Skip to main content
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
The Empty Qwen Image Layered Latent node creates a blank, multi-layered latent representation for use with Qwen image models. It generates a tensor filled with zeros, structured with a specified number of layers, batch size, and spatial dimensions. This empty latent serves as a starting point for subsequent image generation or manipulation workflows.

Inputs

ParameterData TypeRequiredRangeDescription
widthINTYes16 to MAX_RESOLUTIONThe width of the latent image to create. The value must be divisible by 16. (default: 640)
heightINTYes16 to MAX_RESOLUTIONThe height of the latent image to create. The value must be divisible by 16. (default: 640)
layersINTYes0 to MAX_RESOLUTIONThe number of additional layers to add to the latent structure. This defines the depth of the latent representation. (default: 3)
batch_sizeINTNo1 to 4096The number of latent samples to generate in a batch. (default: 1)
Note: The width and height parameters are internally divided by 8 to determine the spatial dimensions of the output latent tensor.

Outputs

Output NameData TypeDescription
samplesLATENTA latent tensor filled with zeros. Its shape is [batch_size, 16, layers + 1, height // 8, width // 8].