At the Adobe Max event, Adobe introduced new video creation tools: Text-to-video, Image-to-Video and Generative Extend. Among them is the ability to create videos from text or images on Firefly. It works as an application as well as integrated into the Premiere Pro editing software. Adobe said this is the first video model designed to be safe for commercial use.
Generative Extend for Premiere Pro is reportedly in beta, allowing users to extend clips by up to 2 seconds in 720 or 1080 mode at 24 FPS.
Additionally, it can be used to enhance audio, extend sound effects and ambient noise by up to 10 seconds, although it does not support dialogue or music. Overall, it is designed for minor editing in video and audio.
The two Text-to-video, Image-to-Video tools are now available in limited version in the Firefly web app.
Text-to-video works similarly to other video engines like Runway and OpenAI's Sora, allowing users to input text descriptions to generate videos. It can simulate a variety of styles including traditional film, 3D animation, and stop-motion.
Image-to-video improves the video creation process by allowing users to include reference images along with text prompts, giving them more control over the output.
Adobe envisions this feature being useful for creating b-roll from photos or visualizing potential footage by uploading stills from existing videos.
All three tools take around 90 seconds to generate their output, but Adobe is working on ways to speed up the process. Adobe is emphasizing the tools are commercially safe, in contrast to other offerings like Runway (which has come under scrutiny for allegedly being trained on a large number of scraped YouTube videos) and Meta, which may have used user videos without their consent.
Adobe doesn't currently charge for AI features beyond the standard software purchase price, but it's possible that its advanced features will be charged at a higher price in the future.
Source: https://kinhtedothi.vn/adobe-ra-mat-cong-cu-tao-video-moi.html
Comment (0)