Adobe has announced the upcoming release of its latest generative AI-powered video creation and editing tool, Firefly Video Model, which will be available in beta later this year. This addition to Adobe's suite of creative tools aims to enhance the company's offerings in the expanding field of AI-driven video generation, a space already populated by competitors such as OpenAI's Sora and Stability AI's Stable Video Diffusion.

The Firefly Video Model enables users to create five-second video clips from a single text or image prompt. Designed to simplify the video creation process, the tool allows for customisation of camera angles, motion, zoom, and panning, according to the previews shared by Adobe.

The tool is able to respond accurately to detailed prompts, a feature that Adobe's VP of generative AI, Alexandru Costin, emphasised in a recent interview with Reuters. "We've invested in making this model reach the level of quality and prompt understanding that videographers expect, and we've made sure it respects guidance much better than other AI video models," said Costin.

According to the company, the Firefly Video Model adheres to intellectual property guidelines. It is trained exclusively on public domain or licensed content, including the Adobe Stock database, which contains 400 million images, illustrations, and videos that are free of intellectual property conflicts.

In the official press release, Adobe states that the Firefly Video Model is particularly adept at generating natural scenes, such as landscapes, plants, or animals, and can be used to create atmospheric elements like fire, smoke, dust, and water. These elements can be layered over existing footage using Adobe's tools, such as Premiere Pro and After Effects, to add dynamic visuals to projects.

Alongside the Firefly Video Model, Adobe is introducing Generative Extend, a tool integrated into Adobe Premiere Pro. This feature can extend existing video clips by up to two seconds, filling gaps in footage with AI-generated inserts.