Skip to main content

Convolution & Mask Processing

Convolution and mask processing shaders in KraftShade enable sophisticated image processing operations like sharpening, embossing, and morphological transformations.

Overview​

Convolution is a fundamental image processing technique that applies a mathematical operation to each pixel based on its surrounding pixels. This technique is the foundation for many effects including blurring, sharpening, edge detection, and embossing.

KraftShade provides several convolution-based shaders that operate on a 3x3 grid of pixels surrounding each pixel in the image. These shaders can be used to create a wide range of effects by simply changing the convolution matrix.

Convolution Base Classes​

Sample3x3KraftShader​

The base class for shaders that sample a 3x3 grid of pixels around each pixel in the input texture. This class handles the sampling logic but doesn't define how the samples are processed.

Convolution3x3KraftShader​

A specialized shader that applies a 3x3 convolution matrix to the image. This is the base class for many convolution-based effects.

Parameters:

  • convolution: GlMat3 representing the 3x3 convolution matrix

[IMAGE: Convolution3x3KraftShader - Left side shows original image, right side shows image with custom convolution matrix applied]

Convolution3x3WithColorOffsetKraftShader​

An extension of the basic convolution shader that adds a color offset to the result, allowing for more sophisticated effects.

Parameters:

  • convolutionMatrix: GlMat3 representing the 3x3 convolution matrix
  • colorOffset: FloatArray (4 elements) representing RGBA color offsets

[IMAGE: Convolution3x3WithColorOffsetKraftShader - Left side shows original image, right side shows image with convolution and color offset applied]

Image Enhancement Shaders​

SharpenKraftShader​

Enhances the details in an image by increasing the contrast between adjacent pixels.

Parameters:

  • imageWidthFactor: Float value related to the image width (default: 0f)
  • imageHeightFactor: Float value related to the image height (default: 0f)
  • sharpness: Float value controlling the strength of the sharpening effect (default: 0f)

Example Use Cases:

  • Enhancing details in photographs
  • Improving text readability
  • Compensating for slight blur in images

[IMAGE: SharpenKraftShader - Left side shows original image, right side shows sharpened image]

EmbossKraftShader​

Creates an embossed effect that makes the image appear raised or stamped.

Parameters:

  • intensity: Float value controlling the strength of the emboss effect (default: 1f)

Example Use Cases:

  • Creating metallic or stamped effects
  • Adding texture to flat surfaces
  • Creating relief-like images

[IMAGE: EmbossKraftShader - Left side shows original image, right side shows embossed image]

Laplacian Filters​

LaplacianKraftShader​

Applies a Laplacian filter to detect edges in all directions. This shader extends Convolution3x3WithColorOffsetKraftShader with a specific convolution matrix.

Example Use Cases:

  • Edge detection
  • Feature extraction
  • Image analysis

[IMAGE: LaplacianKraftShader - Left side shows original image, right side shows image with Laplacian filter applied]

LaplacianMagnitudeKraftShader​

Calculates the magnitude of the Laplacian to detect edges regardless of direction.

Example Use Cases:

  • More robust edge detection
  • Finding boundaries in images
  • Preprocessing for computer vision algorithms

[IMAGE: LaplacianMagnitudeKraftShader - Left side shows original image, right side shows image with Laplacian magnitude filter applied]

Morphological Operations​

Morphological operations are a set of non-linear operations related to the shape or morphology of features in an image.

DilationKraftShader​

Expands bright regions and shrinks dark regions in an image. This is useful for filling in small holes or gaps.

Parameters:

  • radius: Int value controlling the size of the dilation operation (default: 1)

Example Use Cases:

  • Filling in small holes or gaps
  • Expanding features in binary images
  • Connecting nearby objects

[IMAGE: DilationKraftShader - Left side shows original image, right side shows image after dilation]

ErosionKraftShader​

Shrinks bright regions and expands dark regions in an image. This is useful for removing small objects or thin lines.

Parameters:

  • radius: Int value controlling the size of the erosion operation (default: 1)

Example Use Cases:

  • Removing small objects or noise
  • Shrinking features in binary images
  • Separating connected objects

[IMAGE: ErosionKraftShader - Left side shows original image, right side shows image after erosion]

Common Convolution Matrices​

Here are some common convolution matrices that you can use with the Convolution3x3KraftShader:

Box Blur​

1/9, 1/9, 1/9
1/9, 1/9, 1/9
1/9, 1/9, 1/9

Gaussian Blur​

1/16, 2/16, 1/16
2/16, 4/16, 2/16
1/16, 2/16, 1/16

Sharpen​

 0, -1,  0
-1,  5, -1
 0, -1,  0

Edge Detection​

-1, -1, -1
-1,  8, -1
-1, -1, -1

Emboss​

-2, -1,  0
-1,  1,  1
 0,  1,  2

Using Convolution Shaders in Pipelines​

Convolution shaders can be combined with other effects to create sophisticated image processing pipelines. They are particularly effective when used in sequence with color adjustments or blending operations.

Performance Considerations​

When working with convolution operations:

  • Convolution shaders sample multiple pixels for each output pixel, which can be computationally expensive
  • Larger convolution kernels (beyond 3x3) are significantly more expensive
  • For large blur operations, consider using a two-pass approach (horizontal then vertical) instead of a single large convolution
  • Morphological operations like dilation and erosion can be particularly expensive for large radii

Advanced Techniques​

Combining Morphological Operations​

Dilation and erosion can be combined to create more complex morphological operations:

  • Opening (erosion followed by dilation): Removes small objects while preserving the shape and size of larger objects
  • Closing (dilation followed by erosion): Fills in small holes and gaps while preserving the shape and size of objects

Multi-Pass Convolution​

For more sophisticated effects, you can apply multiple convolution passes in sequence to achieve complex transformations that wouldn't be possible with a single pass.