there was no showstopper, just the allocated time was running out.
it was the question of which implementation and what order/filter size should be used. the goal was to find an implementation that optimizes the enormous amount of texture lookups by leveraging the linear filter of the texture sampler.
there are also many filter parameters, this was also a question that needed some research. so if you find one you like, let us know, we can help you to evaluate the implementation.
integrating the filter should be quite easy, the code scaffolding is already there…