Adobe is dipping its toes into generative AI video for the first time, with new features that can turn still images into video and extend the length of existing video clips.

Generative video is rapidly becoming the next AI battleground, with OpenAI introducing its Sora service in selected countries late last year, joining more established services such as Runway and Synthesia.

Now Adobe is adding video to its portfolio of Firefly AI services, offering the ability to:

Alongside the AI video features being added to Premiere Pro, Adobe will launch a new Firefly web application from where users can generate videos. The company claims all the videos generated by the service are “IP-friendly”, meaning creators can add them to their content without fear of being pursued for copyright infringement.

The stunted length of the video clips being generated means it’s highly unlikely that anyone will use Adobe’s service to create full-blown movies or even short films. Instead, Adobe sees the primary use of generative video as b-roll, filling gaps in timelines or providing background images behind titles and credits, for example.

Adobe demonstrated several ways in which the AI could be used to generate VFX, potentially saving money on creating such effects manually. They showed it being used to create smoke effects drawn in a cartoon style, for example, or fireball explosions.

Perhaps the most interesting use of Adobe’s generative video is the ability to convert still images into moving video. The company demonstrated how a still image of a horse photographed on an Icelandic landscape could be turned into a short video clip. An accompanying text prompt asked for the horse to walk towards camera with the wind blowing gently. Within around 90 seconds the Full HD video clip was generated. Although the horse moves slightly unnaturally, it’s unlikely anyone would notice without studying the video very closely.

Adobe’s web app includes a range of controls that allow creators to fine tune the look of the generated videos. They can choose between widescreen and portrait aspect ratios, the latter being more useful for social media. They can choose the camera angle, and also determine the type of camera motion in the clip, for example panning left or right, or zooming in.

The generative video is also being built into Premiere Pro, where it can be used to artificially extend the length of real footage. In instances where you might need a shot to last a few extra seconds, the AI can generate new footage based on what was captured in camera, in much the same way it can extend the size of an image in Photoshop currently.

Other video AI tools being added to the web service include the ability to take hand-made models or sketches and use those as references for generative video, helping filmmakers to visualize how scenes may look when shot. The AI can also translate video or audio into 20 different languages, whilst retaining “an authentic voice”.

The bad news for creators – who might already be paying a hefty monthly fee for Creative Cloud subscriptions – is that the generative video services will incur an additional fee.

Firefly Standard provides 2,000 video/audio credits per month, which Adobe claims will allow creators to generate only 20 five-second Full HD clips, from $9.99 per month. Firefly Pro raises the quantity to 7,000 credits or 70 short video clips, and that will start from $29.99.

It will be available from today.