UNLOCKING MY DOCO DREAMS FROM DVD
This post isn't just another one of my documentary recommendations,
but also a tech post. The reason is that it's a documentary that I
found particularly difficult to get off the internet and into a
format that's usable to me.
To begin with, the tech part [actually this ended up so long that
I'll do the documentary review as a separate "part 2"]. This was
largely a battle with FFmpeg, hampered further by my limited
capacity for large downloads over my home internet connection which
required me to do everything on a VPS.
California Dreamin', in documentary form, was a German production
for their TV station ZDF. Extensive searching revealed this copy
with German narration on YouTube, possibly with the option of some
dodgy auto-translated subtitles, but I've given up on that feature:
https://www.youtube.com/watch?v=-s7VXGIgHNU
https://www.youtube.com/watch?v=ZzWLrFharf0
https://www.youtube.com/watch?v=Upa48HW1Gks
Alternatively it's on the Internet Archive, in a straight ZIP
archive of the files from a DVD release:
https://archive.org/details/2001-california-dreamin-dvd
Even though it was uploaded by IA's best known figure, Jason Scott
back in 2012, it's only had 319 views. Probably because a ZIP of
contents from two DVDs is pretty inconvenient to use, and the
description is entirely in German. Browsing through the ZIP
contents on the IA website however, I found that it included a scan
of the DVD cover, in German as well but apparantly claiming to
include an English ("Englisch") language option.
At 15GB that ZIP would have almost blown my 6 month internet data
quota on its own, so I relaunched one of my free Oracle Cloud VPSs
which come with excessive amounts of storage space. Unfortunately
its installed Ubuntu version had gone EOL while it was
automatically stopped by Oracle due to disuse, but for this it
matters little. On there I downloaded the ZIP and decompressed it
to find apparantly two big VOBs per episode of the five-part
documentary spread over two DVDs. 'ffprobe' showed that they indeed
had multiple audio tracks, matching the list of options on the DVD
cover: German, English, Italian, and French. Easy - 'cat' together
pairs of VOBs, then select the English audio stream to put into a
more compressed set of per-episode English-language video files
small enough for me to download.
First problem is that FFmpeg doesn't show what language the audio
streams are for, so I had to extract snippets of them as MP3s using
the '-map' stream-selection option, download, them, then listen to
the narration after the title music (California Dreamin', of
course). I found the right one in the first VOB, and of course it
would be the same stream number for the others right?... Better
check... nope! In the end it turned out to be something like stream
5 for the first episode but stream 3 for the others.
So I converted the first episode using my normal options plus
mapping audio to the English-language stream, downloaded it, and...
The audio was way out of sync. It was actually a few seconds ahead
of the video, made obvious by the convenient old-fashioned A/V sync
numbers/bleeps at the beginning. Worse, I now discovered that
although it looked like there were two VOBs per episode, they
weren't actually aligned with episode start/end so I had the start
of episode two in my episode one file.
The first problem turned out to be due to my usual '-r 24' option
to set the framerate down to 24fps and avoid wasting data on frames
I won't notice. The original was 25fps and somehow the frame
dropping doesn't work right with VOB files, even though it hasn't
caused me trouble with any other formats. I removed that option and
the audio sync was correct at 25fps.
I couldn't find an easy answer to the other problem, unable to run
graphical DVD ripper programs because "SSH -C -X" over my home
internet is way too laggy and can chew through data itself. But
eventually I discovered that there's a fork of FFmpeg with extra
DVD/BR logic added, triggered via a special "dvd://" file type.
It's called vgtmpeg, and not packaged for Ubuntu so I decided
"security be dammed" and installed the static binary from its
webpage (that's what a throwaway VPS is good for anyway).
http://godromo.com/gmt/vgtmpeg
So much easier! Now it can list all the audio streams along with
their descriptions showing which language they're for. Plus I can
select titles and they're fetched in the correct duration
regardless of how the VOBs are laid out. Plus the audio streams
don't move around between episodes anymore! So I can do the
conversion like this (with my long sequence of specific audio/video
output format options setting bitrate etc. removed):
vgtmpeg -i dvd:///home/ubuntu/downloads/California\ Dreamin\'\,\ 2001/Disc\ 1\ -\ Wilder\ Westen\,\ Goldrausch\,\ Wellenreiter/?title=3 -map 0:0 -map 0:2 -acodec mp3 -vcodec mpeg4 california_dreamin_ep1.avi
Or to do all the episodes on the first disc:
for i in 1 2 3; do vgtmpeg -i dvd:///home/ubuntu/downloads/California\ Dreamin\'\,\ 2001/Disc\ 1\ -\ Wilder\ Westen\,\ Goldrausch\,\ Wellenreiter/?title=$[ $i + 2 ] -acodec mp3 -vcodec mpeg4 -map 0:0 -map 0:2 $i.avi; done
Then I got to download all five episodes while using a little under
1GB of data unstead of 15GB, ready to play on the old hacked video
game console that I use as a video file player on my TV. It's easy
once you know how, but oh how much time I could have saved if the
main FFmpeg project had included this functionality! The fork,
using a GPL2-licenced library, has been around for many years so
it's not like the developers haven't had the opportunity. Do I want
to know why the FFmpeg developers have't integrated this code?
Probably not.
Finally, it crossed my mind to upload these more convenient video
files, or higher bitrate ones that others would find more
acceptable, to the Internet Archive, so other doco lovers needn't
also go to such trouble (and the IA website's own video player
would work). Unfortunately though, that's pretty awkward to do from
the VPS since IA's uploader doesn't work with simple terminal-based
browsers like Links (indeed not much of their website does
anymore). There's a special command-line uploader, but it looks
like it would take some work to figure out using it properly to
handle the description, uploading multiple files, etc.. Plus now
I've talked about the doco here I'll have to make a new 'The Free
Thinker' IA account to do it without harming my anonimity.
If a fellow documentary fan with lots of internet data to burn
wants to upload them to the Internet Archive for me, send me an
email at
[email protected] and I can provide a download link to
get them from the VPS. Or just make them yourself, since I've shown
you how.
- The Free Thinker
