Example

Cool-chic encodes each video frames successively through the cc_encode.py script.

Encoding an image into a .cool bitstream

Encoding an image or a video frame with CoolChic requires to specify

  • The input image path -i and output bitstream path -o

  • A working directory --workdir where logs are written.

  • The number of training (encoding) iterations --n_itr.

  • The decoder configuration --dec_cfg_residue for the neural networks architecture.

  • The rate constraint --lmbda setting the compressed file size.

(venv) ~/Cool-Chic$ python cc_encode.py \
    -i=image.png \
    -o=./bitstream.cool \
    --workdir=./dummy_workdir \
    --dec_cfg_residue=cfg/dec/intra/hop.cfg \
    --n_itr=10000 \
    --lmbda=0.001    # Typical range is 1e-2 (low rate) to 1e-4 (high rate)

More details on encoding images with Cool-chic is available in the encoder documentation.

Encoding a video into a .cool bitstream

Encoding a video requires encoding successively each video frames through the cc_encode.py script. We provide a samples/encode.py script allowing to easily do video encoding.

# Encode the first 33 frames of a video, with an intra at the beginning (0)
# and at the end (-1)
(venv) ~/Cool-Chic$ python samples/encode.py \
    -i myTestVideo_1920x1080_24p_yuv420_8b.yuv \
    -o myTestVideo_1920x1080_24p_yuv420_8b.cool \
    --workdir=./dummy_workdir \
    --lmbda=0.001 \
    --n_frames=33 \
    --intra_pos=0,-1

Decoding a .cool bitstream

Decoding an image or a video with CoolChic requires specifying the input and output paths. We provide a few already encoded bitstreams, ready to be decoded in samples/bitstreams/.

  • The input bitstream path -i and the decoded image path -o

(venv) ~/Cool-Chic$ python cc_decode.py -i=samples/bitstream/kodim14.cool -o=decoded-kodim14.png

More details on decoding images with Cool-chic is available in the decoder documentation.