescut = MR::cutMesh(this->mesh, contours).resultCut;
auto onePart = MR::fillContourLeft( this->mesh.topology, rescut );
auto otherPart = this->mesh.topology.getValidFaces() - onePart;

//then you can do this
Mesh onePartMesh, otherPartMesh;
onePartMesh.addPartByMask( this->mesh, onePart );
otherPartMesh.addPartByMask( this->mesh, otherPart );

Thanks！I have found that the direction of the contour affects the result of MR::fillContourLeft. (Clockwise or anticlockwise)
Is there a way to fix the MR::fillContourLeft function to the part inside the contour?

[czc98]

I want to modify the mesh in place, so I use the deleteFaces function, which still retains a lot of information about points and faces, but makes it no valid.
When I use the following code to extract the information of points and faces, there will be a lot of redundant data, may I ask if there is an efficient way to eliminate the redundant information?

std::vector<float> points;
MR::VertCoords &_points = this->mesh.points;
for (auto &point : _points)
{
    points.push_back(point.x);
    points.push_back(point.y);
    points.push_back(point.z);
}
std::vector<unsigned int> faces;
MR::MeshTopology &topology = this->mesh.topology;
MR::Triangulation tris = topology.getTriangulation();
for (auto &face : tris)
{
    faces.push_back(face[0]);
    faces.push_back(face[1]);
    faces.push_back(face[2]);
}

[Grantim]

You are right fillContourLeft depends on contour orientation. You can distinguish parts by area for example:

auto rescut = MR::cutMesh(this->mesh, contours).resultCut;
auto onePart = MR::fillContourLeft( this->mesh.topology, rescut );
auto otherPart = this->mesh.topology.getValidFaces() - onePart;

auto& smallerPart =  this->mesh.area( onePart ) < this->mesh.area( otherPart ) ? onePart : otherPart;
this->mesh.deleteFaces( smallerPart );

// eliminate the redundant information
this->mesh.pack();