I would like to use NPP to build image pyramids (aka rsets or overviews) where each level is half the resolution of the preceding level. Simply using the Resize function will result in aliasing (due to Nyquist Sampling Criteria). IPP has an antialiased Resize API. Are there any plans to implement something similar in NPP? My current strategy will be to prefilter via a convolution kernel to suppress the higher frequencies prior to resizing, but I will have to determine what that kernel needs to be and this is probably not the most performant approach.