Grok Imagine Model Feature Updates
Introducing Grok Reference-to-Video and Grok Video Extend Nodes
Now available with Partner Nodes are two new features for the Grok Imagine Video model! The Grok Reference-to-Video node allows up to 7 reference images and generates consistent videos using those inputs. The Grok Video Extend node takes an input video of up to 15 seconds and extends the video up to an additional 10 seconds while maintaining coherent motion and audio.
The speed of these models is very impressive and allows for quick generation and iteration. Using Grok Video Extend at 10 seconds, the generation takes ~60 seconds. With Reference-to-Video, 10 second generation times at 720p take ~100 seconds. Check out the examples below to see both of the new features in action!
Get Started
Update ComfyUI to the latest version or visit Comfy Cloud
Find the Grok Reference-to-Video or Grok Video Extend nodes in the Node Library or Templates in the left sidebar.
Drop a node on your canvas and start creating.
 | A guest post by
|
Do we have to do a guess work how much it cost to run GROK? I can't see it at all. I run it locally.