KlingVideoNode - ComfyUI Built-in Node Documentation

This node generates videos using the Kling V3 model. It supports two primary modes: text-to-video, where a video is created from a text description, and image-to-video, where an existing image is animated. It also offers advanced features like creating multi-segment videos with different prompts for each part (storyboards) and optionally generating accompanying audio.

Inputs

Parameter	Description	Data Type	Required	Range
`multi_shot`	Controls whether to generate a single video or a series of segments with individual prompts and durations. When not “disabled,” additional inputs for each storyboard’s prompt and duration appear.	COMBO	Yes	`"disabled"` `"1 storyboard"` `"2 storyboards"` `"3 storyboards"` `"4 storyboards"` `"5 storyboards"` `"6 storyboards"`
`generate_audio`	When enabled, the node will generate audio for the video. Default is `True`.	BOOLEAN	Yes	`True` / `False`
`model`	The model and its associated settings. Selecting this option reveals the `resolution` and `aspect_ratio` sub-parameters.	COMBO	Yes	`"kling-v3"`
`model.resolution`	The resolution for the generated video. This setting is available when the `model` is set to “kling-v3”.	COMBO	Yes	`"4k"` `"1080p"` `"720p"`
`model.aspect_ratio`	The aspect ratio for the generated video. This setting is ignored when an image is provided for `start_frame` (image-to-video mode). Available when the `model` is set to “kling-v3”.	COMBO	Yes	`"16:9"` `"9:16"` `"1:1"`
`seed`	A seed value for generation. Changing this value will cause the node to re-run, but the results are non-deterministic. Default is `0`.	INT	Yes	0 to 2147483647
`start_frame`	An optional starting image. When connected, the node switches from text-to-video to image-to-video mode, animating the provided image.	IMAGE	No	-

Inputs for multi_shot mode:

When multi_shot is set to “disabled”, the following inputs appear:
- prompt (STRING): The main text description for the video. Required. Must be between 1 and 2500 characters.
- negative_prompt (STRING): Text describing what should not appear in the video. Optional.
- duration (INT): The length of the video in seconds. Must be between 3 and 15. Default is 5.
When multi_shot is set to a storyboard option (e.g., "3 storyboards"), inputs for each storyboard segment appear (e.g., storyboard_1_prompt, storyboard_1_duration). Each prompt must be between 1 and 512 characters. The total sum of all storyboard durations must be between 3 and 15 seconds.

Constraints:

The node operates in text-to-video mode when start_frame is not connected. It uses the model.aspect_ratio setting in this mode.
The node operates in image-to-video mode when start_frame is connected. The model.aspect_ratio setting is ignored. The input image must be at least 300x300 pixels and have an aspect ratio between 1:2.5 and 2.5:1.
In storyboard mode (multi_shot not “disabled”), the main prompt and negative_prompt inputs are hidden and not used.

Outputs

Output Name	Description	Data Type
`video`	The generated video file.	VIDEO

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): c329fa5a8609ccd2caec96fc23afae9ba79bdf835faa05450d4e20a7380087f5

​Inputs

​Outputs

Inputs

Outputs