Input latency for one, because the next frame is delayed where the interpolated frame is inserted.

And image quality. The generated frame is, as I said, interpolated. Whether that’s just using an algorithm or machine learning, it’s not even close to being accurate (at this point in time).