On October 15th, according to Finance Associated Press (Edited by Niu Zhanlin), local time on Monday, at the Adobe MAX annual conference, the American computer software company announced that it has begun public testing of a text-based video artificial intelligence (AI) model, joining the ranks of more and more companies trying to disrupt film and television production using generative artificial intelligence. During regular trading hours, Adobe's stock price rose more than 3%.

This technology is called the Firefly video model, and Adobe describes its technology as making a new breakthrough, competing with the text-based video large model Sora launched earlier this year by OpenAI, while the owners of TikTok, ByteDance, and Meta have also released their own text-based video tools in recent months.

Faced with strong competitors, Adobe is betting its future on building models trained with data it has the right to use, ensuring that the output results can be legally used for commercial projects.

In fact, the first test version of Firefly was released in March 2023, mainly for text-generated images. Since its release, it has generated over 13 billion images.

Within a year of its launch, Firefly has been integrated into tools such as Photoshop, Express, Illustrator, Substance 3D, supporting text prompts in over 100 languages.

Now, the Firefly video model expands Adobe's existing generative AI model family. It is currently only available in limited availability to collect initial feedback from a small group of professional creatives, which will then be used to further refine and improve the model.

Adobe, headquartered in San Jose, California, will start opening the Content-Aware Video Generation model to users on the waiting list, but has not disclosed a specific date yet.

Although Adobe has not announced any customers using its video tools, on Monday it stated that PepsiCo's Gatorade under Pepsico will use its image generation model on a website to customize bottles for customers, while Mattel has been using Adobe tools to help design Barbie doll packaging.

Additionally, due to copyright and security considerations, the Firefly video model will temporarily not produce videos involving children or public figures, and the created video content will be labeled as AI-generated.

Ely Greenfield, Chief Technology Officer of Adobe Digital Media, pointed out that Adobe's video tools are designed to enable video creators and editors to use them in their daily lives, with a focus on integrating generated videos with traditional materials.

Greenfield said: "We are truly focused on fine control, teaching the model the concepts used by video editors and cameramen, such as camera position, camera angle, camera movement, etc."