On Sat, 20 Apr 2002 14:01:19 -0700, newvideo@amug.org (Bill Davis) wrote:
[...]
>If you suck your footage into an NLE, then apply an effect such as a cross
>dissolve, many systems will decompress, encode the transition, then
>re-compress the footage back into the DV stream. In that case, compression
>artifacts (while typically minor) can be introduced into the code of the
>transition. This can also be true of other "calculated" effects such as
>color correction, etc.
>
>Most people can't tell the difference between original captured footage
>and stuff decompressed/recompressed through quite a few generations -
>always depending on the nature of the image content. (stuff like highly
>contrasting diagonal lines will reveal compression artifacting quicker
>than something like a human face)
[...]
>Start decompressing/recompressing willy nilly, and you need to start
>paying attention to your signal quality.
[...]

The quality of the image after several
decompressed/recompressed cycles can
vary from barely acceptable to virtually
the same as the original, depending on
the codec quality. Two good ones are the
Canopus (only for Canopus cards), and the
MainConcept. For an example of a
10-generation forced rendering with the
Canopus codec, go here:
www.David-Ruether-Photography.com/original_vs_10th-gen.htm
For examples of 5th generation changes
with other codecs, go to the referred
URL at the site above (I trust it is still
running...), or to the other rendering
examples at: www.adamwilt.com