EmptyFlux2LatentImage - ComfyUI Built-in Node Documentation

The EmptyFlux2LatentImage node creates a blank, empty latent representation. It generates a tensor filled with zeros, which serves as a starting point for the Flux model’s denoising process. The dimensions of the latent are determined by the input width and height, scaled down by a factor of 16.

Inputs

Parameter	Description	Data Type	Required	Range
`width`	The width of the final image to generate. The latent width will be this value divided by 16. The default value is 1024.	INT	Yes	16 to 8192
`height`	The height of the final image to generate. The latent height will be this value divided by 16. The default value is 1024.	INT	Yes	16 to 8192
`batch_size`	The number of latent samples to generate in a single batch. The default value is 1.	INT	No	1 to 4096

Note: The width and height inputs must be divisible by 16, as the node internally divides them by this factor to create the latent dimensions.

Outputs

Output Name	Description	Data Type
`samples`	A latent tensor filled with zeros. The shape is `[batch_size, 128, height // 16, width // 16]`.	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): 337c687b6b3ebb8854618911ab966118940de369ed84bc9c4ed769289b397588

​Inputs

​Outputs

Inputs

Outputs