Enhancing JPEG, PNG, Text Using C# .NET Part 4 - Soft Edge Effect

The Soft Edges effect deflates non-transparent areas of the image by the specified amount, then applies the Gaussian blur effect to smooth the borders. The resulting image is used as a transparency mask for the original image. For example, you can see the result of applying the Soft Edges effect in MS Word:

Ready to Get Started? Download Document Solutions for Imaging Today! 

Let’s reproduce the same image in pure C# and save it as PNG.

The following code creates an image like this and saves it to image1.png:

using System.Drawing;
using System.Numerics;
using GrapeCity.Documents.Drawing;
using GrapeCity.Documents.Imaging;

using var bmp = new GcBitmap(600, 600, false);
using (var g = bmp.CreateGraphics(Color.White))
{
    g.Transform = Matrix3x2.CreateTranslation(100, 100);
    using var path = CreatePath(g, 400, 400);
    g.FillPath(path, Color.FromArgb(204, 51, 153));
    g.DrawPath(path, new Pen(Color.FromArgb(153, 153, 255), 28f));
}
bmp.SaveAsPng("image1.png");

static IPath CreatePath(GcGraphics g, float w, float h)
{
    float a = w * 0.5f;
    float b = h * 0.5f;
    var delta = Math.Min(w, h) * 0.1875f;
    float rx = a - delta;
    float ry = b - delta;
    rx *= rx;
    ry *= ry;
    float m = h / w;
    var ang = Math.Atan(m);
    float xb = a + (float)Math.Sin(ang) * delta * 0.5f;
    float yb = b - (float)Math.Cos(ang) * delta * 0.5f;
    float n = yb - m * xb;
    float c = n - b;
    float A = 1f / rx + m * m / ry;
    float B = (a / rx - m * c / ry) * 2;
    float C = a * a / rx + c * c / ry - 1;
    float D = (float)Math.Sqrt(B * B - A * C * 4);
    var x1 = (B + D) * 0.5f / A;
    var x2 = (B - D) * 0.5f / A;
    if (x2 < x1)
        (x2, x1) = (x1, x2);
    var y1 = m * x1 + n;
    var y2 = m * x2 + n;
    var x3 = w - x1;
    var x4 = w - x2;
    var y3 = h - y1;
    var y4 = h - y2;

    var path = g.CreatePath();
    path.AddEllipse(new RectangleF(0f, 0f, w, h), FigureEnd.Closed);
    path.SetFillMode(FillMode.Alternate);

    path.BeginFigure(new PointF(x1, y1));
    var aSeg = new ArcSegment()
    {
        Size = new SizeF(a - delta, b - delta),
        SweepDirection = SweepDirection.Clockwise,
        Point = new PointF(x2, y2)
    };
    path.AddArc(aSeg);
    path.EndFigure(FigureEnd.Closed);

    path.BeginFigure(new PointF(x3, y3));
    aSeg.Point = new PointF(x4, y4);
    path.AddArc(aSeg);
    path.EndFigure(FigureEnd.Closed);

    path.Close();
    return path;
}

As you can see, the inner and outer edges of the figure are sharp. Let’s make them smooth.

In the first stage, we draw the same figure on a transparent background:

using var bmp = new GcBitmap(600, 600, false);
using (var g = bmp.CreateGraphics(Color.Transparent))
{
    g.Transform = Matrix3x2.CreateTranslation(100, 100);
    using var path = CreatePath(g, 400, 400);
    g.FillPath(path, Color.FromArgb(204, 51, 153));
    g.DrawPath(path, new Pen(Color.FromArgb(153, 153, 255), 28f));
}

The image in GcBitmap now looks like this:

The next step is extracting the Alpha channel from the image to a separate GrayscaleBitmap:

using var gs = bmp.ToGrayscaleBitmap(ColorChannel.Alpha);

If we draw 0xFF values (opaque pixels) as white and 0x00 values (transparent pixels) as black, the GrayscaleBitmap (gs) looks like this:

The magic occurs when we call the GrayscaleBitmap.ApplyGlow method passing a negative value in the first parameter (infRadius). The second parameter specifies the Gaussian blur radius in pixels:

gs.ApplyGlow(-13, 16);

It works the other way around compared to the Glow effect - all non-transparent areas are deflated instead of being inflated; as for the Glow effect:

We can apply the resulting GrayscaleBitmap as a transparency mask to the source GcBitmap with transparent background:

bmp.ApplyTransparencyMask(gs);

Now the image in GcBitmap (bmp) looks like this:

The last step is converting the image background to opaque (the White color in this case). Also, we can save the resulting image to a file:

bmp.ConvertToOpaque(Color.White);
bmp.SaveAsPng("image2.png");

The image2.png file contains the expected result of our manipulations:

Applying Soft Edges can be slow if the deflation radius is large.

MESCIUS' .NET Imaging API Library

This article only scratches the surface of the full capabilities of Document Solutions for Imaging (DsImaging, previously GcImaging). Review our documentation to see the many available features, and our demos to see the features in action with downloadable sample projects.

Integrating this .NET imaging server-side API into a desktop or web-based application allows developers to programmatically create and manipulate images at scale and load, process, and save images across many different formats.

Ready to Get Started? Download Document Solutions for Imaging Today!