I need a specific version of ffmpeg to be able to read video with torchvision (0.12.0+cu113), which currently supports ffmpeg 4.2. I have already tried to upgrade ffmpeg with the next command:
add-apt-repository -y ppa:jonathonf/ffmpeg-4
But it gives me a version 4.3.2. How to install exactly 4.2?
Related
I'm installing pytorch geometric on Google colab. I've done this lots of times before and had no issues but it has suddenly stopped working. I've not changed my code since it worked. Here is how I install it:
!pip install torch==1.8.1 torchvision torchtext
import torch; print(torch.__version__); print(torch.version.cuda)
!pip install torch-scatter -f https://pytorch-geometric.com/whl/torch-1.8.1+cu102.html
!pip install torch-sparse -f https://pytorch-geometric.com/whl/torch-1.8.1+cu102.html
!pip install torch-cluster -f https://pytorch-geometric.com/whl/torch-1.8.1+cu102.html
!pip install torch-spline-conv -f https://pytorch-geometric.com/whl/torch-1.8.1+cu102.html
!pip install torch-geometric
The pytorch version should be 1.8.1+cu102, confirmed by the print statement above. I specify the version when installing with !pip install torch-scatter -f https://pytorch-geometric.com/whl/torch-1.8.1+cu102.html. However, when I import torch_geometric I get the error:
Detected that PyTorch and torch_scatter were compiled with different CUDA versions. PyTorch has CUDA version 10.2 and torch_scatter has CUDA version 11.1. Please reinstall the torch_scatter that matches your PyTorch install.
Why is torch_scatter not compiling with CUDA version 10.2? Is there a way to force it to compile with this version?
you could try to specify the latest wheel version provided by the link you use: https://pytorch-geometric.com/whl/torch-1.8.1+cu102.html (for November 22nd 2021 it is 2.0.8):
pip install torch-scatter==2.0.8 -f https://data.pyg.org/whl/torch-1.8.1+cu102.html
It looks like the latest torch-scatter version in Google Colab is 2.0.9 which is newer than 2.0.8. Therefore, when you run your command it does not do anything thinking that the latest version is already installed.
I want to use avconv for audio related operations on my Ubuntu Bionic 18.04.5 system, but it's not getting installed as a part of the following packages.
sudo apt-get install ffmpeg:i386
sudo apt-get install ffmpeg
I still get..
avconv: command not found
So I tried creating a soft link.
sudo ln -s /usr/bin/ffmpeg /usr/bin/avconv
Now If I run
$avconv
ffmpeg version 3.4.8-0ubuntu0.2 Copyright (c) 2000-2020 the FFmpeg developers......
After updating my mac to high sierra, I can't use ffmpeg
$ ffmpeg
dyld: Library not loaded: /opt/local/lib/libidn2.0.dylib
Referenced from: /opt/local/lib/libgnutls.30.dylib
Reason: image not found
Abort trap: 6
I appreciate any help!
I had the same problem after a security update. I found this Github issue with the same error message for ffprobe, a sister program. The solution was to uninstall and install again.
On macOS, I did:
brew uninstall ffmpeg
brew cleanup
brew install ffmpeg
Depending on your installation, you may need to run brew uninstall --ignore-dependencies ffmpeg (if you have dependencies such as opencv that require ffmpeg) or brew uninstall --force ffmpeg (if you have multiple versions).
And now FFMPEG works:
$ ffmpeg -version
ffmpeg version 4.2.1 Copyright (c) 2000-2019 the FFmpeg developers
...
I tried to update latest release on youtube-dl but I can't update and while I tried got the error like
"It looks like you installed youtube-dl with a package manager, pip, setup.py or a tarball. Please use that to update."
Try:
youtube-dl -U
If you used pip to install youtube-dl in the first place then, use:
pip install --upgrade youtube-dl
If that doesn't work, you can always reinstall it using homebrew:
brew install youtube-dl
To install use:
brew install youtube-dl
To upgrade use:
brew upgrade youtube-dl
Is there a step-by-step guide on how to compile avconv in Ubuntu?
It seems hard to search for any tutorial with avconv compared to ffmpeg.
I finally found out how,
1.) Make a directory avconv-source
mkdir avconv-source
2.) Download and install the x264 library
cd ~/avconv-source
git clone git://git.videolan.org/x264.git x264
cd x264
sudo ./configure --enable-static
sudo make
sudo make install
3.) Download the avconv source
cd ~/avconv-source
git clone git://git.libav.org/libav.git avconv
cd avconv
sudo ./configure
sudo ./configure --enable-gpl --enable-libx264
sudo make
sudo make install
and now you can execute
avconv -i test.mov -c:v libx264 -c:a copy test.mp4
This worked for me on Ubuntu 13.04 and Ubuntu 12.04.3 LTS.
apt-get install libavcodec-extra-53
It worked great for me on Ubuntu 12.04, except that it complained that assembler was missing.
The minimum version is yasm-1.2.0.
If you really want to compile without asm, configure with --disable-asm.
Thanks to HOWTO: Install and use the latest FFmpeg and x264, I did the following as well, and it was OK beyond that.
sudo apt-get install build-essential checkinstall
sudo apt-get build-dep yasm
wget http://www.tortall.net/projects/yasm/releases/yasm-1.2.0.tar.gz && \
tar -xf yasm-1.2.0.tar.gz && cd yasm-1.2.0 && ./configure
make
sudo checkinstall --pakdir "$HOME/Desktop" --pkgname yasm --pkgversion 1.2.0 \
--backup=no --default