Adobe's AI-powered video generation tool will debut this year
Adobe is preparing to introduce its artificial intelligence (AI) model for video creation, Firefly Video. The tech giant has announced that features powered by this innovative model, will be available on the Premiere Pro beta app and a dedicated website before the end of 2024. Three key features - Generative Extend, Text to Video, and Image to Video - are currently in beta testing, but are expected to be publicly accessible soon.
Firefly's features and their applications
Generative Extend, a feature that allows users to lengthen any input video by two seconds, will be incorporated into the Premiere Pro beta app later this year. Also, Firefly's Text to Video and Image to Video models, which generate five-second video clips from prompts or input images, will also be launched on a dedicated website within the same timeframe. These tools have the potential to revolutionize content creation by significantly reducing production time.
Firefly aims to enhance existing workflows
Ely Greenfield, Adobe's CTO of Digital Media, stated that there is a "huge appetite" for tools like Firefly that can complement/accelerate existing workflows. He highlighted the popularity of Firefly's generative fill feature in Adobe Photoshop, describing it as "one of the most frequently used features we've introduced in the past decade." This indicates a growing trend toward integrating AI technology into creative processes.
Generative Extend explained
Generative Extend is designed to seamlessly add extra two seconds of footage to existing videos. It works by taking the last few frames of a scene and running them through Firefly's Video model, to predict the next couple of seconds. For audio, this feature can replicate background noise but not voices/music due to licensing requirements from the music industry. This tool could prove useful in situations where a scene ends prematurely and needs to be extended for transition/fade-out purposes.
Firefly's text to video and image to video features
Firefly's Text to Video and Image to Video features allow users to input a text or picture prompt, and receive up to five seconds of video in return. These AI video generators will be accessible on firefly.adobe.com, with potential rate limits. Adobe has also implemented safeguards in its video models, preventing them from generating videos that include nudity, drugs, alcohol, or public figures like politicians and celebrities.
