Novel view synthesis from an in-the-wild video is difficult due to challenges
like scene dynamics and lack of parallax. While existing methods have shown
promising results with implicit neural radiance fields, they are slow to train
and render. This paper revisits explicit video repres