anibali/pywebp

Python bindings for WebP

Python

81

119 commits

updated Aug 4, 2026

See the code

README

WebP Python bindings

Build status License PyPI GitHub

Installation

pip install webp

On Windows you may encounter the following error during installation:

conans.errors.ConanException: 'settings.compiler' value not defined

This means that you need to install a C compiler and configure Conan so that it knows which compiler to use. See https://github.com/anibali/pywebp/issues/20 for more details.

Requirements

  • Python 3.8+

Usage

import webp

Simple API

# Save an image
webp.save_image(img, 'image.webp', quality=80)

# Load an image
img = webp.load_image('image.webp', 'RGBA')

# Save an animation
webp.save_images(imgs, 'anim.webp', fps=10, lossless=True)

# Load an animation
imgs = webp.load_images('anim.webp', 'RGB', fps=10)

If you prefer working with numpy arrays, use the functions imwrite, imread, mimwrite, and mimread instead.

Advanced API

# Encode a PIL image to WebP in memory, with encoder hints
pic = webp.WebPPicture.from_pil(img)
config = WebPConfig.new(preset=webp.WebPPreset.PHOTO, quality=70)
buf = pic.encode(config).buffer()

# Read a WebP file and decode to a BGR numpy array
with open('image.webp', 'rb') as f:
  webp_data = webp.WebPData.from_buffer(f.read())
  arr = webp_data.decode(color_mode=WebPColorMode.BGR)

# Save an animation
enc = webp.WebPAnimEncoder.new(width, height)
timestamp_ms = 0
for img in imgs:
  pic = webp.WebPPicture.from_pil(img)
  enc.encode_frame(pic, timestamp_ms)
  timestamp_ms += 250
anim_data = enc.assemble(timestamp_ms)
with open('anim.webp', 'wb') as f:
  f.write(anim_data.buffer())

# Load an animation
with open('anim.webp', 'rb') as f:
  webp_data = webp.WebPData.from_buffer(f.read())
  dec = webp.WebPAnimDecoder.new(webp_data)
  for arr, timestamp_ms in dec.frames():
    # `arr` contains decoded pixels for the frame
    # `timestamp_ms` contains the _end_ time of the frame
    pass

Features

  • Picture encoding/decoding
  • Animation encoding/decoding
  • Automatic memory management
  • Simple API for working with PIL.Image objects

Not implemented

  • Encoding/decoding still images in YUV color mode
  • Advanced muxing/demuxing (color profiles, etc.)
  • Expose all useful fields

Developer notes

Setting up

  1. Install uv.
  2. Create a development environment with all dependency groups:
    $ uv sync --all-groups
    

Running tests

$ uv run pytest

Cutting a new release

  1. Ensure that tests are passing and everything is ready for release.
  2. Create and push a Git tag:
    $ git tag v0.1.6
    $ git push --tags
    
  3. Download the artifacts from GitHub Actions, which will include the source distribution tarball and binary wheels.
  4. Create a new release on GitHub from the tagged commit and upload the packages as attachments to the release.
  5. Also upload the packages to PyPI using Twine:
    $ uv run twine upload webp-*.tar.gz webp-*.whl
    
  6. Bump the version number in pyproject.toml and create a commit, signalling the start of development on the next version.

These files should also be added to a GitHub release.

Known issues

  • An animation where all frames are identical will "collapse" in on itself, resulting in a single frame. Unfortunately, WebP seems to discard timestamp information in this case, which breaks webp.load_images when the FPS is specified.
  • There are currently no 32-bit binaries of libwebp uploaded to Conan Center. If you are running 32-bit Python, libwebp will be built from source.
image
pillow
python3
webp

anibali/pywebp

Python bindings for WebP

Python

81

119 commits

updated Aug 4, 2026

See the code

README

WebP Python bindings

Build status License PyPI GitHub

Installation

pip install webp

On Windows you may encounter the following error during installation:

conans.errors.ConanException: 'settings.compiler' value not defined

This means that you need to install a C compiler and configure Conan so that it knows which compiler to use. See https://github.com/anibali/pywebp/issues/20 for more details.

Requirements

  • Python 3.8+

Usage

import webp

Simple API

# Save an image
webp.save_image(img, 'image.webp', quality=80)

# Load an image
img = webp.load_image('image.webp', 'RGBA')

# Save an animation
webp.save_images(imgs, 'anim.webp', fps=10, lossless=True)

# Load an animation
imgs = webp.load_images('anim.webp', 'RGB', fps=10)

If you prefer working with numpy arrays, use the functions imwrite, imread, mimwrite, and mimread instead.

Advanced API

# Encode a PIL image to WebP in memory, with encoder hints
pic = webp.WebPPicture.from_pil(img)
config = WebPConfig.new(preset=webp.WebPPreset.PHOTO, quality=70)
buf = pic.encode(config).buffer()

# Read a WebP file and decode to a BGR numpy array
with open('image.webp', 'rb') as f:
  webp_data = webp.WebPData.from_buffer(f.read())
  arr = webp_data.decode(color_mode=WebPColorMode.BGR)

# Save an animation
enc = webp.WebPAnimEncoder.new(width, height)
timestamp_ms = 0
for img in imgs:
  pic = webp.WebPPicture.from_pil(img)
  enc.encode_frame(pic, timestamp_ms)
  timestamp_ms += 250
anim_data = enc.assemble(timestamp_ms)
with open('anim.webp', 'wb') as f:
  f.write(anim_data.buffer())

# Load an animation
with open('anim.webp', 'rb') as f:
  webp_data = webp.WebPData.from_buffer(f.read())
  dec = webp.WebPAnimDecoder.new(webp_data)
  for arr, timestamp_ms in dec.frames():
    # `arr` contains decoded pixels for the frame
    # `timestamp_ms` contains the _end_ time of the frame
    pass

Features

  • Picture encoding/decoding
  • Animation encoding/decoding
  • Automatic memory management
  • Simple API for working with PIL.Image objects

Not implemented

  • Encoding/decoding still images in YUV color mode
  • Advanced muxing/demuxing (color profiles, etc.)
  • Expose all useful fields

Developer notes

Setting up

  1. Install uv.
  2. Create a development environment with all dependency groups:
    $ uv sync --all-groups
    

Running tests

$ uv run pytest

Cutting a new release

  1. Ensure that tests are passing and everything is ready for release.
  2. Create and push a Git tag:
    $ git tag v0.1.6
    $ git push --tags
    
  3. Download the artifacts from GitHub Actions, which will include the source distribution tarball and binary wheels.
  4. Create a new release on GitHub from the tagged commit and upload the packages as attachments to the release.
  5. Also upload the packages to PyPI using Twine:
    $ uv run twine upload webp-*.tar.gz webp-*.whl
    
  6. Bump the version number in pyproject.toml and create a commit, signalling the start of development on the next version.

These files should also be added to a GitHub release.

Known issues

  • An animation where all frames are identical will "collapse" in on itself, resulting in a single frame. Unfortunately, WebP seems to discard timestamp information in this case, which breaks webp.load_images when the FPS is specified.
  • There are currently no 32-bit binaries of libwebp uploaded to Conan Center. If you are running 32-bit Python, libwebp will be built from source.
image
pillow
python3
webp

Project health

Overall

2.7/10

Maintained

1/10

Code review

1/10
OpenSSF Scorecard

Significant stargazers

Calum Young

34 followers · starred Mar 2025

Brice Gros

5 followers · starred Apr 2020