grav1an

grav1an is a binary which creates psychovisually optimized AV1 video using Vapoursynth, Av1an, Grav1synth, and MKVToolNix. Currently, it supports the encoders svt-av1-psy and rav1e. This was originally a Python script made by Ironclad, so credits to them for creating the original script. The AV1 anime discord server has the original script & support.

Todo:

• TVDB support
• Rentry support
• Comparison images
• Uploading torrents
• Separate rescale into optional feature
• Dependency checks at compile-time

Dependencies:

Bolded dependencies are required.

• FFmpeg/FFprobe (enable x264 to detect offsets)
• Vapoursynth
• Av1an
• svt-av1-psy/rav1e (at least one is required)
• MKVToolNix
• Grav1synth (required if --no-grain is unset)
• opus-tools (required if --original-audio is unset)
• BestSource/LSMASHSource/dgdecnv (at least one required)
• vszip (required if --ssimu2-algo isn't set to ssimulacra2_rs)
• vs-mlrt (required for scaling)
• vsjet (required for scaling)
• vodesfunc (required for scaling)
• lvsfunc (required for scaling)
• resize2 (required for scaling)
• descale (required for scaling)

Make sure that all binary dependencies are in your PATH environment variable.

Installing:

1. Install Cargo if you haven't already.
2. Clone & enter this repo

git clone --depth 1 --single-branch https://github.com/SuspiciousDuck/grav1an
cd grav1an

3. Build/Install

cargo install --path .

4. Profit

# Add $HOME/.cargo/bin to your PATH if you haven't already
grav1an

Usage:

Basic Example:

# This uses the directories ./show & ./show_out as the input & output directories.
# -w specifies the amount of workers to use when encoding (not threads).
# -n specifies the name of the show, which is used when creating output files.
# --no-torrent specifies that a resulting .torrent file shouldn't be made.
# This will encode 4 fast passes in order to target a SSIMULACRA2 score of 80 in the final encode!
grav1an -i ./show -o ./show_out -n Show -w 4 --no-torrent