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 LatentBlend node combines two latent samples by blending them together using a specified blend factor. It takes two latent inputs and creates a new output where the first sample is weighted by the blend factor and the second sample is weighted by the inverse. If the input samples have different shapes, the second sample is automatically resized to match the first sample’s dimensions.

Inputs

ParameterData TypeRequiredRangeDescription
samples1LATENTYes-The first latent sample to blend
samples2LATENTYes-The second latent sample to blend
blend_factorFLOATYes0 to 1Controls the blending ratio between the two samples (default: 0.5)
Note: If samples1 and samples2 have different shapes, samples2 will be automatically resized to match the dimensions of samples1 using bicubic interpolation with center cropping.

Outputs

Output NameData TypeDescription
latentLATENTThe blended latent sample combining both input samples