SD-video-to-HD-video-convertor

Objective/Aim

To develop a prototype of a video conversion process that can convert SD videos to HD videos using diffusion (image to image, Inpainting) type models, while preserving the context of the video and being efficient.

Requirements

• Python 3.x (python>=3.6 (>=3.9 in macOS arm))
• OpenCV: For video frame extraction and image processing.
• NumPy: For numerical operations.
• MoviePy: For compiling image sequences back into video format.
• Real-ESRGAN: For upscaling image frames.

Installation

1.To use this package, simply install it via pip:

• pip install realesrgan-ncnn-py moviepy
• pip install numpy
• pip install opencv-python

2.If linux user the you also need to run this:

• apt install -y libomp5 libvulkan-dev

Path

You need update input_video_path variable . Change the input_video_path = “path_of_your_video” Example in this repo: input_video_path = "input.mp4”

GPU

Realesrgan(gpuid=1) : make sure to change the “gupid” according to your gpu setup. By default gpuid = 0 which select default gpu for your system. In my case i am using gpuid=1, because I want to use that gpu.

If you don’t; know which one to choose then run “run.py” it will show all gpus of your system with their id’s and you can chose from them. Shown in below figure.

Usage

1. Prepare the input video: Place the SD resolution video that you want to upscale in the same directory where run.py file is present . For example, an input.mp4 file is already present in the this git repo.

2. Run the script: Ensure you are in the project directory and run the run.py script:

 python run.py 

3. View the output: The resulting HD video will be saved in the build directory as output_hd_video.mp4.

4. Recorded working of the code.

https://drive.google.com/file/d/1EpBf5wz2kqC0Z4CIwtj5rEaaa6Zq_Zqm/view?usp=drive_link