oavif

oavif is a tool for target quality AVIF encoding using fssimu2, a fast perceptual image quality metric.

Usage

oavif | [version]

usage:  oavif [options] <in> <out.avif>

options:
 -h, --help
    show this help
 -v, --version
    show version information
 -s, --speed u8
    encoder speed (0..10) [9]
 -t, --score-tgt f64
    target SSIMULACRA2 score (0..100) [80]
 --quality-alpha u8
    quality factor for alpha (0..100=lossless) [100]
 --max-threads u8
    maximum number of threads to use (1..255) [1]
 --tile-rows-log2 u8
    tile rows log2 (0..6) [0]
 --tile-cols-log2 u8
    tile columns log2 (0..6) [0]
 --auto-tiling 0/1
    enable automatic tiling [1]
 --tune str
    libaom tuning mode (ssim, iq, ssimulacra2) [iq]
 --tenbit 0/1
    force 10-bit AVIF output [1]
 --tolerance f64
    target quality error tolerance (1..100) [2]
 --max-pass u8
    maximum search passes (1..12) [6]
 -q, --quality u8
    quantizer (0..100), bypasses search

Input image formats: PNG, PAM, HEIC, HEIF, JPEG, WebP, or AVIF

A script is included in the scripts/ directory to test oavif's performance on a directory of images.

./scripts/measure.py --help
usage: measure.py [-h] [--tolerance TOLERANCE]
                  images_dir oavif_path output_csv

Measure oavif performance on a directory of images

positional arguments:
  images_dir            Directory containing input images
  oavif_path            Path to oavif binary
  output_csv            Output CSV file path

options:
  -h, --help            show this help message and exit
  --tolerance TOLERANCE
                        Tolerance value for oavif encoding

Compilation

Native (Zig)

Compilation requires:

• Zig 0.15.1
• libavif
• libwebp
• libjpeg-turbo
• libspng
• libheif

git clone https://github.com/gianni-rosato/oavif
cd oavif
zig build --release=fast

The oavif binary will be emitted to zig-out/bin. To install system-wide on macOS or Linux, run the following command as root:

zig build --release=fast --prefix /usr/local

Docker

A multi-stage Dockerfile is provided to build a fully static oavif binary in a controlled environment.

Workflow:

• build stage (Ubuntu):

  • Pull zlib and nasm from ubuntu repository
  • installs Zig and build tools
  • builds libjpeg-turbo, libwebp, libspng, and libavif (with local libaom) as static libraries into an isolated prefix
  • builds oavif with zig build, linking against that prefix

• final stage (distroless):

  • copies only the oavif binary into a minimal runtime image

Build via make:

• the Makefile builds the Docker image
• extracts the resulting oavif binary from the image to the current directory

Typical usage:

make
bin/oavif --version

This requires only Docker on the host; Zig and all dependencies are handled inside the container.

Tip

You can use the container in your containerized environment (e.g synology or Kubernetes) to automate a compression workflow.

License

oavif is governed by the Apache 2.0 License.