The group of settings at the top are used to configure the image, shader and render mode. The image setting lists the four images than can be used as the background. The shader dropdown configures which shader to use; a the first choice allows for user-entered kernels, while the remaining three are preconfigured edge detection filters. render mode controls how to diplay the image. The capture frame button will grab the current rendered image an use it for future render operations — this can be used to apply multiple convolutions in sequence.
The first bank of settings below the image control which kernel to use. A list of presets can be found in the kernel drop down. If the the split rgb option is selected, separate kernals can be specified for each RGB component. When checked the full kernel option will apply a full 81 element kernel for each pixel. When unchecked, the kernel will be rendered in two passes. First a vertical 9x1 kernel will be applied, followed by a horiztonal 1x9 kernel. Not all kernels can be expressed in the separable form. When the kernel preset custom is selected, custom kernel values can be typed into the input boxes.
The next bank of settings below the image control the output color for each pixel. After applying a convolution, the pixel color is scaled by factor in each of the RGB components. An RGB bias is then added on. The origin field controls where the convolution is applied, relative to the source pixel. scale is used to control the x and r area sampled by the convolution.