Get answers to frequently asked questions about using Generative Extend in Adobe Premiere Pro.
Video
These requirements are on the source media.
Video source media resolution:
- 16:9 Aspect Ratio (Landscape/Standard):
- Minimum: 640×360 (360p)
- Maximum: 3840×2160 (4K UHD) or 4096×2160 (4K Cinema)
- Native resolutions supported by AI Model: 1280×720, 1920×1080, 3840×2160, 4096×2160
- 9:16 Aspect Ratio (Portrait/Vertical):
- Minimum: 360×640 (360p portrait)
- Maximum: 2160×3840 (4K portrait) or 2160×4096 (4K Cinema portrait)
- Native resolutions supported by AI Model: 720×1280, 1080×1920, 2160×3840, 2160×4096
When your clip's resolution doesn't match the natively supported resolutions, Generative Extend will automatically generate the extension at the nearest higher supported resolution. Your original clip remains unchanged.
Video Source Media Frame Rate and Bit Depth:
- 12-60 fps
- Bit Depth: Source clips can be 8-bit, 10-bit, or 16-bit, but extensions are generated in 8-bit when the sequence is set to 8-bit
- Color Space: Source clips can be SDR or HDR, but extensions are generated in SDR when the sequence is set to SDR
- Your original clip remains unchanged in its native bit depth and color space
Audio
- Generative Extend cannot create or extend spoken dialog. During extension, existing dialog will be muted to preserve the integrity of the original speech content.
- Due to the complexity of musical structures and potential copyright concerns, clips containing music are not eligible for extension.
- Currently, only mono and stereo audio are supported. Surround sound and 5.1 audio formats are not compatible with Generative Extend.
Video clips can be extended up to 2 seconds, while audio can be extended up to 10 seconds.
Generative Extend is currently unavailable to some users in the following circumstances:
- Generative Extend is powered by Adobe Firefly, which isn't available in Russia, Belarus, and China. Users in these countries cannot access this feature.
- Users who are part of K-12 and educational organizations may experience limited or no access to Generative Extend.
- Some enterprise users, particularly those with a CCE v3 license, don't have Firefly services enabled. Therefore, they will not be able to access Generative Extend.
This dialog appears when your clip’s resolution doesn’t match one of the native resolutions that the Generative Extend AI model supports.
Selecting Continue will generate the extension at the next highest supported resolution. Your original clip remains unmodified.
The Generative Extend model natively supports the following resolutions:
- 16:9 Aspect Ratio (Landscape/Standard): 1280×720, 1920×1080, 3840×2160, 4096×2160
- 9:16 Aspect Ratio (Portrait/Vertical): 720×1280, 1080×1920, 2160×3840, 2160×4096
This dialog appears when your clip's framerate exceeds 30 fps. The Generative Extend model natively generates extensions between 12-30 fps.
Selecting Continue will generate the extension at 30 fps. Your original clip remains unchanged.
This dialog appears when your clip's bit depth is higher than what the Generative Extend AI model natively supports. The AI Model generates extensions in 8-bit format.
Selecting Continue will generate the extension in 8-bit format. Your original clip remains unchanged.
This dialog appears when more than one of your clip's specifications (resolution, framerate, bit depth, and/or dynamic range) isn’t compatible with Generative Extend’s supported formats.
Selecting Continue will generate the extension using the closest compatible format that the AI model supports. Your original clip remains unchanged.
HDR (High Dynamic Range) content typically uses 10-bit or higher bit depths to capture extended color ranges and brightness levels. You can use Generative Extend on HDR content when your sequence is set to SDR; however, in the extended portion:
- You'll lose the extended color range (wide color gamut).
- Dynamic range information will be compressed.
- The extended portion will be in standard dynamic range (SDR).
- Color and brightness matching may not be perfect.
It depends on your project needs and where you'll use the final video.
Extend if:
- Your content is for web, social media, or standard broadcast.
- Color precision isn't critical for your project.
- Check your output requirements: If your final delivery is 8-bit anyway, conversion won't impact your end result.
Consider alternatives if:
- You're working on a professional color-graded project.
- The clip has critical color work that can't be compromised.
Quick decision check: If your final delivery format is 8-bit anyway (which covers 90% of content), the conversion won't impact your result.
Bit depth determines how many colors your video can display. Higher bit depths (like 10-bit or 16-bit) can show more subtle color gradations and are often used in professional video production. 8-bit is the standard for most consumer video and web content.
Your original clip remains unchanged in its original bit depth, while the AI-generated extension is created in 8-bit. This may cause color banding in smooth gradients (like sunsets or skin tones) within the extended portion.
Throughout our extensive testing, we’ve succeeded with all sorts of clips, including humans, animals, landscapes, and more. Video clips that seem to work best have less contextual change in the shot, such as a fast-sweeping camera movement. Audio clips with consistent noise or ambiance also perform well.
Clips that do not work well yet include archival clips with a lot of film grain or other noise. We look forward to your feedback to learn more.
Generative Extend can add up to 2 seconds for video content and up to 10 seconds for audio content. This means you can extend your video clips by an additional 2 seconds, while audio elements like room tone or background sounds can be stretched for up to 10 seconds.
Generative Extend is a premium generative AI feature in Premiere Pro. Premium generative AI features are more computationally intensive (currently, audio and video generative features) and are accessible through Adobe Firefly standalone plans and select Creative Cloud plans. For a limited time, you can try Generative Extend with complimentary generations, after that, generative credits will be required.
Your Creative Cloud membership includes assistive AI features inside Premiere Pro, including Caption Translation, Enhance Speech, Speech to Text, and many others.
While Generative Extend is primarily designed for raw footage that has been shot, it can technically work on clips with text overlays or graphics. However, it's important to note that this isn't its intended use case. Using it on such clips might produce interesting and potentially unexpected results.
Yes, Generative Extend can work with slow-motion and time-lapse footage.
Generative Extend is designed to analyze and understand camera movements in the original clip and will continue these movements in the extended portion. This means if your clip has a pan, tilt, zoom, or other camera motion, the AI will attempt to seamlessly extend this movement in a natural way.
Each extension is saved as an H264 MP4 or .wav file in the Captured and Generated scratch disk assigned in your project settings. To set this location open Project Settings > Scratch Disks. Here, you can specify a new location for your generated media, allowing you to manage your storage according to your workflow and space requirements.
At any point, if you want to replace the AI-generated frames with original media, right-click on the AI-generated label and select Revert to Original. You can also Unlink the audio and video and revert either of them to the original media if the original media exists in the project.
No, dialogue isn't extended. During the extension process, dialogue is muted. Generative Extend only extends the room tone, sound effects, and background audio. This approach helps maintain the integrity of the original dialogue while allowing for seamless extension of the ambient audio elements.
Generative Extend takes a short segment (typically between 1.5 - 3 seconds) of your clip and sends it to a cloud AI model. The AI analyzes this segment and creates an extension based on it. The tool then returns this AI-generated extension as a new clip for you to use in your project. Your original media remains untouched.
No, your media isn't training Adobe's AI model. We understand the importance of your content and privacy. Generative Extend uses your clip solely to create the extension you've requested without using it to improve the AI. Learn more about our generative AI commitments.
If you don't like the result, don't worry - you can easily regenerate your extensions. Right-click on the AI-generated label and select Regenerate to create a new variation. You can do this multiple times until you're satisfied with the outcome. Also, we'd greatly appreciate it if you could provide feedback on whether the result is good or bad. Your input helps us improve the tool and ensure it meets your needs better in the future.
Yes, you can generate multiple variations with Generative Extend. This feature allows you to explore different possibilities for your extended content. To create a new variation, right-click on the AI-generated label and select Regenerate. You can repeat this process to produce several unique extensions, giving you more creative options to choose from for your project.
Right-click on the AI label and select Good result or Bad result. In the dialogue box that appears, you can provide the reason why the result was good or bad. This direct feedback helps us improve the tool and better understand your needs. It's a quick and easy way to share your experience and suggest improvements directly within the software.
Craft the perfect story with Premiere Pro
Find the best-in-class video-editing tools all in one place.
