Adobe Firefly AI Adds Video Generation Capabilities to Power Professional Video Creation
Adobe recently launched the Firefly video generation model, a major upgrade to its suite of generative AI tools. The model supports the creation of high-quality videos from text prompts and is now integrated into Premiere Pro. The Firefly Video Model is trained on commercially safe data, ensuring its output is suitable for commercial use.
Core Features of the Firefly Video Model
The Firefly Video Model can generate 5-second long videos at 1080p resolution from user text prompts, with a smooth playback of 10 frames per second. Users can access this feature in the Firefly web app or through the Extensions panel in Premiere Pro. Adobe states the model is designed for video editing professionals, offering controls for camera movement, motion effects, and style customization. It is currently in a limited preview, and interested users can apply for access via a waitlist.
Deep Integration with Premiere Pro
Within Premiere Pro, the Firefly Video Model can extend existing video clips using the Generative Extend feature or generate new B-roll footage. Users can create videos directly on the timeline with text prompts, significantly boosting workflow efficiency. Adobe emphasizes that all generated content comes with Content Credentials, which document the creation process for easy tracking and verification.
Commercial Safety and Training Data
The Firefly Video Model is trained on licensed content from Adobe Stock, public domain material, and other licensed third-party content, ensuring it is copyright-safe. The model is not trained on customer-uploaded Creative Cloud content. Output videos are tagged as synthetically generated, supporting the C2PA standard. Adobe is committed to continuously improving the model’s performance and plans to extend video length and resolution in the future.
Industry Reaction and Future Outlook
Several tech media outlets, including TechCrunch and The Verge, have reported on the launch, confirming the Firefly Video Model was officially announced on November 28, 2024. The preview is available to users of the Firefly for Business plan and is expected to drive the adoption of AI in video production. Adobe has stated that it will iterate on features based on user feedback to strengthen the role of generative AI in professional software.
