smooth-frames-web-ui
Web UI tool for smoothing videos using RIFE ncnn Vulkan.
Demo
Input video
input.mp4
Output video
output.mov
Setup
Setup script (macOS)
You can run this shell script to perform all setups at once.
This project uses rye as the Python project management tool.
Note that running the setup script will install rye on your machine!
(Note that if it is already installed, the installation process will be skipped.)
What is executed
- Install
rye - Creating .env files
- Download
RIFE ncnn Vulkan - TODO: Not yet, See Manual ->
Installffmpeg (supports OpenH264)
Manual (Windows or macOS or Linux)
First you need to install RIFE ncnn Vulkan in your environment.
Release binaries for each OS are available in GitHub releases, so please download them.
Next, put the path to RIFE ncnn Vulkan in your .env file.
Copy the .env_sample with the name .env.
Then specify the path.
Example setting .env:
RIFE_NCNN_VULKAN_PATH =./rife-ncnn-vulkan/rife-ncnn-vulkan
Then install the Python dependencies.
This project uses rye.
This program requires that FFmpeg be installed in the execution environment. In addition, OpenH264 must be supported by FFmpeg in order to export in H.264.
Here is the installation command assuming macOS.
curl --proto '=https' --tlsv1.2 -sSf https://raw.githubusercontent.com/shinshin86/setup-mac-ffmpeg-with-openh264-ciscobinary/main/setup-mac-openh264.sh | bash
Usage
Run the following command to launch Gradio's Web UI.
Access to localhost:7860.
About codecs
On the Web UI, previews can only be displayed in webm format (with the codecs currently available for selection).
Therefore, the default is to export files in webm (VP9 codec).
However, mp4 (MPEG4-V codec) has a smaller file size after conversion.
You can change these codecs in the drop-down list.
Google Colab
Does not work as expected on Google Colab
See this issue for details.
#1
To use on Google Colab, execute the following command
!apt-get install -y libvulkan-dev
!git clone https://github.com/shinshin86/smooth-frames-web-ui.git
%cd smooth-frames-web-ui
!wget https://github.com/nihui/rife-ncnn-vulkan/releases/download/20221029/rife-ncnn-vulkan-20221029-ubuntu.zip
!unzip rife-ncnn-vulkan-20221029-ubuntu
!echo "RIFE_NCNN_VULKAN_PATH =./rife-ncnn-vulkan-20221029-ubuntu/rife-ncnn-vulkan" > .env
!pip install gradio python-dotenv
!python launch.py --share
Special thanks
I came up with this program after learning about RIFE ncnn Vulkan.
