Convolve Node¶

The Convolve node performs a 2D convolution operation on an image using a custom Image Kernel. This allows for a wide range of effects such as blurring, sharpening, and edge detection, depending on the kernel values provided.

The node multiplies a neighborhood of pixels around each image pixel by the kernel values and sums the results to compute the output pixel color. This process is often used in image processing to emphasize or suppress certain spatial features.

Inputs¶

Image: The image to be processed.
Kernel: The convolution kernel used to modify the image. This can be provided as a constant value or an input image where pixel values define the kernel weights.
Normalize Kernel: A factor that determines whether the kernel values should be automatically normalized (i.e., divided by their total sum) to maintain consistent brightness.

Properties¶

Kernel Data Type

Determines how the kernel is defined:

Float:: A numeric kernel defined by the Kernel input socket.
Image:: A full image used as the kernel, allowing for more complex filtering patterns.

Outputs¶

Image: The resulting image after applying the convolution operation.

Usage¶

The Convolve node can be used for a variety of image processing effects:

Blurring – Use a kernel with evenly weighted positive values.
Sharpening – Use a kernel with a high central value and negative surrounding values.
Edge Detection – Use kernels such as Sobel or Laplacian to enhance image edges.

Tip

When designing custom kernels, ensure the kernel size and values are balanced to avoid excessive brightening or darkening. Enabling Normalize Kernel is often recommended to keep overall image brightness consistent.