Meta introduced its text-to-image generation AI Make-a-Scene in July, which like Dall-E and Midjourney uses machine learning algorithms (and huge databases of scraped online artwork) to create fantastic renderings of written prompts . On Thursday, Meta CEO Mark Zuckerberg revealed Make-a-Scene’s more animated contemporaries, Make-a-Video.
As the name suggests, Make-a-Video is “a new AI system that allows humans to turn text prompts into short, high-quality video clips,” Zuckerberg wrote in a meta-blog on Thursday. Functionally, Video works the same as Scene – it relies on a mix of natural language processing and generative neural networks to turn non-visual prompts into images – it just pulls content into a different format.
“Our intuition is simple: learn how the world looks and is described from paired text-image data, and learn how the world moves from unsupervised video footage,” a team of meta-researchers wrote in a research paper , which was released Thursday morning. In this way, the team was able to reduce the time required to train the video model and eliminate the need for paired text-video data while “preserving the breadth (variety in aesthetics, fantastical representations, etc.) of today’s imaging models. ”
Like almost all of Meta’s AI research, Make-a-Video is released as an open-source project. “We want to think carefully about how we build new generative AI systems like this,” noted Zuckerberg. “We openly share this generative AI research and results with the community to get their feedback and will continue to leverage our responsible AI framework to refine and evolve our approach to this emerging technology.”
As with seemingly every Generative AI released, the potential for abuse of Make-a-Video is not small. To stay one step ahead of potential nefarious shenanigans, the research team preemptively sanitized the Make-a-Video training dataset with all NSFW imagery and toxic wording.
All products recommended by Engadget are selected by our editorial team, independently of our parent company. Some of our stories contain affiliate links. If you buy something through one of these links, we may receive an affiliate commission. All prices are correct at time of publication.
https://www.engadget.com/metas-new-make-a-video-can-generate-quick-movie-clips-from-text-prompts-130042758.html?src=rss Meta’s new Make-a-Video AI can generate quick movie clips from text prompts