Adobe enhances Firefly Video with new features, partner models, and AI sound tools

Adobe Firefly

Adobe has unveiled major upgrades to its Firefly Video Model, introducing improved motion fidelity, advanced video controls, and integration with leading generative AI partners — all aimed at giving creators more precision, flexibility, and speed across their workflows.

It brings smoother transitions and cinematic flow, ideal for storyboarding or full-length videos.

One standout feature is Composition Reference for Video, which lets users upload a reference clip and have Firefly generate new visuals that mimic its structure — streamlining content repurposing and multi-scene continuity.

Adobe has also added Generate Sound Effects (beta), allowing users to create custom audio from text prompts or even their own voice. This tool adapts to vocal rhythm and energy, placing sounds with cinematic timing to match visual actions. The sound generation is commercially safe, like all Firefly tools, trained only on licensed content.

In a major ecosystem expansion, Runway Gen-4 Video, Google Veo3 with Audio, Luma AI’s Ray 2, and Pika 2.2 are now integrated into Firefly, giving users a choice of models based on their creative needs. Upcoming additions include Topaz Labs’ upscalers and Moonvalley’s Marey.

Firefly now supports multiple aspect ratios — vertical, square, widescreen — tailored for platforms like TikTok or YouTube, without additional editing.

Also in beta is Text to Avatar, turning written scripts into avatar-led videos for presentations or educational content — a big win for content creators and marketers.

To simplify prompting, Adobe has added Enhance Prompt, which automatically refines user inputs to yield more accurate results without the guesswork.

All videos can be exported directly to Adobe Express for social-ready content or Premiere Pro for deeper editing.