One major difference is that training/optimization for the other domains is on a large training set, while here it is per scene.
That makes training time more important for radiance fields in my opinion. But the trend seems to be in the opposite direction?

A nice video from Veritasium on the topic. People are just prone to clickbait, but if it helps to educate more people it is helpful to use it :)

youtu.be/S2xHZPH5Sng?...