Given we can't even afford high res video wouldn't 3D be meaningless :)

It's not a 3D problem, it a lot of other things problem. Cost of teachers, creating environments that give automated re-enforcement. Making students actually learn / turn up.

Fiction often glosses over this bit. Ready Player One was really about kick ass AI / bandwidth needed to make the environments work. Computer games are already at graphic limits. It's just taking to many people to create their environments. There was an example article on HN this week. Can't find link.

It depends... How abstract is the picture you're going to transfer? The mid/low res video of the teacher would take a lot more bandwidth than a static texture + moving model. (or even partial static/dynamic texture for the body/face, as long as we don't hit the uncanny valley problem) Video with someone drawing a graph would take more b/w than the actual 3d model of the same graph.