Sony DCR-SR100 and MPEG 2 Issues

Jul 8, 2004
London, Ontario, Canada
I am thinking about the the new 30GB hard drive based Sony DCR-SR100 simply because it seems more convienient.

Video is secondary to shooting stills for me, and I like the notion of not having to deal with tapes, being able to directly back up to dual layer dvd in a laptop, and it looks like I might even be able to use it as a HD recorder for capturing TV shows/transferring old analog video tapes for transfer to dvd (though I am not exactly sure yet if it will function as a "digital VCR").

I saw on another thread here that MGEG 2 is an issue as it is compressed - how would MPEG2 compare in quality to the other current formats? To hi-8? Just a little worse than DVC?

I also saw that there is some software to decompress - anyone have any insights?

Here is the only review I can find - I like the fact it apparantly has very good low light performance:

Please also keep in mind I am still shooting in LOW 8mm and I don't bother selling my video.


My research is bringing to light some of the issues with mpeg2 - quality and ability to edit.

I have to decide on the quality issues and here is a quote I found on a DVD forum that relates to editing and a program called VideoReDo:

Apparantly it will support *better* editing of mpeg2 but there still will be lost data (same as going to mp3 from wav). I still need to think about whether the trade offs are worth the convienice.

All further thoughts condisered.


The post I refer to (by "smarty")

VideoReDo really is an excellent program, and does reconstruct a 30 frame
per second time base for editing. The left and right arrow keys do indeed
step through the video content in 1/30th of a second clicks.

However, the mpeg2 encoding scheme does not ensure discrete and unique and
separate frames at this rate. Rather, it only encodes video whose motion
vectors meet or exceed pre-determined thresholds. For the mpeg2 encoders
which have user-specified motion estimation controls, then choosing the
right thresholds for the specific type of video content could guarantee that
the mpeg2 compressed stream does indeed have the necessary frame to frame
detail to permit editing on these 30 samples per second to achieve absolute
accuracy. For the vast majority of situations, especially in a consumer DVD
camcorder, these thresholds are NOT user specified. More importantly, by the
time VideoReDo sees the video content, the camera's encoder has made a lot
of premature and incorrect assumptions about what to keep and what to
discard, and has thrown away 90% of the "avi" sampled content which can
NEVER be accurately regenerated. This is the very reason why multi-pass
mpeg2 encoders, which first parse the video and build predictive and
backward (B and P frame) candidates on the first pass before doing the
actual mpeg2 encoding on the 2nd pass look so much better.

My point before as well as now is that avi / linearly sampled and lightly
compressed (non intraframe compressed) video from a miniDV camcorder has
nearly 10 times the data to work with, and no "rash assumptions" about what
a cheap hardware encoder would keep or discard as it runs "on-the-fly"
attempting to write a DVD and throw away 90% of the data in the process. No
amount of editing software sophistication can recover the original detail,
despite appearances to the contrary.

Does any of this matter in real, day to day editing? My opinion is that it
matters a great deal for critical editing work and not much at all for most
users who are merely doing home movies.