AI Video Edit by Sora in Premiere Pro

Since its debut on Mac in late 1991 with “Premiere,” Adobe’s Premiere Pro has long been one of the most popular video editing software in the world, widely adopted by film editors and independent filmmakers globally. Now, as generative AI technology rapidly advances, Premiere Pro is set to undergo its most significant transformation in its 33-year history.

Adobe recently announced an update to Premiere Pro that will incorporate plugins from emerging third-party AI video generation models, including OpenAI’s Sora, along with competitors Runway ML Gen-2 and Pika 1.0.

This integration not only brings these AI tools to a broader audience but could also mark a revolutionary shift in the video production landscape if a substantial segment of Adobe’s 33 million Creative Cloud subscribers embraces these new capabilities.

With these enhancements, Premiere Pro users will be able to edit and process live-action footage captured with traditional cameras and seamlessly blend it with AI-generated content. For instance, imagine shooting a scene where an actor flees from a monster and using AI technology to generate the monster itself—no props or costumes needed. Both pieces of footage can be edited and merged within the same editor into a single video file. This applies equally to more sophisticated animation processes (combining pure computer-generated animation with frame-by-frame drawing), where animated content can easily be blended with AI materials provided in Premiere Pro.

As Adobe points out on their official website: “Early explorations suggest that professional video editors could enrich their projects by incorporating rich content outputted by the OpenAI and Runway video generation models integrated into Premiere Pro. We’ve also observed that Pika Labs can be used in conjunction with Generative Extend tools to add several seconds of extended footage after a shot finishes.”

Adobe explains this exciting concept further: “Looking ahead, we envision thousands of specialized models emerging, each excelling within its niche market. Decades of AI application experience at Adobe show that seamlessly integrating generated content into everyday workflows unlocks the greatest value of this technology. For most Adobe customers, generative AI will become a new starting point and source of inspiration for creative exploration.”

Adobe has also released a preview video showcasing the video editing workflow with these third-party AI video generators.

No timeline or specific release details have been confirmed yet…

Adobe has not yet announced when these third-party AI video generators will be integrated into Premiere Pro, and the details appear to be still under discussion, particularly since many third-party tools involve subscription fees. There are few early preview videos currently available, with none from Sora.

Moreover, Adobe mentioned its own internal generative AI products, such as Firefly and Generative Fill, emphasizing that their models are trained on data owned or licensed by Adobe, including contributions from Adobe Stock images.

This ties back to Adobe’s AI indemnity pledge, focusing on positioning Adobe as a trustworthy provider of AI tools and products, helping enterprise clients mitigate concerns about potential legal issues and widespread criticism from some artists and creators regarding AI applications.

It is speculated that third-party tools used in Adobe Premiere Pro might not be covered by Adobe’s indemnity pledge. However, Adobe has indeed stated, “We are committed to providing content credentials— including labeling for free, open-source technologies used in online content creation, clearly marking assets generated by AI applications. This allows users to track the content creation process and understand which AI models are used to create video content on the Adobe platform.”

Just last week, Bloomberg reported that Adobe had trained its own Firefly using images from Midjourney, a competing AI art generator that relies on the open-source AI model Stable Diffusion, which uses publicly scraped and copyright-protected data for training.

As of now, Firefly is still applicable only for video generation

Recently, Adobe also announced a text-to-image version of the Firefly generator model, expected to arrive in Premiere Pro later this year, establishing a new “generative AI workflow” and feature set within the software.

This includes “Generative Extend,” allowing video editors and filmmakers to seamlessly add new frames to extend clip durations without reshooting any footage—a potentially significant cost-saving feature. This technology also helps achieve smoother transitions and prolonged endings on specific moments or actions.

Firefly for Video will also enhance “object detection and removal,” highlighting specific objects in videos (including props, characters, costumes, scenes, etc.) and allowing AI models to continuously track them across different frames. Based on this, users can use generative AI to edit these objects into new ones, quickly changing a character’s costume or props, or even completely removing the object across multiple video clips and camera angles.

Additionally, Firefly for Video will include a text-to-video image generator, claimed to compete with Sora, Runway, Pika, and Stable Video Diffusion. However, it remains to be seen how this model will perform in accurately replicating user prompts compared to more established AI video generators.

Film creators and creative professionals express enthusiastic approval

Although only a preview version is currently available, Adobe’s new generation of AI integration features crafted for Premiere Pro has already received a warm reception among filmmakers and social media creatives, especially those who have already experimented with AI in video production.

Jason Zada, filmmaker and founder of Secret Level Studio, commented on LinkedIn: “If this technology proves effective, AI will help every practitioner enhance their efficiency.”

Film director Kevin K. Shah also replied to the same post, stating: “This will be really beneficial for my live-action filmmaking.”

AI industry leader and former member of the Google Maps AR/VR team, Bilawal Sidhu, posted on X (formerly Twitter), noting that the introduction of third-party AI video models into Premiere Pro is “definitely something to look forward to for creatives, as AI-generated models have been producing some truly amazing content and it’s absolutely necessary to incorporate this new capability into video editing tools.”

Leave a Reply

Your email address will not be published. Required fields are marked *

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.