Notebook and tools for end-to-end automation of music video production with generative AI
Repository Stats
208 stars
36 forks
Repository Activity
This repository has been starred 208 times and forked 36 times, showing its popularity in the developer community.
README Preview
# Video Killed The Radio Star [](https://colab.research.google.com/github/dmarx/video-killed-the-radio-star/blob/main/Video_Killed_The_Radio_Star_Defusion.ipynb) ## Requirements * ffmpeg - https://ffmpeg.org/ * pytorch - https://pytorch.org/get-started/locally/ * vktrs - (this repo) - `pip install vktrs[api]` * stability_sdk api token - https://beta.dreamstudio.ai/ > circular icon in top right > membership > API Key * whisper - `pip install git+https://github.com/openai/whisper` ## FAQ **What is this?** TLDR: Automated music video maker, given an mp3 or a youtube URL **How does this animation technique work?** For each text prompt you provide, the notebook will... 1. Generate an image based on that text prompt (using stable diffusion) 2. Use the generated image as the `init_image` to recombine with the text prompt to generate variations similar to the first image. This produces a sequence of extremely s