go to pdf article 
print article 
2 | 3
HDV and the Z1 in context
What is the Z1E?
The Z1E and the FX1E are new 3 chip camcorders that utilize HDV format recording. Their form factor is based on prosumer DV camcorders, with fixed lens and flip out screen.
The FX1 was launched in late 2004. Z1E has XLR inputs and more camera controls than FX1 and records both 50i and 60i.
What is HDV?
A MPEG 2 recording format allowing recording of 720p and 1080i formats.
1080i format consists of 1440 x1080 pixels, sampled 8 bit 4:2:0 and then compressed 60:1 to tape. Data rate of 25mbs
720p format consists of 1280 x 720p sampled 8 bit 4:2:0 Data rate of 19mbs
The inter frame compression takes place across six frames.
Bear in mind that Z1 and HDV deck output analogue 4:2:2 HD 1920 x1080 signals (YPb Pr) by "upconverting" 1440 x1080 to 1920x1080.
The HDV standard was established by Sony, Canon, Sharp, JVC and others.
How can 25mbs compete with 185mb of HDCAM?
Pinnacle Systems suggest that a 50mbs MPEG inter frame compression comes close to 185mb HDCAM (which uses intra frame compression).
It is apparent that inter frame has advantages that are acceptable most of the time.
Also HDCAM records 1440 x1080, not 1920x1080.
Generally the picture becomes softer as you progress down the line of HDCAM SR to HDCAM to DVCPRO HD to HDV. So called I-frame compression is about 2.5 times less efficient than IBP but is easier to encode, decode and process. So an HDV image encoded at 25 Mb/s (IBP) is roughly equivalent to an I-frame image encoding of about 60 Mb/s.
So higher bit rate may not necessarily provide the "better" image quality.
In reality when the Z1 pans on a wideshot the image goes slightly soft. The MPEG processing runs out of bandwidth to record all of the detail from frame to frame. During a pan virtually every part of the picture changes from one frame to the next. When the camera stops panning MPEG builds up the resolution of the picture within a few frames.
The Sony FX1 and Z1 cameras rock!
The camera isn't bad it's the format that is the weakest link…..
The FX1 and Z1 cameras have three 1/3 inch 16x9 CCDs with 960 x1080 pixels.
That is 960 pixels across by 1080 lines. Or put another way 960 columns with 1080 pixels stacked in each column.
But the HDV format allows for 1440 x1080 pixels to be recorded. So what's up?
Sony increase the pixel count (apparently) by off-setting the green CCD 1/2 a pixel width to create a 1440 x 1080 Y output.
960 x1.5 equals 1440.
At least that's the most likely story until Sony spill the sushi...
The HDV recording standard of 4:2:0 sampling gives us 1440 x 1080 for Y and 720 x 580 for colours.
However since the colour is 'uprezzed' from the 1080 x 960 CCD the figures of 540 x 430 "native" colour should be considered.
After this 4:2:0 sampling the data is compressed to tape 8-bit with a compression ratio many times greater than DV. This is where the artifacts breed! As a comparison the HDW f900 has three 2/3 inch 1920x1080 pixel CCDs, HDCAM recording then sub samples to 1440 x 1080 pixels, colour samples 3:1:1 to create 1440 x 1080 for Y and 480 x1080 for colours then samples 8-bit and compress 4:1 to tape.
So when testing the Z1 camera it is vital to watch the replay, not just plug it into a monitor. Hooking the Z1 up to external HD recorders will have its place.
Does the Z1 do Progressive scan?
Technically speaking no. But from a viewers perspective yes.
The Z1 has a CF settings of 24, 25 and 30.
Sony is guarded as to how it achieves "24 frames per second" and is making a point of not saying it is not "24p".
There is much speculation as to how 24 frames per second is created from a interlace CCD. Including some that say it is created from a 60i scan in MPEG processing.
I can say that at 25CF motion strobing is very similar to f900 at 25psf and I would have no hesitation in using 25CF.
It appears that in 25CF mode each frame is created from the interlace scan by dropping one field and doubling the lines of the remaining field. This technique has been the mainstay in PAL countries to create a filmic look from video for over a decade.
Only now it appears that it can be achieved in camera rather than in post!
The result is a drop in vertical resolution, but less than you might expect.
