Meta announced their video generation model, Movie Gen, that can create high-definition video and also edit existing footage with text prompts. Movie gen is also able to to generate high-fidelity, synced audio, including ambient sounds, sound effects, and background music. It won't be available yet, but it presents new ways creators can be empowered to make high-quality videos and sounds, as well as increasing concerns among creatives like filmmakers and artists.
It's been several months since OpenAI announced Sora, as well as Google with Veo, and now Meta has joined the train with their own video generation model, Movie Gen. It is one of the pros of having competitors in a field of technology as it causes innovative developments, but it is more interesting how much AI has grown since two years ago when AI generation hit the mainstream.
Similar to image generation, Movie Gen can generate videos using text inputs and even edit already existing video footage. And it adds AI-generated sounds to match the video, including background music and sound effects.
Movie Gen can generate videos up to 16 seconds at 16 frames per second from still images and videos using optional prompts. And it can generate high-quality audio for up to 45 seconds. For example, from an image of a young lady, Movie Gen generated a video using the prompt "A woman DJ in a pink jacket spins records with a cheetah by her side."
 |  |
Many of the videos Meta Movie Gen has generated are not only of high-quality, but they have an impressive level of consistency with representing humans in features and behavious. "Our model achieves state-of-the-art results when it comes to creating personalised videos that preserve human identity and motion." Meta says in a blog.
Having AI generate videos that are potentially usable is already in itself a feat, but to have AI-generated sounds come along with it—that even sounds really good—and synced up is another level of creation that will enable people to do a lot more without even being able to organically do so themselves.
As much as AI video generation may change the way videos are made with time, it also raises more concerns, if not exacerbates the existing ones with image generation. There have been reports of data scraping of many YouTube videos to rain some of these existing AI models and reports of AI abuse of AI that violate privacy terms across many platforms.
If at this point AI is able to generate imagery and sounds with this level of detail, then one cannot really say what it'll be able to do in a few years to come. The future where AI drives the world appears closer than we imagine. Movie Gen, like OpenAI's Sora and Google's Veo, will not be publicly available to use yet.
Creatives such as audio creators, filmmakers, and even actors and artists are mostly concerned about how these AI technologies affect them at their jobs. Meta, however, shares in a blog their intentions to "take a collaborative approach" and "work closely with them to integrate their feedback to help enhance inherent creativity."
Monitise your content on Hive via InLeo and truly own your account. Create your free account in a few minutes here! Here's a navigation guide.
Image credits: Meta
Posted Using InLeo Alpha
