|
|
Posted by Richard Crowley on 09/20/06 05:54
"Toby" wrote ...
> Some of you might remember our discussion about dropouts in DV a
> couple of months ago. I found something rather interesting in this
> month's IBE:
>
> "MPEG-2 video uses 4:2:0 sampling (one Cb sample, one Cr sample and
> four Y samples) for each pixel (except 4:2:2 and High profiles).
> Samples are then grouped into blocks (or squares) of 16 x 16 pixels, a
> macroblock. If the data from any of these samples is lost, the decoder
> cannot reproduce the block.. When this occurs in an I frame,
> subsequent B and P frames may also be affected, causing the data
> corruption to continue until the next good I frame, which could be a
> second or more in long GOP encoded data. These errors are very
> apparent to the viewer. These errors are usually a result of
> transmission problems (RF or IP) or 'drop-outs' in videotapes."
>
> So obviously there are drop-outs in DV, which--while rarer than in
> analogue perhaps--can be egregious and are of concern to the industry.
Well, of course there are dropouts in DV, but that does not
really follow logically from the description of handling
dropouts in MPEG2. And since there is no temporal
compression in DV, dropouts last only one frame each.
Navigation:
[Reply to this message]
|