Our media encoding process

It occurred to me that I started posting to debugfs as if I just left off from my old domain (a story for another time). I had something in the order of 300 articles on the site so naturally posts on this site don't have the context they had on the old site.

When I first started debugfs I talked about the Handbrake command-line script I run on our KODI server to shrink the size of Blu-rays (since they can be huge). I didn't really get into the details of the whole process and I've since changed how I rip and encode media.

When I buy a DVD/Blu-ray the first step I take is to back-up the media using MakeMKV. MakeMKV is great for dumping both Blu-ray and DVD content to a .mkv file. I prefer .mkv over .mp4 because I love subtitles and the .mp4 container only lets you "burn" one subtitle into the file. Files in an .mkv container can contain as many subtitles as the DVD/Blu-ray has. I normally rip these on my desktop workstation which has a late 2013 AMD A8-5600K APU with 16GB RAM, a 500GB SSD, and an NVidia GTX 650 Ti Boost video card. I then use Filezilla to SFTP them to the 8TB drive in our i7-2600 (non-K)-based KODI server. The i7-2600 is a much better processor than the A8-5600K with almost double the compute power despite the fact that both have similar clock speeds and core count (the i7 has more threads).

I added an inexpensive Corsair H60 water cooler not too long ago because I planned on re-encoding a lot of video on our KODI machine, it's been doing a wonderful job at keeping the CPU cool despite all 8 threads ramping up to almost 100%.

I wrote a simple script that uses the handbrake-cli package to re-encode the .mkv files and embed all subtitles in the new .mkv container file. I looked at the resulting files on a 1080p TV and couldn't tell the difference. My original script looked very different.


I've since modified the script to look something like this - hb.sh:

#!/bin/bash
for x in *
do
     HandBrakeCLI -i "$x" -o "${x%.*}-264.mkv" --preset="H.264 MKV 1080p30" --all-subtitles
done

 Note that you have to have the handbrake-cli software installed, this doesn't use the GUI version of Handbrake. I still use the graphical version of Handbrake occasionally, but I've found myself using it less and less. The reason? Speed.

Lately I found myself with a large backlog of un-ripped television shows. Almost two months ago I bought the first season of Happy Days because my son and I were watching Night Shift and I thought it would be nice for him to know Henry Winkler as I remember him. We found the first season on sale at one of the movie specialty stores in Cambrige Mall ($10 new if I remember correctly). But I never got around to ripping the media. Also on the slate were several other shows that I managed to all rip in under 2 hours last night using MakeMKV.

MakeMKV preserves the contents of the disc (minus closed captions which seem to be a problem for both Handbrake and MakeMKV) but this means large file sizes. I used my A8-5600K APU-based workstation to rip the media with MakeMKV, then I used Filezilla to transfer them over to our KODI server in the living room. That server has a much more powerful i7-2600 CPU (which by today's standards is pretty middling). I then ssh into the KODI server and run the hb.sh script above to re-encode the video.

It's important to note that the script takes ALL files in the current directory and tried to encode/compress them to the 1080p30 format. If you have .srt subtitle files you should probably temporarily move them to another folder. The script creates another file with the same filename that has -264.mkv appended to the end.

so:
jolly-frog.mkv

would end up as:
jolly-frog-264.mkv

Really this probably isn't saving any time over just using Handbrake to encode from scratch, but it frees my time up since I don't have to babysit Handbrake. The process is: rip on workstation, transfer to server, encode on the server.

The only downside to this process is that the graphical version of Handbrake has the ability to add tags to files, but this is something I haven't really used much and that KODI doesn't seem to care as much about as it does proper naming.

Update:  Because I ssh into our KODI server to run the script I used to have to leave that SSH session open. So what I've done is put all the files I want to newly encode in a subdirectory called 111/. I then run:

nohup hb.sh &

This lets me detach from the session and shut down my (A8-5600K-based) workstation PC.