[CinCV TNG] cinelerra-gg Speed Fade Automation oddity

Good Guy good1.2guy at gmail.com
Mon Sep 11 01:56:45 CEST 2017


If you have a good sound card, your dog might hear it.
The frequencies will be multiplied by the speed factor, like 25.
So, first problem is that the render duration must be reduced
by the speed factor.  Normally, the render ends as the cursor
passes the end of the track.  In this case, the media ends way
before this.  Next, if you are using audio, you have to have data
that can be re-sampled effectively, such as s16le pcm raw data.

I used the ffmpeg aeval plugin on a silent track and created
raw audio which was 25 times too slowwww (using t/25).  Then
I replayed/rendered the data using a/v speed 25 and I used
the in/out markers to reduce the playback/render duration by
a factor of 25.  The video zooms by... and the audio now is
audible because it was sampled in a human frequency range.

I have uploaded a tarball with the original mp3 reference media,
the render session with ffmpeg aeval plugin, and the rendered
result.  Works for me.

download link: http://www.datafilehost.com/d/53832e67
here are the directions to reproduce the experiment:

1) untar fred3.tar in /tmp
2) run cin, open session /tmp/fred3.xml
3) click F_aeval show icon, plugin parameters pops up
  notice that the t variable is always used as t/25
4) close popup
5) enable speed auto graphic curve, (press 9 in mwindow)
6) double click video speed curve and set speed to 25
7) since the unit sample rate is 25x, change the timeline
units to frames (rightt mouse timeline, set time format - frames)
set the in/out markers [ ] to [0, 133] (133= 3331/25)
8) Press R (render), render gui appears
9) set the "render range" to in/out
10) Select file format FFMPEG / qt
11) press audio wrench, select s16le.qt, ok
12) press video wrench, select mp4.qt, ok
13) set "Select a file to render" as /tmp/fredx.qt
14) click ok to render.

Now, open and render /tmp/fredx.qt with cin, or your media player.
This has been known to work at least once.

On Sun, Sep 10, 2017 at 3:50 PM, Frederic Roenitz <ml1 at aimvis.de> wrote:

> Hi GG,
> the rendered time lapse video has audio problems.
> I tested this with an one hour video and different speed auto
> constellations.
> Easiest test case: One speed handle at 0:00:00:00 for each
> track. Selecting about 3 seconds of the clip. Render selection.
> extremely rough approximations:
> speed 15: audio ok
> speed 20: inconspicuous audio dropouts
> speed 25: 50% audio dropouts
> speed 40: 90% audio dropouts
> speed 50 and above: 100% audio dropout
> And if I add (later in the timeline) another speed handle with high
> speed and one frame later speed 1, the speed 1 section is missing
> audio completely, too.
> Please help.
> Kind greetings
> Frederic
> _______________________________________________
> Cinelerra mailing list
> Cinelerra at lists.cinelerra-cv.org
> http://lists.cinelerra-cv.org/cgi-bin/mailman/listinfo/cinelerra
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cinelerra-cv.org/pipermail/cinelerra/attachments/20170910/cdcae274/attachment.html>

More information about the Cinelerra mailing list