Adobe Launches AI-Powered Video Tools in Premiere Pro and Firefly

Adobe has officially entered the realm of generative AI video with the launch of its Firefly Video Model. This new technology brings a suite of AI-powered tools directly to creatives, including features inside Premiere Pro that allow users to extend clips and generate video content from images and text prompts.

Generative Extend: Fixing Footage with AI

The first of these tools, Generative Extend, is now available in beta for Premiere Pro users. Generative Extend helps address common issues, like footage that’s too short or eye-line shifts, by extending the beginning or end of clips by up to two seconds. While it’s ideal for minor fixes, it could save users from needing to reshoot footage to correct small mistakes. It can also smooth out audio transitions, extending ambient sound and effects by up to ten seconds, though it doesn’t work on dialogue or music.

Text-to-Video and Image-to-Video: Generative AI on the Web

In addition to Premiere Pro, Adobe is rolling out two other video generation tools in its Firefly web app. Text-to-Video allows users to input a text description to create short video clips, emulating various styles like real film, 3D animation, and stop motion, and can further be refined with camera angles, motion, and shooting distance. Image-to-Video offers even more control, enabling users to upload reference images alongside text prompts to create b-roll or visualize reshoots.

Limitations and Comparisons to Other AI Video Generators

Currently, these AI-generated videos are capped at five seconds and 720p resolution at 24 frames per second, meaning they aren't yet suitable for full-scale production. By comparison, OpenAI’s Sora promises to generate up to one-minute videos with high visual quality, though it hasn’t been released to the public yet. Adobe’s Firefly tools, however, stand out for their “commercially safe” nature, having been trained on legally sourced content, unlike other models that have faced scrutiny for using scraped data. Text-to-Video, Image-to-Video, and Generative Extend each take approximately 90 seconds to process.

A Boost for Commercial Viability and Content Ownership

A unique feature of Adobe’s Firefly model is its ability to embed Content Credentials in AI-generated videos. These credentials help disclose the use of AI in creating or editing videos and clarify ownership rights when published online. This transparency could be a crucial advantage for businesses concerned with ethical content creation.

AI Video Tools at Adobe MAX Conference

The announcement of Adobe’s Firefly Video Model and its related tools came at the Adobe MAX conference, where the company also revealed a host of other AI-powered features across its creative software suite. While these AI tools are still in beta, they are publicly accessible—offering a glimpse into the future of AI-driven video production.

What This Means for AI Video Generation

Adobe’s entry into the generative AI video space signals a major shift in the industry, as established creative tools begin integrating AI capabilities directly into production workflows. This move could accelerate the adoption of AI-driven video generation across professional settings, bridging the gap between AI experimentation and mainstream use, while setting new standards for ethical content creation.