Adobe plans to release a new AI video editing tool called Firefly Video Model. The company will offer a limited beta version later this year to help videographers with brainstorming and editing tasks.
Firefly Video Model joins Adobe's existing Firefly suite, which already includes tools for creating images, designs, and vector graphics.
This marks Adobe's first venture into AI-powered video models, putting it in competition with offerings from startups like Runway and Pika Labs. It might also compete with Sora, but OpenAI's first video model is still in research mode.
Alexandru Costin, Adobe's Vice President of Generative AI, told Reuters that the tool can generate a five-second video clip from a text or image prompt. Users can also specify camera angles, pans, moves, and zoom effects.
Adobe claims that its Firefly video tool will accurately follow prompts and outperform other video models. The company says it has worked closely with the video editing community over the past several months to refine the Firefly video model.
Training only on public and licensed data
Adobe notes that the video model, like its image models, is trained only on public domain and licensed content that the company has permission to use, and does not use Adobe customer content in its training data.
However, there has been some controversy over Adobe's use of AI-generated images uploaded to Adobe Stock for AI training, which would in turn mean that it is training on copyrighted material, somewhat indirectly.
In addition to the Firefly video model, Adobe is also introducing Generative Extend, a tool that will be available in Adobe Premiere Pro video editing software. It allows users to add two seconds to an existing clip by generating an appropriate insert to fill in gaps in the footage.
Interested parties can join a waiting list here to receive updates and access to the beta version of the Firefly video model.
