Re: Editing HV20 HDV Footage

David,

It seems we are talking about two very different scenarios when we try to
compare our tests / results.

I have been doing a very simple 92 second movie, consisting of two clips,
each 45 seconds long joined together with an intervening 2 second crossfade.
This yields a 92 second long movie in which the first segment of
*****approximately***** 45 seconds is unchanged *****normally***** from the
input clip to the output clip, rerendering needs to be done to make the
crossfade, and then another *****approximately***** 45 seconds is then
unchanged from the input clip. There is no need *****normally***** to
rerender any part of the first or last approximately 45 second pieces.

I use the word "approximately" since the crossfade is very likely to cut a
GOP at each end, thereby creating the need to rerender the 2 GOPS
surrounding the crossfade. There are 15 frames in the long HDV GOP, and thus
another half second of video on each side might be rerendered, leaving a
total of 3 seconds out of 92 to rerender, 2 for the cross fade and .5 apiece
for in and out GOPs. The rest of the video is totally, entirely, and
absolutely a mere file transfer, with no degredation or other changes
whatsoever. (Strictly speaking the video and audio streams need to be
re-interleaved, new time / presentation stamps written, new headers created)
but the video content is literally the same as the input, no more and no
less.

I use the word *****normally***** above to distinguish the Trial version of
the software from the actual program. Since each and every frame of the
input clips as well as the crossfade now is output with an Adobe logo in the
upper right quadrant, I contend (and Richard agrees) that the Trial version
is thus forced to remake each and every frame / all GOPs for the entire 92
second movie. This, as I am sure you understand, requires uncompressing the
original GOPs, additively combining each 1440 by 1080 frame with the logo,
and recompressing the new frames, for the entire 92 second video, but only
in the Trial version.

Since the output file now contains a stationary, large Adobe logo in the
upper right quadrant lasting for 92 seconds rather than the original moving
video, occupying nearly one quarter of the entire frame, the new output
video should actually be a lot smaller, since the motion estimation vectors
in this quadrant collapse to zero valued scalars. The time required to
recompute and regenerate this extra step is what Richard and I contend
should add a lot of additional rendering time.

It is entirely possible that Adobe Elements makes no attempt to distinguish
regions of the clip which are unchanged from those which are changed, and
thus, as you say, rerenders everything. It is a bad strategy, and has no
upside, but they may chose to do this, in which case the actual program I
receive from Amazon will still take a lot of time unneccesarily to rerender
the entire 92 second clip even though the Adobe logo / watermark is not
present.

For the scenario / tests I have conducted, I am expecting the program to
only generate, at most, 90 new frames (3 seconds max by 30 frames per
second) and consider 88 seconds a relatively nominal time to do so, given
the additional need to do the other 2670 frames of essentially disk file
input, output, and stream regeneration.

I would certainly welcome your measurement of this very same 92 second
scenario, since it appears that you have perhaps been applying the crossfade
across nearly all of the 45 second clips rather than merely juxtiposing the
clips with a brief crossfade between them.


Smarty





.

Relevant Pages

  • RE: Movie Maker Problems
    ... pane - That's how you'll avoid missing the exact frame. ... Wait for the bars to cool off. ... Right click the 2nd part's video segment. ... > between clips. ...
    (microsoft.public.windowsxp.moviemaker)
  • Re: Whats Next, After Exhausting MM?
    ... One recent project I completed was a 5-minute demo video for my ... Then I'd combine clips from all video sources ... with "clean" audio from the mixer board and sometimes loop in audience ... can either frame them perfectly, and have them move out of the frame ...
    (microsoft.public.windowsxp.moviemaker)
  • Re: Whats Next, After Exhausting MM?
    ... One recent project I completed was a 5-minute demo video for my ... Then I'd combine clips from all video sources ... with "clean" audio from the mixer board and sometimes loop in audience ... can either frame them perfectly, and have them move out of the frame ...
    (microsoft.public.windowsxp.moviemaker)
  • Re: Whats Next, After Exhausting MM?
    ... One recent project I completed was a 5-minute demo video for my ... Then I'd combine clips from all video sources ... with "clean" audio from the mixer board and sometimes loop in audience ... can either frame them perfectly, and have them move out of the frame ...
    (microsoft.public.windowsxp.moviemaker)
  • DirectShow MPEG-1 keyframe detection
    ... Subject: DirectShow MPEG-1 keyframe detection ... total frames, the IPB frame ratio is 1:5:2 which gives me ... > capability to a DirectShow video editing application that ... the correct sample time for that I-frame would have put it ...
    (microsoft.public.win32.programmer.directx.video)