Grok: Video generation
Input text prompts or an initial image frame to generate up to 15-second videos with synchronized audio using the Grok model.
About
This ComfyUI workflow generates short videos—up to 15 seconds—using the Grok model, with an automatically synchronized audio track. At its core is the GrokVideoNode, which accepts either a pure text prompt (text-to-video) or a starting frame from LoadImage (image-to-video). The node handles inference and returns a ready-to-save video clip that pairs the visuals with audio produced by the model. SaveVideo then writes the result to disk as a standard video file, preserving the embedded audio when present.
What you provide
- Reading the workflow
Runs in ComfyUI. This workflow can't run on this page. Open it in Comfy Cloud with everything loaded, or download it for your own ComfyUI.
Try in CloudQuestions & answers
For text-to-video, enter your prompt in GrokVideoNode and leave LoadImage disconnected. For image-to-video, load a still in LoadImage and connect it to GrokVideoNode so the model uses the image as the starting frame.
This workflow is designed for clips up to 15 seconds, matching the Grok model’s intended range. Keep your duration at or under 15 seconds for reliable results.
GrokVideoNode produces a video with synchronized audio, and SaveVideo preserves it when saving. If you need custom audio, export the video and replace the soundtrack in your video editor. The provided workflow does not include an audio import/replace node.
If GrokVideoNode exposes a seed or randomness control, set and reuse the same value across runs. Also keep prompts, duration, and the starting image (for image-to-video) unchanged to maximize repeatability.