A Guide to Generative Extend in Premiere Pro (beta) | Adobe Video x @filmriot
Education
Introduction
As editors, we often face the challenge of needing just a few extra frames or seconds to make our projects flow better, whether for smooth transitions or to fill in gaps with audio ambiance. This issue is common across nearly all projects, which is why Adobe has introduced the Generative Extend feature in the newest Premiere Pro beta. This innovative tool allows you to generate additional video or audio frames seamlessly, revolutionizing the editing process. Let's take a closer look at how it works.
Getting Started with Generative Extend
To use Generative Extend, your clips should be either 1920 x 1080 or 1280 x 720, and they cannot exceed 30 frames per second. If you're working with 4K footage, you'll need to export it down to 1080p before importing it into your sequence. In the toolbar, you'll find the new Generative Extend icon located under the text tool. By tapping on it, you're able to extend your video clip by a few seconds, and a processing bar will appear as the tool generates additional frames.
Upon testing, we found that while results can vary, clips with less dynamic change tend to work better, such as slow-motion shots or time lapses. For instance, a slow-motion video of someone running can be extended easily with minimal noticed shifts during transitions.
Performance Insights
The tool does have some limitations. Using it on clips containing a lot of grain or complex motion, like a running dog, may yield less satisfactory results. You'll notice where details fade or flickering occurs, but the overall performance is impressive for something still in beta. As a general rule, the longer you use the feature, the higher the chance of encountering AI artifacts, particularly in footage with faces.
If the initial result doesn’t meet your expectations, you can simply right-click the AI-generated label to regenerate a new version. Adobe also allows you to provide feedback on the output quality to help improve the tool over time. For easy access, extended files are stored in a subfolder called Generative Assets within your project folder.
Audio Enhancements with Generative Extend
Where Generative Extend truly excels is in audio applications, particularly for extending room tone, ambient sounds, or sound effects. This feature understands the context of the audio to generate seamless extensions. For example, if a car pass-by sound cuts off prematurely, the extend tool can effectively generate the trailing noise of the engine fading into the distance. The same applies to other sounds, like ocean waves, ensuring they aren't abruptly cut short.
However, there are a few restrictions to note for audio: clips must be a minimum of 3 seconds long, and those containing music are not eligible for extension. Importantly, generative extend will mute spoken dialogue and focus solely on extending the room tone.
Adobe has clearly stated that none of your media will be used to train the Firefly model, ensuring that your content remains private while using this feature.
Conclusion
The introduction of the Generative Extend feature marks a significant advancement in editing capabilities, enabling creators to refine their projects without the interruptions of traditional editing limitations. While not flawless, the rapid development and performance of this beta tool hint at its potential to become an essential aspect of our editing toolkit. If you're eager to try out Generative Extend, be sure to access it through the beta version of Premiere Pro.
Keyword
Generative Extend, Premiere Pro, Adobe, editing, video, audio, frames, room tone, ambient sounds, AI artifacts, beta version.
FAQ
Q1: What is the Generative Extend feature in Premiere Pro?
A1: Generative Extend is a new tool in the beta version of Premiere Pro that allows users to generate additional frames for video and extra seconds for audio, helping to create smoother transitions and fill gaps in edits.
Q2: What are the requirements for using Generative Extend on video clips?
A2: Clips must be either 1920 x 1080 or 1280 x 720 and cannot exceed 30 frames per second.
Q3: Are there limitations on the audio clips for Generative Extend?
A3: Yes, audio clips must be at least 3 seconds long, and the feature will not extend music or spoken dialogue.
Q4: How can I provide feedback on the output of Generative Extend?
A4: You can right-click the AI-generated label and select options for feedback, indicating whether the output was good or poor.
Q5: Will Adobe use my media to train its AI models?
A5: No, Adobe ensures that none of your media will be used to train its AI models, maintaining user privacy.