EmptyLTXVLatentVideo - ComfyUI Built-in Node Documentation

The EmptyLTXVLatentVideo node creates an empty latent tensor for video processing. It generates a blank starting point with the specified width, height, length, and batch size, which can be used as an input for video generation workflows. The node produces a zero-filled latent representation whose spatial dimensions are 32 times smaller than the configured width and height, and whose frame count is compressed by a factor of 8.

Inputs

Parameter	Description	Data Type	Required	Range
`width`	The width of the latent video tensor (default: 768, step: 32)	INT	Yes	64 to MAX_RESOLUTION
`height`	The height of the latent video tensor (default: 512, step: 32)	INT	Yes	64 to MAX_RESOLUTION
`length`	The number of frames in the latent video (default: 97, step: 8)	INT	Yes	1 to MAX_RESOLUTION
`batch_size`	The number of latent videos to generate in a batch (default: 1)	INT	Yes	1 to 4096

Note: The latent video is compressed compared to the input dimensions: the spatial dimensions (width and height) are divided by 32, and the frame count (length) is divided by 8 and rounded up to the nearest whole number. The step values for width, height, and length help keep these divisions even.

Outputs

Output Name	Description	Data Type
`samples`	The generated empty latent tensor with zero values in the specified dimensions, along with a spatial downscale ratio of 32	LATENT

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): 0b1e57baf9730d852b03b6bccbb8a033e2be9b9cd2420a0aa3638c31f6d3cd26

​Inputs

​Outputs

Inputs

Outputs