[materialpoint]

I thought the point of a sliding window, especially a fixed one, is to add the next element and subtract the oldest one from a running sum, to achieve O(1) running time except the start cost. This is what an efficient constant-time median filter in image processing would do, such as Perreault and Hébert's Median Filtering in Constant Time, and Weiss' Fast Median and Bilateral Filtering do.

Edit: I can't see the visualizer explaining this part.

[neumannbd]

Thanks for the feedback. I can't speak to the image processing references, as I'm still learning, but I will check them out and see if they make sense in the app.

Regarding the purpose of the sliding window, you are correct for commutative operations in fixed windows, like maintaining a sum where each operation is O(1). When you're talking about min/max, median, and other operations where adding or removing would not be constant time, as with a priority queue or heap.

Can you elaborate on which aspects you felt are missing from the visualizer?

[materialpoint]

This reply looks like classic and corteous AI when contradicted, and proves it haven't read nor understood the references, which clearly indicate how to achieve O(1) for min, max and median, even with greater dynamic range than 8-bit, and also clearly show how lacking the visualization is.