Hi, I know about point spread filtering and filter kernels, nyquist, aliasing, and so on. There's a lot of text about these as well. However, there is one info I just cant figure out or find anywhere. I have an image, with dimensions w1 x h1. I want to scale this to w2 x h2. w2 and h2 can be arbitrary, no restrictions of any kind, aspect can be different, it doesnt have to be w1/2 h1/2, nothing. Now, *how* do I choose the appropiate filter kernel size? This is important so I filter everything above the right frequency... but the math for this procedure is never explained. Anybody has an idea?