VFX Blender Add-on for removing objects from a video using Deep Learning
This Add-on for Blender 2.8x is utilizing Deep Learning methods to remove objects from videos. The technique to do this is based mainly on the following repository: https://github.com/nbei/Deep-Flow-Guided-Video-Inpainting. I've modified it so it can be used more easily. The changes are available in my fork.
If you don't have a Nvidia GPU in your system or your if are willing to sacrifice quality for faster computation check out the onion branch of this repository where I integrated another method for video inpainting into Blender.
Create a mask for the object you want to remove in the Clip Editor
. This Add-on is compatible with my AutoMask Blender Add-on that uses Deep Learning to track a mask in a video.
The Add-on is heavily dependent on python libraries. You can either install the dependencies to the Blender python or if python is already installed on the system and the python version is compatible with the Blender python version you can also install the dependencies on your system. In the latter case you need to replace #r'PYTHON_PATH'
with the path to your python site packages in the __init__.py
file.
.zip
filezip
file as Add-on in Blender (Do not yet activate it since the dependencies are missing)pip install -r requirements.txt
weights
. Follow these instructions.Q: The CuPy installation failed
A: Try installing a binary package (e.g. pip install cupy-cuda100
).
The software is for educaitonal and academic research purpose only [1, 2].
Deep-Flow-Guided-Video-Inpainting, LiteFlowNet original, LiteFlowNet pytorch, DeepFill