Runway Launches Expand Video Feature: How to Expand Videos on Runway?
The video generation AI company RunwayML has introduced the "Expand Video" feature. Users can generate additional content based on the original footage by inputting text prompts, allowing for flexible adjustments to the video aspect ratio. The system is capable of maintaining visual consistency while expanding the frame.
With multiple expansions, users can achieve dynamic camera effects such as "crash zoom" and "pull-back shot," transforming static images into dynamic sequences with a cinematic feel.
It is important to note that the "Expand Video" feature is currently being gradually made available to Gen-3 Alpha Turbo users.
Expand Video in Runway
The "Expand Video" feature launched by Runway allows users to extend video frames while maintaining visual consistency. This tool supports text prompts and reference images, enabling creative expansion in new directions. Users can chain multiple expansions to achieve dynamic camera effects like "crash zoom" and "pull-back shot," turning static images into dynamic cinematic sequences.
Expand Video reimagines and reconstructs existing videos into different formats. Moreover, Expand Video does not delete details; instead, it adds content to the edges of the input video to create new formats.
Choosing the Right Video Material
When creating videos, we need to select a guiding image that matches the text prompt, with dimensions of at least 620 pixels wide and 620 pixels high, and an aspect ratio close to traditional formats.
This image should ensure that the subject is centered in the frame to highlight it. Additionally, to maintain the image's clarity and focus, there should be no unnecessary graphics or text elements that could distract the viewer. This helps convey core information better and enhances the visual effect.
How to Use Expand Video on Runway for Creation?
- Select the Video to Expand
First, navigate to the "Generate Video" section in the Runway Dashboard, then select "Expand Video" from the left toolbar. Next, choose the video you want to expand. The expand video feature will reconstruct the input video into either horizontal or vertical formats. Selecting an input video with a width greater than height will produce vertical output, while height greater than width will yield horizontal output. For square input videos, you can choose either vertical or horizontal output.
- Configure Expand Video Prompts
- Using Text Prompts: When writing prompts for Expand Video, describe what we would like to see in the expanded area. While the input video will automatically guide most of the scene details, we can briefly reference it to ensure visual consistency. However, we should avoid detailing the existing video's content.
- Using Guiding Images: The best results will come from using an expanded image of the input video’s first frame as a guiding image. When guiding generation with an input image, there is no need to include text prompts.
- Generate and Reiterate Output
Before starting generation, we can hover over the "Duration" icon to view the estimated credit cost. After confirming the credit cost, click "Generate" to begin processing the expansion video creation.
Conclusion
To enhance user creation experiences, major platforms are continuously innovating and launching new features. Runway's launch of Expand Video and XXAI's integration of multiple popular AI models provide users with comprehensive solutions.
When creating with the Expand Video feature, tall videos typically produce horizontal output, while wide videos yield vertical output, and square videos allow for flexible output size selection.
In most cases, videos can be smoothly expanded without prompts. However, for more control, we can add text or image prompts. If using a guiding image, it is best to use an image reference from the first frame of the video for optimal results. Using a guiding image that does not fully match may result in brief visual jumps.
Of course, we can also achieve multiple video expansions by clicking the dropdown menu next to "Reuse Settings" and selecting "Expand Video" on previously expanded outputs to transform static images into dynamic visuals.