Would recursively adding more t's give you more control to how you can interpolate your curve? it seems also that adding more t's would force the curve to be flatter. Is there a way to avoid making the curve flat but still recurse with more t's?
I think adding more ts will not make the curve flatter but smoother. Every t will produce a point x(t), and only if we have enough x(t) points, we can assume they form a curve.
Sainan is correct. The more samples of t, the closer our curve approximates the true curve.
This seems quite computationally expensive (if you start with many points as opposed to just four) and it's not parallelizable like we discussed in previous lectures. Is calculating these curves usually a significant factor in evaluating performance?