Adobe introduces Firefly-powered text-to-video AI tools for video generation

Adobe has unveiled its forthcoming text-to-video generative AI tools powered by Firefly, introducing capabilities for video production, including a feature that creates video clips from still images. Users can modify these generated clips with simulated camera controls to change angles, motion, and shooting distance. Additionally, the text-to-video feature enables the creation of video clips from textual descriptions, while an image-to-video feature generates clips using reference images, beneficial for creating supplementary B-roll footage or filling gaps in production.

The quality of the generated videos is on par with OpenAI’s Sora model, which Adobe is considering integrating into Premiere Pro. Videos generated through Firefly will have a maximum duration of five seconds, and its training on openly licensed, public domain, and Adobe Stock content ensures "commercially safe" usage, mitigating copyright infringement issues.

Initially, the text-to-video and image-to-video features will be available in beta as a standalone Firefly application later this year. Eventually, the Firefly video model will be integrated into Adobe Creative Cloud and Adobe Express applications. Adobe also previewed a "Generative Extend" feature for Adobe Premiere Pro, designed to extend existing video footage similar to Adobe Photoshop Generative Expand tool for images. The release date for this feature is expected later this year.