Video Archival
Table of Contents
I keep a copy of each DVD image or a decrypted backup of Blu-rays. This makes re-ripping easy, and might let me ignore discs rotting and such.
Workflow
Idealized workflow (future hbr additions)
In this workflow the mythical hbr gui does some semi-automated matching like musicbrainz picard does. The big difference is there is no central database for DVD/BD chapter info like there is for music releases. So instead we'll rely on the user pasting a blob of text with like "Title hh:mm:ss" or something and try to match chapters with their titles.
Ideally I'd like to depend directly on libhb, but it's not built as a standalone library. Alternatively keep relying on parsing HandBrakeCLI --scan and maintain the current hbr file format with some concessions made towards primarily using HandBrake's Video Profiles for encoder settings.
DVD
Play first with VLC to handle CSS encryption. Then exit and use ddrescue to make a copy of the disc:
vlc /dev/sr0 ddrescue /dev/sr0 /path/to/movie.iso
This doesn't decrypt, but VLC and Handbrake can do it themselves via libdvd / libdvdcss.
Blu-ray
MakeMKV's backup works great for copying Blu-ray contents and decrypting the AACS keys. Files are stored in video/iso/<movie title>/<Blu-ray Label>. The process is GUI driven and straightforward. Just make sure the decrypt option is checked when you start a backup.
When it comes to matching content names with titles on blu-ray you need a decent player. VLC has some support, but it's practically useless. There are a couple free windows players that do the job with some hiccups. Try DVDFab Player 5 or Leawo Blu-ray Player. DVDFab is preferable for it's ability to access to chapter numbers and such. I ran into some incompatibilies with certain studio's blu-ray menus. Sometimes you just have to do a lot of guess and check using the playlist files and context clues to properly match content.
Encoding with hbr and HandBrake
Creating hbr files:
hbscan.py
I created hbscan to parse HandBrakeCLI –scan output and create hbr outfile sections for each title. It is best suited for non-episodic releases (mostly movies) because it doesn't insert episode/season keys. It doesn't generate the initial config section. I usually copy that section from a previous hbr file. It would look something like this:
[CONFIG] input_basedir=/mnt/video/iso/<movie name>/ output_basedir=/mnt/video/movies/<movie name>/ iso_filename=<filename> type=movie name=<movie name> year=<movie year>
Then you run hbscan.py on the iso or blu-ray directory and redirect output to the hbr file:
hbscan.py /path/to/movie.iso >> test.hbr
For multi-disc releases you probably want hbscan.py –-setiso to set the iso_filename for each outfile section. hbscan.py can also exclude titles under a minimum length (usually intros, piracy warnings, etc.) Check hbscan.py –help for a quick overview.
This gives you a pretty good start on the hbr file. Most of the rest of the work is:
- Remove unwanted outfile sections
- Adjusting subtitle and audio track ordering
- Remove unwanted subtitle/audio tracks
- Setting names for each title
- Setting an extra type for non-feature content.
hbr-gen
Whereas hbscan takes the disc structure and lets you match content hbr-gen sort of does the opposite by working from a list of content that you match with disc titles. It also has simpler modes to generate a fixed number of outfile sections.
Running hbr:
hbr takes one or more hbr files as input, and runs encodes for each OUTFILE section.
Sometimes it's a good idea to verify the commands hbr will run first with the -d/--debug option:
hbr -d test.hbr
A normal run to actually encode is just:
hbr test.hbr
Other useful options are -e/--episode to specify a particular episode, -y/--overwrite to overwrite a previous encode without confirmation, and -n/--skip to only encode when a previous encode doesn't exist.
Checking files for corruption or bad encodes
Testing with ffmpeg and find
Change the file extension to suit.
find . -name "*.mkv" -exec sh -c "ffmpeg -v error -i '{}' -map 0:1 -f null - 2>'{}.log'" \;
Solution found via:
Testing with Spek (for audio, one file at a time)
Spek analyzes the audio spectrum and draws a spectrograph. It can be useful for detecting audio issues. It's a bit slow and not batchable, but the extra analysis might be useful.