In this video, the kernel for a low pass filter is given to be $[\frac{1}{2},\ \frac{1}{2}]$, while the kernel for a high pass filter is given to be $[-1,\ 1]$.
Interpreting these as essentially vectors, I find it a bit strange that they have different magnitudes. I would've guessed that regardless of scaling, flipping the sign of one of the components is what makes the kernels fundamentally different, just as would be the case for the task of selecting two vectors which (each taken independently) span the diagonals of $\mathbb{R}^2$.
Why is this? Does scaling the kernel of a filter matter, or is it only the span of the kernel (in vector lingo) which effects the qualitative outcome of the operation?