The 48MP sensors on these drones use what’s called a QuadBayer filter array, as opposed to the Traditional Bayer array. Because of this, it is able to combine four pixels into one using what is called pixel-binning. It uses this method to achieve 12MP photos even though the sensor itself has ~48 million physical photodiodes.
Lossless digital zoom, where you are fluidly zooming in from 1x to 2x, will never be possible with current sensor technology. The only way to implement “lossless” digital zoom is if the sensor has a QuadBayer Array and the “zoom” jumps from 1x (using 4-to-1 pixel binning) to 2x (using just the central ~4000x3000 photodiodes without pixel binning). However, this will still not be completely lossless because interpolating a transitional Bayer Array from a QuadBayer array using array conversion will create some slight color artifacts (unnoticeable to most) and more noise. But it will be almost lossless. This same method can be used for video with 48MP QuadBayer sensors, if the manufacturer implemented it, and if you are recording in 1080p. None of the DJI drones implement this type of digital zoom as far as I am aware.
On the other hand, one can shoot video in 4K and make their production resolution 1080p, which will give them the ability to fluidly zoom all the way to 2x in post (cropping) while not losing any detail because 4K cropped to 2x is 1080p.
On a side note, the QuadBayer color filter array in the
Mini 4 Pro is also how one is able to choose between 12MP and 48MP photo capture. When using 12MP photo mode, the 48MP sensor uses 4-to-1 pixel binning to create 12 million huge photodiodes by combining every four photodiodes (2x2) into one big photodiode. When changing into 48MP photo mode, it uses each photodiode individually, but because it has a QuadBayer array, it needs to use array conversion to convert the QuadBayer to a Traditional Bayer, so 48MP photos have slightly more noise and very very slightly worse color rendering, which most people won’t notice at all.
Shooting video uses only 12MP mode so it also uses 4-to-1 pixel binning and then crops the 4:3 aspect ratio to 16:9.
I am not sure how it shoots 1080p but it is most likely using over-sampling to turn the source 4K data from the sensor into 1080p footage.
View attachment 174183