Expect text-to-video and image-to-video capabilities across Adobe’s CreativeCloud with the release of the Firefly video model beta “later this year”.

The company claims “rapid innovation” since the launch of Firefly in March last year, including new models in imaging, design and vectors. These power Photoshop’s ‘generative fill’ feature, used for a claimed 12 billion images and subject of some controversial modifications, as well as ‘generative remove’ in Lightroom, ‘shape fill’ in Illustrator and ‘text-to-template’ in Express.

Now Adobe is preparing for the introduction of video features, and is taking names for the launch of a beta version of the Firefly Video Model. Interest is high, with a video update viewed 167,000 times in two days.

Adobe says it will power workflows in tools such as Premiere Pro, and assures users it has been designed with creators’ rights in mind, adding that the model is “designed to be commercially safe and is only trained on content we have permission to use”. A video says this will be limited to content in the public domain or licensed.

AI will be applied to editorial tasks such as “navigating gaps in footage, removing unwanted objects from a scene, smoothing jump cut transitions, and searching for the perfect B-roll takes time”. On its website, it shows a number of applications in which text prompts have been used to create sophisticated video. They include an image of woman wearing glasses, these “adorable monsters”, and a cinematic closeup of a reindeer in a snowy forest at sunset. The developer says all were generated in under two minutes.