(Credit: Adobe)
Adobe today announced the availability of the Firefly Video Model in public beta, which allows users to generate video content from text prompts or existing images.
The company claims that it’s the only commercially safe option—that is, it doesn’t violate any intellectual property rights, so companies can use it without fear of being sued. Adobe also announced new Firefly plans that include not only this video-generation capability, but also the updated web app’s other tools such as image generation, vector generation, 3D image editing, and the also-new audio and video translation tools.
Generate Video Content With Firefly
The new Generate Video feature can produce up to 5 seconds of 1080 video. For comparison, a ChatGPT Plus account ($20 per month) includes Sora video generation with the same 5-second limit at a resolution of only 720p. For 1080p, you need the $200-per-month ChatGPT Pro account, which also increases the duration limit to 20 seconds. Google’s Veo generative video AI isn’t yet available, but the search company says it will be capable of 4K content for over 2 minutes. Meta’s free Make-a-Movie feature produces 5-second videos at resolution of 1,024-by-1,024, but the company has demonstrated 1080p content of 16 seconds.
(Credit: Adobe)Text prompts can be up to 1,000 words, and you can add your own image as a reference for creating the AI video. You get a good degree of control over your video’s look in Firefly, with choices like shot size (close-up, medium, or long shot), camera angle (including aerial, eye level, and top down), and motion (things like zooming in and out and panning left or right). You also choose a frame rate and aspect ratio, with 24fps and 16:9 the defaults. One thing we’ll have to see is whether Firefly generates improved prompts for you the way VideoGen does.
As noted above, Adobe claims that the video you generate with Firefly won’t impinge on anyone’s intellectual property, and it doesn’t use your input for training. Adobe has a documented set of AI Ethics principles. The videos created in Firefly will come with Content Credential metadata to let the world know that the content was generated by AI rather than in the real world.
3D and AI Translation Capabilities in Firefly
Aside from video, a generative AI capability Adobe announced today is Scene-to-Image. This allows 3D creators to perfect a 3D sketch into a high-resolution image, changing angles and perspectives and render production-ready assets.
Firefly can also now translate Spoken words from either video or audio files into any of more than 20 languages. The voices created will match the tone and acoustic features of the original.
New Firefly Web App Plans
(Credit: Adobe)Starting today, users will be able to buy subscriptions to Firefly. A Firefly Standard plan costs $9.99 per month and gets you 2,000 AI credits per month and up to 20 5-second 1080p video generations. For $29.99 per month, the Pro plan includes 7,000 credits and up to 70 5-second generations. A Premium plan for larger organizations is coming soon, with higher limits.
Firefly Video in Creative Cloud Apps
Adobe has already announced or even implemented several Firefly-powered tools in its other Creative Cloud applications such as Photoshop and Premiere Pro. With today’s announcement, the company announced the availability of a “limited taste” of the new AI features for Creative Cloud subscribers. Premiere Pro users are also getting the previously announced Generative Extend feature, which uses AI to lengthen video clips to fit their project.