This image illustrates the crop modes of the Panasonic GH4, which uses a very similar 16MP MFT sensor:
The sensor areas:
- Black (4608 x 3486) is the full 16MP sensor, used only for photos. This is a 2.0x crop.
- Magenta (4608 x 2592) is the full width of the sensor, but the top and bottom bands are not used, in order to get the 16x9 HD image ratio. This is used for 1920 x 1080 (HD) video. This is a 2.0x crop.
- Blue (4096 x 2160, or 3840 x 2160) is the part of the sensor used for 4K video. The crop mode is either 2.2x crop or 2.3x crop.
The question is always: "Why don't they use the full sensor width for 4K video?" The answer: because when they use the blue portion, each pixel in the resulting image is derived from one and only one photosite on the sensor, and this increases image quality.
The follow-up question: "But for 1080 video, they sample more pixels and combine them; why wouldn't they do the same for 4K?" The answer: for 1080 video, each pixel in the image is produced by combining 2.4 pixels by 2.4 pixels. In other words, each pixel in the final image is derived from about 6 photosites on the sensor. But in 4K mode, there just aren't enough pixels to do this; it could only use 1.2 x 1.2 pixels. There isn't enough data to benefit from combining, so they don't, and the resulting image is superior.
The natural next question: "Why don't they use a sensor that has twice the number of pixels as 4K video requires, and oversample those pixels to combine them into a 4K video image?" The answer; that's a 36 megapixel sensor, which would be much much much more expensive, and would require much much much faster hardware to decode. So wait about 5 years, and that's what we'll all be flying.