I don't think that's needed. I think the whole 4k 16:9 area of the sensor is continuously read at 60fps when recording 1080/60i.
The evidence I have for that is some info from the manual. When recording video, you can also take pictures (without disturbing the video). In 1080/60i, those pictures will become 3840x2160. In 1080/30p, they will become 1920x1080, and in 720/60p. they become 1280x720. This suggests that: in 1080p, sensor is read with 2x2binning. In 720p, sensor is read with 3x3 binning. In 1080i, the whole 3840x2160 area is read.
Manual also says:
"1,920 × 1,080/60i (59.94 fields/s *)
...
* Sensor output is about 60 fps"
The benefit of using binning, is that the rolling shutter is faster. The downside is a softer image. In 1080i, every pixel of each line, is likely made of "super pixels" i.e a pixel in the image is made from a square of 4 pixels from the bayer pattern, which means it contains all three colors without using data from its "neighbors" (which makes it sharp). This should make 1080i sharper than 1080p, if my theory is right.
I also looked in the manual of Nikon 1 V2/J3. They have a new sensor, with a resolution of 4608x3072. When recording video, you can take stills of the full resolution. This suggests that the whole sensor is continuously read, and cropped/resized in software for the video. I think that this should produce really nice 1080p. (If it's not all destroyed by the compression.)
V1 probably used an early version of this sensor:
http://www.aptina.com/products/image_sensors/ar1011hs/
V2 may use this exact sensor, or an early version of it.
http://www.aptina.com/products/image_sensors/ar1411hs/
These sensors look so great. I hope you can get them in an affordable, and raw recording, camera soon.
Last edited: