Offset with holes

[sethhillbrand]

Hi Angus and others

tldr; the ClipperOffset appears to behave differently in Clipper2 vs Clipper1. Is this intentional and is there a way to simulate the Clipper1 behavior?

In Clipper1, if we called ClipperOffset with a negative value, the outline got smaller while the holes got larger

In Clipper2, calling ClipperOffset with a negative value, both the outline AND the holes get smaller

The code in Clipper1 is

ClipperOffset c;
PolyTree solution;

for( const POLYGON& poly : m_polys )
{
    for( size_t i = 0; i < poly.size(); i++ )
    {
        c.AddPath( poly[i].convertToClipper( i == 0, zValues, arcBuffer ),
                   jtSquare, etClosedPolygon );
    }
}

double coeff = 1.0 - cos( M_PI / aCircleSegCount );
c.ArcTolerance = std::abs( aAmount ) * coeff;
c.MiterLimit = 2.0;
c.MiterFallback = miterFallback;
c.Execute( solution, aAmount );

In Clipper2, the code is:

ClipperOffset c;

for( const POLYGON& poly : m_polys )
{
    for( size_t i = 0; i < poly.size(); i++ )
    {
        c.AddPath( poly[i].convertToClipper2( i == 0, zValues, arcBuffer ),
                   JoinType::Square, EndType::Polygon );
    }
}

double coeff = 1.0 - cos( M_PI / aCircleSegCount );
c.ArcTolerance = std::abs( aAmount ) * coeff;
c.MiterLimit = 2.0;
c.MiterFallback = miterFallback;
Paths64 solution = c.Execute( aAmount );

[sethhillbrand]

I'm going to answer my own question here. Just took some searching.

In Clipper2, we can't be as loose with adding holes. They need to be grouped with a specific path. So, changing the Clipper2 code to the following seems to work.

for( const POLYGON& poly : m_polys )
{
    Paths64 paths;

    for( size_t i = 0; i < poly.size(); i++ )
        paths.push_back( poly[i].convertToClipper2( i == 0, zValues, arcBuffer ) );

    c.AddPaths( paths, joinType, EndType::Polygon );
}

[Rafa350]

Yes, run ok in C# too.