On Monday, Adobe introduced Firefly Video Mannequin, a brand new AI-powered text-to-video era software that may create novel movies from written prompts. It joins comparable choices from OpenAI, Runway, Google, and Meta in an more and more crowded subject. In contrast to the competitors, Adobe claims that Firefly Video Mannequin is skilled solely on licensed content material, doubtlessly sidestepping moral and copyright points which have plagued different generative AI instruments.
Due to its licensed coaching knowledge roots, Adobe calls Firefly Video Mannequin “the primary publicly accessible video mannequin designed to be commercially secure.” Nonetheless, the San Jose, California-based software program agency hasn’t introduced a normal launch date, and through a beta take a look at interval, it is solely granting entry to individuals on a ready checklist.
An instance video of Adobe’s Firefly Video Mannequin, supplied by Adobe.
Within the works since at the least April 2023, the brand new mannequin builds off of strategies Adobe developed for its Firefly picture synthesis fashions. Like its text-to-image generator, which the corporate later built-in into Photoshop, Adobe hopes to intention Firefly Video Mannequin at media professionals, akin to video creators and editors. The corporate claims its mannequin can produce footage that blends seamlessly with historically created video content material.
