Hey I know its dumb but do I unwrap this how I have or have each face as
a seperate thing? I only mention because someone said to have uvs stop
at hard edges, so wouldnt that mean I have loads of single faces, which Ive also been told to avoid. The highpoly has beveled edges btw.

Replies
Depending on the mesh, if each face had its own UV shell, it would foremost be a waste of texture space due to the combined amount of padding necessary. Iirc correctly, more UV shells effectively increase vertex count when rendering - potentially increasing performance cost - again depending on the specific mesh. If you really want to know, best read up on rendering.
I think a reasonable approach here is to use hard edges where they are beneficial (steep angles) and split corresponding UVs. But of course, when working on a project, follow its guidelines.
To our dear friend Jake Gyllenhaal : what about trying things out ?
One important thing to note: splitting UVs on hard edges is only really necessary if you’re baking normal maps.
IMHO this is a case of taking an assumption (which isn't necessarily wrong of course) and running with it without really putting it in context.
We've all seen polygon density affecting performance : importing a sculpt into a regular 3d program can makes it crawl ; and upgrading a video card allows to handle scenes with more complexity. And similarly, models get reduced for LODs or for games to be ported to low-end hardware.
But the above is orders of magnitude different from adding a few hard edges and their matching UV splits. Sure enough this is increasing the vertex count - but not to a point that it makes it even a factor for performance. Or if it was, I'd love to see an example of that ... but that would require a scene already pushed to the limit geometry-wise ... which is not viable in production anyways.