Re: Tim Kolbs article on HD formats

[pom_boarder] "This has confused me somewhat, how can DVCPROHD be compressed 7-1 and be 40Mbs yet HDV is compressed 35-1 and be only slightly less. That would mean that HDV carries a lot more information yet we know that isn't the case."

Since I'm answering in the thread...I look at compression ratios a bit differently than manufacturers do. Compression ratios are usually quoted as how compressed the data set is AFTER 'bit rate reduction', or as Sony euphemistically states "prefiltering."

When we refer to these resolution reductions and color subsampling, those are done BEFORE any actual "compression" takes place where there is a routine run to eliminate redundancies to optimize file size. So while perhaps 1280x720p24 4:4:4 may not compute to be 7x 100 Mbits/s, 960x720p24 4:2:2 probably does...in Sony HDcam's case it's a case of uncompressed being 1920x1080Psf 23.97 4:4:4 vs what is left when the footage hits the actual compression engine after "prefiltering", which is 1440x1080PsF 23.97 3:1:1. Sony also quotes HDcam as being 7:1 compression even though uncompressed, even at 4:2:2 works out to be something on the order of 1.2 Gbits/s, making 135 Mbits/s closer to 10:1 if the comparison is done this way.

35:1 for HDV works out roughly assuming that uncompressed is something less than 900 Mbits/s, which is probably on target as HDV's native frame size is 1440x1080, not 1920x1080, I guess that works, but then for HDcam to truly be 7:1, you could only be considering uncompressed to be slightly higher, so obviously they are starting with a 1440x1080 frame to reason HDcam's compression ratio too...

HDV includes temporal compression (and as Jan points out, I was referring to XDcam as 35 Mbits/s not HDV) and therefore gains efficiencies. Even though all things cannot be equal, if all things WERE equal with a compression system that had two identical versions other than one version was employed temporally and spatially and one was allowed to use spatial compression alone, there IS an increased effectiveness visually with temporal compression employed. It's still a pain to edit, no question and the robustness of the tapes and the quality of the camera heads and all that are part of it, but looking at only image quality resulting from compression, at any given bitrate, image quality would be better using temporal and spatial compression vs using spatial compression alone. It's why temporal compression is employed for distribution...

The problem with temporal compression is that it's hard to edit and a tape dropout (I've not experienced one yet, but I know it happens) is a GOP long. Combine this with the fact that for the purposes of this discussion, all the temporally compressed video we are speaking of is at a very low bitrate and the I-frame compression we're talking about is much higher. A significant amount of the image quality differences are camera, codec design and bitrate-based as opposed to specifically temporal vs spatial compression based.

As I've said before, sure DVC ProHD is a better, more robust codec than HDV or XDcam, but the extent to which this is true may be more flexible than most have thought.





TimK,
Director,
Kolb Productions,

Creative Cow Host,
Author/Trainer
www.focalpress.com
www.classondemand.net