The smoothness issue you’re seeing on the sides and rear is actually a really common frustration with this workflow, thresholding alone isn’t designed to capture cortical surface detail well, especially the sulci and gyri. It’s not something you can fix just by adjusting the threshold values.
Getting the cortical folds to come out properly for 3D printing involves a few extra steps beyond what most tutorials cover, and the extension route you found has some compatibility nuances that trip people up.
I work with brain MRI segmentation and have done similar pipelines before. If you want I can take a look at what you have and point you in the right direction, feel free to DM me.