One of the big leaps in the artificial intelligence space is video generation. The industry has seen the launch of tools like OpenAI’s Sora, Meta’s Movie Gen, Runway, and Kling AI, ensuring strong and healthy competition. Now, Adobe has announced the availability of Firefly, its AI-powered video generator.
Adobe’s AI-powered video generator can extend clips
Currently, Adobe Firefly offers three main tools: Generative Extend, Text-to-Video, and Image-to-Video. Starting with Generative Extend, it’s available in beta as part of the Premiere Pro suite. As its name suggests, the feature is capable of extending video segments, making it ideal for correcting errors in the final or middle sections.
Generative Extend can only extend the section of the video you need by two seconds. Therefore, its primary purpose is to apply corrections to minor errors rather than to complete footage. While two seconds may not sound like much, it can be the difference between using a piece of footage or having to re-record it. The tool can output 720p/1080p video at 24 FPS.
The feature also extends audio by up to 10 seconds. However, it will only work for audio effects and ambient sound, not for dialogue or soundtracks.
Text and image-based video generation
Adobe’s Text-to-Video and Image-to-Video are available in limited-access public beta from Firefly’s web app. Text-to-Video works like most AI-powered video generators available on the market. The tool is able to generate an output based on a text prompt. Text-to-Video also includes options to emulate certain styles of footage. For example, you can choose between cinematic clips, 3D animation, and stop motion. You can also simulate things related to the “camera,” like angles, movement, or distance.
On the other hand, Image-to-Video creates videos by combining a picture and a text prompt. You can use this to generate clips from an image you extracted from your footage. However, the results of this tool are currently improvable.
Firefly is “commercially viable” for creators
Adobe stresses that its Firefly video generator model is “commercially viable.” In other words, the suite of tools allows users to commercially exploit the content it generates without worrying about potential lawsuits. The company assures that it trained the model based on content for which it had full authorization to use. This may be a factor in a segment embroiled in controversies related to copyright.
Lastly, Firefly generates clips that support Content Credentials, a standard that helps identify whether an image or video originated from AI.