Adobe Firefly adds AI sound, avatars, and motion tools for video creation

Adobe enhances Firefly with AI-generated sound effects, avatars, improved motion fidelity, and new editing tools to simplify and speed up video creation workflows

Adobe has rolled out a major update to its Firefly video model, introducing features aimed at giving creators more control and flexibility. The update enhances motion fidelity, introduces AI-generated sound effects, supports avatar-led video creation, and expands format support for streamlined video production. New partner AI models are also being integrated to boost capabilities across creative workflows.

Smoother motion and flexible formatting

Among the most significant upgrades is improved motion fidelity. The Firefly Video Model now delivers more natural and fluid movement – particularly in scenes involving nature, weather dynamics, and animation (2D and 3D).

 

The update also makes it easier to switch between vertical, horizontal, or square formats to suit different platforms and publishing needs.

 

Smarter tools to speed up workflows

Adobe is adding several features to simplify the video editing process:

• Composition reference for video: Users can upload a reference video and describe their vision; Firefly will generate a new clip that mirrors the layout and style.

• Style presets: Choose from predefined aesthetics such as anime, claymation, or line art to apply a visual tone instantly and consistently.

• Keyframe cropping: Upload the first and last frame with the desired crop to have Firefly auto-generate a video to match the specified format—without switching apps.

AI sound and avatars

New tools allow users to bring audio and characters into videos using simple prompts:

• Generate sound effects (beta): Users can create custom sounds—from roars to ambient nature—using text or voice prompts. Firefly syncs them to videos with cinematic timing. The output can be exported to Adobe Express or Premiere Pro.

• Text to avatar (beta): This feature turns written scripts into avatar-led videos. Users can select from a library of avatars, customise the background, and choose different voice accents.

Expanded model integration

Adobe is also expanding Firefly’s capabilities by integrating partner generative AI models into the platform:

• Coming soon to Firefly Boards:
• Topaz Labs: Image and video upscalers
• Moonvalley’s Marey

Coming soon to generate video:

• Luma AI’s Ray 2
• Pika 2.2 (currently available in Boards)