This video describes how can we manipulate the number of channels in the output by having different number of filters for convolution.