xAI has introduced a video extension capability for its Grok video model, allowing users to lengthen existing video content. This feature generates new video segments that pick up from the final frame of an input video, guided by a text prompt describing the desired continuation.
Users can provide the source video via a public URL, a base64-encoded data URI, or a file_id from the Files API. The input video must be in .mp4 format, encoded with supported codecs such as H.265, H.264, or AV1.
Key Points
- The Grok video model extends existing videos using a text prompt.
- Input videos must be
.mp4files with codecs like H.265, H.264, or AV1. - Input video duration must be between 2 and 15 seconds.
- The extended portion can range from 2 to 10 seconds, with a default of 6 seconds.
- Output video resolution is capped at 720p, matching the input's aspect ratio and resolution.
- The
durationparameter controls only the length of the extended segment.
Context
According to xAI, the video extension functionality is part of the broader Grok model capabilities, which also include video generation, image-to-video, and reference-to-video features. The Grok model supports various input methods for source videos, including direct URLs and file_ids from its Files API, as detailed in the xAI documentation.
Why It Matters
This capability offers builders a method to programmatically extend video content, potentially streamlining workflows for dynamic content creation or automated video editing tasks. The ability to specify continuation with a text prompt could enable more flexible and scalable video production.
What To Do
- Review the xAI documentation for the Grok video model's extension feature.
- Test the video extension in the Playground with various input videos and text prompts.
- Note the constraints on input video duration (2 to 15 seconds) and extension duration (2 to 10 seconds).
- Compare the output resolution and aspect ratio to ensure it meets project requirements.
