The cinematic mode is done by recording a depth map (calculated using the ISP and Neural Engine?) alongside the regular video. On older phones you could do this

The cinematic mode is done by recording a depth map (calculated using the ISP and Neural Engine?) alongside the regular video. On older phones you could do this using the SDK, but iirc it was a much lower resolution depth map (like 480p or less?). Given that they've also added ProRes encode/decode, it's likely fair to assume that they've redone the entire codec hardware. So yeah, this is my long-winded way of agreeing with you that it is indeed more efficient.