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 #20 for more details.
- Python 3.8+
import webp
# 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.
# 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
- Picture encoding/decoding
- Animation encoding/decoding
- Automatic memory management
- Simple API for working with
PIL.Image
objects
- Encoding/decoding still images in YUV color mode
- Advanced muxing/demuxing (color profiles, etc.)
- Expose all useful fields
- Install
mamba
andconda-lock
. The easiest way to do this is by installing Mambaforge and then runningmamba install conda-lock
. - Create and activate the Conda environment:
$ conda-lock install -n webp $ mamba activate webp
- Install PyPI dependencies:
$ pdm install -G:all
$ pytest tests/
- Ensure that tests are passing and everything is ready for release.
- Create and push a Git tag:
$ git tag v0.1.6 $ git push --tags
- Download the artifacts from GitHub Actions, which will include the source distribution tarball and binary wheels.
- Create a new release on GitHub from the tagged commit and upload the packages as attachments to the release.
- Also upload the packages to PyPI using Twine:
$ twine upload webp-*.tar.gz webp-*.whl
- 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.
- 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.