Adobe has announced new upgrades for its Firefly video model, aiming to deliver smoother and more natural motion generation. The enhancements include improved transitions for both 2D and 3D elements, creating lifelike motion for characters as well as objects like bubbles, leaves, and clouds.
New AI Models and Tools
Alongside this upgrade, Adobe has added four new third-party AI models to Firefly Boards. These include Topaz Labs’ Image and Video Upscalers and Moonvalley’s Marey. Luma AI’s Ray 2 and Pika 2.2, which were already part of the platform for image generation, will now gain video generation capabilities. These tools are designed to enhance creativity and offer more flexibility to users.
A notable addition is the ability to upload a video as a composition reference. This allows Firefly to follow the structure and format of the uploaded video, producing a more consistent output. Users can also now apply Style Presets such as claymation, anime, 2D, or line art, ensuring the generated content matches the desired aesthetic.
Keyframe cropping is another new feature, enabling users to upload the first and last frames of a video so that Firefly can create content that aligns with the chosen format and aspect ratio.
Generate Sound Effects and More
Adobe has also introduced a beta feature called Generate Sound Effects, which allows users to create custom audio tracks using voice or text prompts. Users can control timing and intensity to match the energy of the video, ensuring a seamless blend of visuals and sound.
In addition, Adobe’s new Text to Avatar feature converts scripts into avatar-based videos. Users can choose from pre-listed avatars, customize backgrounds, and even select speech accents, making content creation faster and easier.
These new features, combined with Firefly’s improved motion fidelity, strengthen Adobe’s position in the AI-powered video generation space. While some tools will initially be available only for paid subscribers, they promise to significantly streamline video creation for professionals and creators alike.
Also Read: OpenAI’s ChatGPT Agent Can Now Do Complex Tasks Alone
