In late 2025, Layer incorporated the Kling O1 Video Edit model, which is a new way of modifying specific areas of a video. This requires specific wording and inputs in order to modify a video in the way the user desires. We have added it to our Workflows as an experimental Edit Video node so that our users may incorporate it into their workflows.
As of January 2026, there is only one model available for Edit Video nodes: Kling O1.
To set up an Edit Video node, you will want to upload a video or pass in videos into the node. Kling O1 video edit can only edit videos, so you will need a Create Video node or a reference video node attached to it to edit.
The best way to use the Video Edit node is to include one or more images as reference groups, then reference them in the prompt. Kling O1 expects you to reference these groups and their images as @Image1 or @Image2 when writing prompts, and you can incorporate them into natural language when making your prompt. Here's an example:
"Take the following video and replace the running character with @Image1. Then, replace the car they are chasing with @Image2. Keep all other elements of the video the same."
This prompt would replace a running character in the video with whatever was uploaded in the first reference group. It would then replace a car in the video with whatever was uploaded in the second reference group. You can use this Video Edit node after a video has been generated, but there are limits to the length of the video: A video must be at least 3s long and cannot be longer than 12s, so you may need multiple video edit nodes to modify the entire contents of a 30s or 60s stitched video.