Basics and General Workflow

Preparation

downloading a source, looking at the video, some decisions (resolution(s) for the release, audio codec, group-specific requirements)

Writing the Script

imports, source filter (mention lsmash, ffms2), examples for resizing, debanding, AA. with images if possible

Encoding the Result

$ vspipe.exe script.vpy -y - | x264.exe --demuxer y4m --some example --parameters here --output video.264 -

Editors for VapourSynth usually have inbuilt support for encoding scripts you wrote. Use %encode --y4m <clip_variable> in Yuuno or the GUI provided by VSEdit.

Transcoding Audio

As said earlier, only qaac will require configuration, everything else can simply be extracted wherever you like. Nonetheless, it is easier to add every binary you use to the PATH environment variable, and the rest of this guide will assume you’ve done exactly that.

Decoding audio with FFmpeg and piping it out

Basic ffmpeg usage is simple. You just need to specify your input file and your desired output file, like this:

ffmpeg -i "input.dtshd" "output.wav"

This command will decode the DTS-HD MA audio file and encode it as a WAVE file. This command doesn’t specify any options so ffmpeg will resort to using its defaults.

For transcoding from DTS-HD MA only one option is needed:

ffmpeg -i "input.dtshd" -c:a pcm_s24le "output.wav"

The -c:a pcm_s24le parameter will tell ffmpeg to encode its output with a depth of 24 bits. If your source file is 16 bits, change this parameter to pcm_s16le or simply skip it because 16 bits is the default.

The command above will decode the source file and save a resulting WAVE file on your drive. You can then encode this WAVE file to a FLAC or AAC file, but there is a faster and more convenient way to do that: piping. Piping skips the process of writing and reading the data to and from a file and simply sends the data straight from one program to another.

To pipe from ffmpeg, specify the output format as WAVE using the -f option, replace the output filename with a hyphen and place a pipe symbol at the end, which will be used to separate the ffmpeg command from your encoder command, like this:

ffmpeg -i "input.dtshd" -c:a pcm_s24le -f wav - | {encoder command}

Encoding to FLAC

To encode to FLAC, we will use the flac command line program:

flac -8 --ignore-chunk-sizes --bps 16 "input.wav" -o "output.flac"

-8 sets the encoding level to 8, the maximum level. This will result in the best compression, although at the expense of encoding speed. FLAC encoding is fast, so just stick with level 8.

--ignore-chunk-sizes is needed because the WAVE format only supports audio data up to 4 GiB. This is a way to work around that limitation. It will ignore the length field in the header of the WAVE file, allowing the FLAC encoder to read files of any size.

--bps 16 specifies the “bits per sample” to be 16. As noted earlier, 16 bits are enough for our purposes and FLAC additionally has the downside that all 24-bit samples are padded to 32-bit samples in the format, meaning 25% of the storage used is completely wasted.

To encode audio piped from ffmpeg, replace the input filename with a hyphen and place the whole command after the ffmpeg command, like this:

ffmpeg -i "input.dtshd" -c:a pcm_s24le -f wav - | flac -8 --ignore-chunk-sizes --bps 16 - -o "output.flac"

Encoding to AAC

First, set up qaac:

• Go to its download page and download the newest build (2.70 at the time of writing) and makeportable.zip.
• Extract the x64 folder wherever you want qaac to be, then extract contents of makeportable.zip inside it.
• Download the iTunes setup file (iTunes64Setup.exe) and move it to the x64 folder.
• Run the makeportable.cmd script
• You are done. You can now delete the iTunes installation file.

To encode from a file, use the following command:

qaac64 --tvbr 91 --ignorelength --no-delay "input.wav" -o "output.m4a"

The --tvbr 91 option sets the encoding mode to True Variable Bitrate (in other words, constant quality) and sets the desired quality. qaac has only 15 actual quality steps in intervals of 9 (0, 9, 18, ... 127). The higher the number, the higher the resulting bitrate will be. The recommended value is 91, which will result in a bitrate of about 192 kbps on 2.0 channel files, enough for complete transparency in the vast majority of cases.

--ignorelength performs the same function as --ignore-chunk-sizes in FLAC. --no-delay is needed for proper audio/video sync¹.

To encode audio piped from ffmpeg, replace the input filename with a hyphen and place the whole command after the ffmpeg command:

ffmpeg -i "input.dtshd" -c:a pcm_s24le -f wav - | qaac64 --tvbr 91 --ignorelength --no-delay - -o "output.m4a"

Encoding to Opus

Encoding a file with opusenc will look like this:

opusenc --vbr --bitrate 160 --ignorelength "input.wav" "output.opus"

--vbr sets the encoding mode to Variable Bitrate (which is the default but it never hurts to be explicit), while --ignorelength does the same thing as in qaac.

As you may have noticed, opusenc uses bitrate control rather than some kind of constant quality mode². Instead of an abstract number that corresponds to a quality target like with x264‘s CRF mode and qaac‘s TVBR mode, you give the encoder your preferred resulting bitrate and it chooses the constant quality target itself. The recommended bitrate is 160 kbps for 2.0 channels and 320 kbps for 5.1.

Encoding audio piped from ffmpeg works the same as for previous encoders—just replace the input filename with a hyphen:

ffmpeg -i "input.dtshd" -c:a pcm_s24le -f wav - | opusenc --vbr --bitrate 160 --ignorelength - "output.opus"

Muxing

mkvtoolnix