Adobe, a leading developer of creative software, has announced the launch of a new generative AI video model now in public beta. The company claims it to be the first commercially secure AI-powered video creation model available on the market.
This model is integrated into the latest version of the Firefly web application, offering a suite of professional-grade creative tools that enable creators to produce artwork anytime and integrate it seamlessly into Adobe Premiere Pro.
The new generative video capabilities are powered by Adobe Firefly Video Model, a robust proprietary AI framework capable of generating video clips from text prompts or images. It supports features such as camera angle control for framing shots, creating high-quality B-roll footage, and crafting imaginative elements or custom motion designs based on descriptions.
The model has been optimized to support 1080p resolution initially with a low-resolution variant for rapid iteration. A 4K version suitable for professional production will soon follow.
This innovation empowers creative professionals to effortlessly generate B-roll footage to fill timelines or add missing atmospheric elements matching their sequences while maintaining visual continuity by locking the start and end frames of videos, ensuring consistent colors and character details.
"Firefly is designed for creatives who demand unparalleled creative control and IP-friendly tools, making it safe and efficient for production use," said David Wadhwani, President of Adobe's Digital Media Business.
Adobe emphasizes that the Firefly model is trained exclusively on content Adobe has licensed rights to use, including over 200 million images and 22 million videos from Adobe Stock, public domain materials, and no customer content. Leading brands like Gatorade, IBM, IPG Health, and marketing agencies such as Dentsu Group and Stagwell have joined the early access program for Adobe's Video Model beta.
In addition, Firefly now allows users to create professional-quality images from 3D sketches and reference shapes within the web app. The new scene-to-image feature, currently in testing, enables artists to build structural references when writing text-to-image prompts using 3D tools to craft visual guides.
For creators targeting international audiences, Adobe has introduced audio and video translation capabilities that translate spoken dialogues into multiple languages while preserving the original voice. This tool reduces the time spent on dubbing and audio mixing by maintaining consistency in voice, tone, and acoustics across translations into more than 20 languages.
To support these new models and offerings, Adobe is providing early access to Firefly at special pricing with two new subscription plans: Firefly Standard and Firefly Pro. Both offer unlimited access to Firefly image and vector generation functionalities.
The Standard plan provides users with 2000 monthly video/audio credits and up to 20 five-second 1080p video generations starting at $9.99 per month. The Pro plan increases this to 7000 monthly video/audio credits and 70 five-second 1080p video generations starting at $29.99 per month.
