i think the original one is on a curved surface, so i just did it on a curved surface and recorded all the steps i did as a JI https://gfycat.com/LinearMadKoodoo
"...my low poly topology is atrocious. I've tried redoing it a few times, but I always end up with a similar result." hmm...well to start with there's a whole heap of redundant vertices or edges that are not contributing too the overall object's silhouette/shape. EDIT: So firstly I'd advise desolving those continuous…
I am serious but you know i wanted to avoid this 3 edges goin for all the lower surface, coz this all is bend in letter C, when i got 3 edges goin threw all the surface it gives me an hard edge everywhere. ok never mind just separated the outer element from the lower surface.
well you could calculate the limit surface of the catmull-clark subd realtively easy, either recursively, inputing the amount of iterations you will do, or non recursively to get a theorethical limit surface.