Adobe has recently released its AI video generator for public beta after its limited early testing period last year. Users can access Adobe’s Firefly Video Model on the Firefly web app page, which also features Adobe’s other generative AI tools.
The Firefly Video Model features two ways to generate AI videos. The first is text to video, which allows users to input a description of the video they wish to generate. The second feature is the image to video tool that enables users to upload a reference image in addition to a text prompt.
The videos take around 90 seconds to generate and have a maximum length of five seconds. Adobe has also improved the output quality from 720p to 1080p. According to Adobe, there are plans for a faster ideation model in the future.
In addition to the public beta release of the video generator, Adobe has updated its Firefly web app to include integration with Creative Cloud apps. This revamp makes it more convenient to create, edit, and transfer generated content.
Paying users also have access to additional tools, namely the scene to image and translator tools. Naturally, Adobe has included two new Firefly subscription plans, Firefly Standard and Firefly Pro. Firefly Standard currently costs US$9.99 (~RM44) a month and allows the user to generate up to 20 five-second 1080p videos while Firefly Pro is priced at US$29.99 (~RM134) a month for up to 70 five-second 1080p clips.
Currently, the AI video market is ripe with competition. Among Adobe Firefly’s competitors is OpenAI’s Sora, which can generate clips with a length of up to 20 seconds. Meanwhile, Google has claimed that its own generative AI video model, Veo, can produce consistent and coherent videos that depict realistic movement.
Adobe claims that Firefly is the only IP-friendly and commercially safe generative AI model as it was trained using material acquired with permission. This may give Firefly an edge over its rivals among the more conscientious users who wish to avoid issues with copyright.
(Source: Adobe [1], [2], The Verge)
Nurul Kamil contributed to this article.
