On Sun, 19 Aug 2001 12:37:40 -0700, Bob Diaz wrote:

{most of a very interesting post deleted, for brevity...]
>AVIO / Sequel / ScreenPlay Comparison, Part 11, Image Quality
>
>NATIVE DV EDITING:
>
>As long as we do not have to decompress and re-compress Native DV, any
>copy made is still first generation quality. One could load Native DV
>into a system, do cuts only editing, dump it back to DV, load it back
>into the system again, and still have first generation footage. For
>those with limited storage space, Native DV editing allows users to load
>in raw footage, select/trim the footage they want, and save it back to
>tape. This pre-editing will provide a final tape of just the footage
>you intend to use.
>
>Now the bad news, DV is VERY sensitive to multi-generation compression
>noise. While I don't have an AVIO DV to test (I just have the AVIO ST),
>I have seen second and forth generation DV images. Second generation DV
>shows a VERY slight compression noise. If you don't know what to look
>for and are not looking closely, you're likely to miss it. Forth
>generation DV compression noise is more noticeable and easier to spot.
>If you are looking for the DV compression noise, don't look at the
>details, rather look at large areas of a solid color, like the sky.
>
>Any time a title, effect or transition is used, the video must be
>decompressed, processed, and compressed again. That results in a
>generation loss. Everyone's idea of what is good enough for them is
>different, but I believe that the majority of users will find the VERY
>slight second generation compression noise created during a title,
>effect or transition, to be too minor to worry about. Once the tittle,
>effect, or transition is over, the image is back to first generation DV
>quality. However, the level of compression noise generated during
>multi-layered (Native DV) effects may pose a problem.
>
>I've found that MPEG-II and Wavelet handle multi-generation layers much
>better than Native DV. If your video requires multi-layered effects,
>you may wish to consider using a compression system other than DV for
>editing.
>
>NOTE: I do NOT have an AVIO DV to test Native DV with. Thus, all I
>could do was to base my conclusions on seeing Native DV with other
>equipment. I hope that AVIO DV Users will try the following tests: (1)
>Take Native DV Footage and add a Rectangle to the image. Take the clip
>created and add another rectangle to it. Do that for 2 more
>iterations. That is the "easy" multi-generation test. (2) Take a clip
>and lower the brightness by 3%. Now take the clip created and raise the
>brightness by 3%. Do that for 2 more iterations. That is the "tough"
>multi-generation test. Be sure to look in areas that have a solid
>color, like a sky, for compression noise and share your findings.
[....]
>Bob Diaz
>bdiaz@my-deja.com

Wouldn't the results of multigenerational rendering of DV
depend on the quality of the codec used? I did a forced
rendering for 10 generations using the Canopus Raptor codec
and alternating brightness values, with only minor losses
(see: www.David-Ruether-Photography.com/original_vs_10th-gen.htm),
yet at another site, some earlier-version DV-codecs performed
poorly with only five generations - though later versions of
these were generally improved... (see:
http://members.home.net/dgcom/MiniDV/DVcompressors.htm).
(It appears that QT5 has been substituted for the labeled
QT4.1 in the chart - and the results are far better than
they were for the earlier version, but still well short
of the MainConcept and Canopus codecs for both image
quality and render speed. Fortunately, MC is available
for OCHI-card users...).