Creative software developer Adobe Inc. today launched into public beta test a new generative artificial intelligence video model, which the company claims is the first commercially safe AI video creation model on the market.

It’s available in a new version of the company’s Firefly web application, which offers a suite of professional-grade creative tools, allowing creative professionals to produce artwork from anywhere, and in Adobe Premiere Pro.

The new generative video capabilities are powered by the Adobe Firefly Video Model, the company’s powerful proprietary generative AI model capable of producing video clips from text prompts or images. It can use camera angles to control shot composition, create professional-quality B-roll, craft imaginative elements from descriptions and custom motion design.

The model has been customized to support 1080p resolution from the start, with a lower resolution model for high-speed iteration and a 4K version for pro-level production launching soon.

This will allow creative professionals the opportunity to easily generate B-roll to fill in timelines or missing atmospheric elements to shots that match their reels. According to Adobe, the model will allow them to lock in the first and last frames of a video to preserve visual continuity, keep colors and character details consistent and more.

“Firefly is designed for creative professionals looking for unmatched creative control and IP-friendly tools that can be used safely and effectively in production,” said David Wadhwani, president of Adobe’s digital media business.

The company emphasized that the Firefly model was only trained on content Adobe has permission to use. This includes licensed content from Adobe Stock — which exceeds 200 million images and 22 million videos — public domain content and never customer content. Leading global brands such as Gatorade, IBM Corp., IPG Health, and marketing agencies including Dentsu Group Inc. and Stagwell Inc. have joined Adobe in early use of the Video Model beta.

Adobe also provided professionals the ability to create professional quality images from 3D sketches and reference shapes in the Firefly web app. The new Scene to Image capability, now in beta mode, allows artists to build structure references while writing Text to Image prompts to create visual guides using 3D tools.

For creators trying to reach international audiences, Adobe introduced audio and video translation of spoken dialog into multiple languages while maintaining an authentic voice. Using this tool, professionals can spend less time dubbing performance and audio mixing. It can maintain the voice, tone and acoustic match when translating content into more than 20 languages.

To support the new model and offerings, Adobe is making access to Firefly available today with early access pricing for two new Adobe Firefly plans: Firefly Standard and Firefly Pro. Users get unlimited access to Firefly image and vector generation features with both. Standard provides users with 2,000 video/audio credits per month and up to 20 five-second 1080p video generations starting at $9.99. Firefly Pro ups that to 7,000 video/audio credits per month, 70 five-second 1080p video generations and starts at $29.99.

Image: Adobe