As far as I recall, Super-Resolution only works on footage with a certain kind of aliasing which is sort-of where the detail "gets hidden". It also works best on footage without lots of motion, as that can mess up the temporal portions of the algorithm.
In most cases, you end up having to rely on image interpolators, and I have never seen any that do a good enough job at magnifications higher than, about 250%. The best one I've seen for video is an Avisynth one called nnedi3, but it is slow.And you still need to follow it up with some kind of detail-faking script.
You can also go nuts and try some glacial-speed single image interpolators like the ones on this page: