Still vs Video mode of DSLRs

[Theodebernacius]

In terms of pixel numbers, DSLRs can capture much larger pictures in still than in video. I was wondering how the sensor was used in both case. I can think of 2 possibilities.

1) In video mode the part of the sensor used is a smaller rectangle inside the whole sensor.

2) In video mode the whole width and height of the sensor are used but some kind of subsampling is taking place. For instance, several pixels could be averaged into one or only one out of several pixels could be read.

In the 1st case, the video image is like a cropped version of the corresponding still. In the 2nd case, still and video cover exactly the same space but the video image should lose in sharpness.

 

[Cracker Funk]

Line-skipping.

This article describes it nicely (as well as a couple other short-comings of DSLR):

http://ninofilm.net/blog/2010/04/17/536/

 

[chilipie]

The latter is closest. Canon DSLRs use a technique called "line-skipping" to save on processing power - every third horizontal line of the sensor is read, one after another (this is called a rolling shutter), and the image is then resized. This technique is why moiré and aliasing can be such a problem. Some DSLRs (like the 600D/T3i) can also use your first option, which is effectively a digital zoom without the usual loss in quality.

 

[Theodebernacius]

Very interesting, thanks for the replies!