ffmpeg upscale videos with the sr filter using a docker container
ffmpeg's sr filter applies super-resolution methods based on convolutional neural networks to upscale videos.
The sr filter uses Libtensorflow and Cuda but the version of Libtensorflow must match your version of Cuda which must in turn match your NVIDIA driver and ffmpeg must be built with the tensorflow option.
This can make it tricky to set up, but luckily someone has built a docker image with ffmpeg built with all requirements preinstalled
In this video i show you how to install the ffmpeg-tensorflow docker image and use it with docker or podman, install the nvidia-container-toolkit which is used to upscale video with ffmpeg inside the docker container using your gpu
I also cover a workaround needed for systems running selinux like Fedora to allow the container access to the filesystem
ffmpeg sr filter
https://ffmpeg.org/ffmpeg-filters.htm...
super-resolution-podman.
https://github.com/NapoleonWils0n/cer...
ff-super-resolution script
https://raw.githubusercontent.com/Nap...
ffmpeg-tensorflow
https://github.com/MIR-MU/ffmpeg-tens...
video stack exchange
https://video.stackexchange.com/quest...
docker
https://hub.docker.com/r/miratmu/ffmp...
Watch video ffmpeg upscale videos with the sr filter using a docker container online without registration, duration hours minute second in high quality. This video was added by user NapoleonWils0n 14 July 2023, don't forget to share it with your friends and acquaintances, it has been viewed on our site 1,253 once and liked it 21 people.