How to Build a Netflix VOID Video Object Removal and Inpainting Pipeline with CogVideoX, Custom Prompting, and End-to-End Sample Inference

This tutorial shows how to set up a Colab environment and run Netflix's VOID video object removal pipeline.

It installs dependencies, clones the void-model repo, and checks for GPU availability using PyTorch.

The workflow downloads a CogVideoX inpainting model and the void_pass1 checkpoint, then runs sample inference.

An optional OpenAI step can generate a refined background prompt if the user provides an API key.