Adobe invites you to "adopt technology" with the new Firefly video generator.

Adobe has introduced video generation capabilities on its Firefly AI platform just before its Adobe MAX event. Starting today, users can first experiment with Firefly's video generator on Adobe's website or try out the new Generative Extend feature, which utilizes artificial intelligence, in the beta version of Premiere Pro. On the Firefly site, users can access text-to-video or image-to-video models, both capable of producing clips of up to five seconds of AI-generated video. Access to this web beta version is free, although it may have usage limits.

The company has trained Firefly to create animated content and photo-realistic media, depending on the specifications of the instructions. Additionally, Firefly can generate videos that include text, an area where AI image generators have faced challenges. The Firefly web app features settings to adjust camera angles, movement intensity, shot angles, and sizes. Meanwhile, in the beta version of Premiere Pro, users can explore the Generative Extend feature, which allows them to extend video clips by up to two seconds, generating an additional beat in a scene while continuing the camera and subject movements. The background audio will also be extended, marking the first public glimpse of the AI audio model Adobe has been working on.

In pre-launch tests, Firefly's Generative Extend feature showed more impressive results compared to the text-to-video model, standing out for its practicality. Although the text-to-video and image-to-video models do not seem to have the same level of quality as offerings from competitors like Runway and OpenAI, Adobe has focused its efforts on AI editing features rather than video generation as a way to cater to its user base.

It is crucial for Adobe's AI features to maintain a delicate balance with its creative audience, as the company aims to lead in a market saturated with new startups and advanced AI technologies. However, many creatives are concerned about the possibility of these tools displacing their traditional work. In response, the Generative Extend feature uses AI to solve existing problems in video editing rather than creating new clips from scratch.

Adobe has taken measures to be mindful of creatives' concerns, compensating photographers and artists with $3 for every minute of video they contribute to train its Firefly AI model. Nevertheless, there remains skepticism about the use of AI tools. In defense of these technologies, it is noted that the demand for personalized and individualized content in businesses is growing, encouraging creators to leverage these tools to boost their productivity without compromising their passion for creativity.

Firefly will also have a system to automatically insert watermarks in the metadata of the generated videos, indicating that they are AI-produced. Despite this, the videos will not display visible labels that clearly identify them as such to the public. Adobe has designed Firefly to generate media considered "commercially safe," meaning it avoids creating content that includes elements like drugs, violence, political figures, or copyrighted materials. With free access to this video model, it will be interesting to observe its performance in practice.