Edge 444: Learn About Movie Gen: Meta AI's Amazing Audio-Video Generation Model
The new model represents an important milestone open source video and audio generation.
Video is rapidly becoming one of the next frontiers in generative AI. From OpenAI’s Sora startups like Runway or Pika to the new efforts from Google, most of the large video generation models remained closed-source. There are open-source efforts such as Stability AI’s Stable Video, but the quality is not at the same level as that of the large commercial alternatives. The barrier of entry for pretraining video models is fundamentally larger than language or image, as datasets are more scarce and difficult to produce.
Can open-source AI compete in the video generation space?
Recently, Meta unveiled some of their recent work in video and audio generation with the publication of the Movie Gen research paper. The initial indications look both promising and very impressive. Take a look: