Adobe is introducing new generative AI filmmaking tools that offer fresh methods for producing sound effects and managing video outputs. In addition to the traditional text prompts for its Firefly AI models, users can now make use of onomatopoeic (vocal sound imitation) recordings to create custom sounds and utilize reference footage for guiding actions in videos generated by Firefly. The beta version of the Generate Sound Effects tool facilitates audio creation based on recorded visuals, giving users better control than Google’s Veo 3 technology. The intuitive interface mirrors a video editing timeline, coordinating sound effects with visual content seamlessly. For instance, recording “clip clop” sounds while playing a video of a horse trotting leads to several sound effect choices based on accompanying text descriptions. This development builds upon Adobe’s Project Super Sonic, which was announced at its Max event in October.
Although it does not generate speech, the tool effectively produces impact sounds like breaking twigs and footsteps. The Firefly Text-to-Video generator is also receiving additional controls, such as the Composition Reference feature, allowing users to upload videos together with their text prompts to accurately mirror the original footage’s layout, yielding better outcomes than depending solely on text. Furthermore, keyframe cropping enables the uploading of initial and final frame images to assist in video generation, and new style presets offer a variety of visual styles for quick selection, limited though they may be to the Firefly video AI model. Despite showing areas for enhancement, such as a “claymation” style reminiscent of early 2000s animations, Adobe continues to bolster support for various AI models. Alexandru Costin, the Generative AI lead, suggested that similar capabilities might be extended to third-party AI models, highlighting Adobe’s determination to lead in creative software while navigating competition from OpenAI and Google.
The ainewsarticles.com article you just read is a brief synopsis; the original article can be found here: Read the Full Article…
