1920x1080 CMOS?
 

I was just reading this article:
Streamingmedia.com: Choosing a Camera for Online Video Production

And at one point it said: "HVR-Z5U is reportedly the first Sony HDV camera with sensors equipped to capture full 1920x1080 HD resolution."

Now I thought that the Z5 & Z7 had a lot of the same internals, which would mean that if the Z5 had 1920x1080 chips, the Z7 & S270 would have 1920x1080 chips.

Now I know all three of these cameras record HDV which is 1440x1080, but does anyone know if the Z7 has 1920 chips? I can't seem to find any info after a short search.

The dimensions of the chips are quite complicated, since the (square) pixels are rotated 45 degrees compared to normal layouts - so the pixel corners point horizontally and vertically. Count pixels corner to corner and the chip dimensions seem to be 960x540 (0.5megapixel) - but in reality there are two such interleaved arrays, so they are 1 megapixel in total. Thinking of a tiled floor with interleaving black and white tiles (500,000 of each colour tile!!) may help to visualise it!!

Effective maximum resolution is a bit more difficult to grasp because of the effective horizontal and vertical overlapping of rows/columns of pixels, but it's effectively halfway between 1920 and 960 horizontally (1440) and 1080 and 540 vertically (810).

Which is the clever thing about the design. They've concluded that 0.5megapixels are less than ideal, but 2 megapixels on a 1/3" chip makes each pixel just too small. Conventional 1 megapixel designs (such as 960x1080, as with the Z1) tend to be assymetrical in sharpness - this pattern enables a 1 megapixel chip to have equal horizontal and vertical resolution.

The article does state that the Z5 "uses three 1920x1080 CMOS sensors" which is incorrect. (It's the EX1/3 that does that.) It's not really possible to define it in a AxB form for the reasons above, but best to think of it as 1 megapixel, same number as the Z1.

There is also a difference between sensor sites and what the DSP creates as pixels. If you draw squares in the centre of the diagonal sensors and then draw squares comprising the remaining corners of adjacent sensors you will get 1920x1080 equally square created pixels. Which is what the DSP creates from the sensor data. It also creates, from this array, 1440x1080 non square pixels for HDV tape recording.

Ron Evans

Wow, thanks guys for the great information. I now know why that Z5 comment seemed fishy.

FWIW, the V1 uses the same pixel shift technology to produce 1920 x 1080 in the camera and through the HDMI output. And like it's newer bretheran it actually records in 1440 x 1080.