Artificial intelligence can already do a bunch of things with text and image generation from basic prompts, but the future of the technology pivoting towards realistic video generation has already been teased. Adobe is the first to market with this next step, with the launch of the Firefly Video Model. Available now in limited public beta, Adobe Firefly will let you turn text prompts, static images, and short video clips into transformed AI-generated video, and also allow for a lot more creative control over how the video is generated.
Firefly text-to-video lets you use text prompts, multiple camera controls, and reference images to generate AI-created B-roll to add into video timelines to fill gaps. Adobe states that this can come in handy for professional video production requirements, including clips that can be directly used, or even clips to suggest creative intent for large-scale production in place of 'placeholder' elements.
Not only does this work for animation-style clips, but Adobe Firefly will also be able to generate realistic-looking video clips entirely out of text prompts or reference images. Prompts can also help to adjust camera angles, lighting, temporal elements, colour grade, moods, and general video styling. Take a look at some of the sample clips generated by Adobe here.
Adobe Firefly's video generation features will eventually operate on a premium subscription model, but for the time being it's available for free through the limited public beta. Interested users can join the waitlist to receive an invitation to try the new features. Adobe also states that Firefly has been used to generate over 13 billion images since March 2023, and is being rapidly adopted by leading brands and organisations.