Video texture LOD - questions on possible approaches

I’m trying to implement a version of a video texture LOD - with a mesh with that video texture as a material switching to higher resolution of the video when a user gets closer to it.

If I change the underlying HTMLVideoElement width and height properties as the user gets closer, will this ultimately affect the amount of pixels that are drawn to the texture?

What about having a few different hidden html video elements, each with a source loaded from a video file with a specific resolution, and pausing/resuming the corresponding video and loading that into the texture. Has someone tried this approach?

Any other suggestions?