Wow!
As for the question whether sensor size or pixel size matters ... Gimbal said it all.
I may even simplify the answer even more:
First, let me make sure that we all understand that the video output will be produced via downsampling. There is no center area cropped.
My question only was if the downsampling is via sub- or supersampling. Where I explained the terms in the second post.
Now, here we go (video = 1080p fixed size video):
- for Supersampling, the sensor size will determine noise in video.
- for Subsampling, the pixel size will determine noise in video!
Point is, while the dSLR sensor is much bigger, its pixels aren't, compared to a HD camcorder.
Originally posted by Philippos I agree that the entire surface of the sensor should (and I'm sure it will) be used - otherwise it's meaningless.
If someone knows how the burst mode in K20d works, I believe we will have our answer.
Well, just accidentily, I once posted the most detailed analysis of the K20D burst mode in this forum.
It is here:
https://www.pentaxforums.com/forums/pentax-dslr-discussion/21941-k20d-burst-m...esolution.html
The result is rather shocking: The K20D subsamples 768x1024 from the full sensor, i.e., it picks 3 colors from an array of 6x3 pixels. And it does it badly, producing unnecessary chromatic aberration artifacts.
So, in short: K20D uses subsampling
My earlier analysis is one reason why I am now critically questioning the quality of video to be expected from a dSLR...
Of course, subsampling is allright for LV.
Originally posted by nixcamic Even sub-sampling will still give better noise than your average camcorder
I computed it above: Only if camcorder uses a sensor SMALLER than 1/2.3" ...
Originally posted by Philippos Unfortunately, it is not only a matter of size, it is also a matter of good electronics.
Don't expect wonders.
Noise (at the pixel level) is the sum of read-out noise (the electronics) and Shot noise (from counting a discrete number of photons). The former is already rather low in modern cameras. The latter is only determined by two factors: pixel surface and quantum efficiency. The latter is already rather good as well (>50% in scientific devices). So, there is no other way than larger pixels. Sorry that physics is so unforgiving (and forget the black silicon exaggerations).
Last edited by falconeye; 04-30-2009 at 03:57 PM.