Adobe Firefly Video Model teased; will compete with OpenAI's Sora

Adobe is all set to take over the AI video generation market with its latest Firefly Video Model teased recently. The upcoming generative AI video tool by Adobe got a teaser yesterday showcasing its capabilities. The teaser shows the text-to-video results of the Adobe Firefly Video Model that was announced by Adobe in April 2024.

Adobe Firefly Video Model

The model will let the users generate video clips using text descriptions and further modify the results with multiple camera controls that simulate camera angles, shooting distance, and even motion. In addition to that, an image-to-video feature was also demonstrated for the Firefly Video Mode teaser that can generate video clips using reference images. As suggested by Adobe, the latest tool in their inventory will help the users create B-roll footage or patch the gaps in a production timeline.

If the teased footage is anything close to the final result capabilities of Adobe Firefly, then OpenAI’s Sora is going to get tough competition from the Adobe offering. In a report by The Verge, it has been said by Adobe’s Generative AI VP Alexandru Costin that the videos created by image-to-video and text-to-video will have a maximum length of five seconds.

On the other hand, Sora offers up to one minute-long video generation which is almost 20 times more than what has been claimed for Adobe Firefly. However, Adobe Firefly has the upper hand over Sora when it comes to being commercially safe as the model is being trained on openly licensed, public domain, and Adobe Stock content only.

This could also reduce the risk of copyright infringement to a great extent. Moving forward, Adobe also showed some additional slips of the ‘Generative Extend’ feature that can extend the length of a video by adding frames to it. Until now, there is no concrete information related to the release of the Adobe Firefly Video Model. However, the company has mentioned that the new model will be launched later this year.