Skip to main content
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

ParameterDescriptionData TypeRequiredRange
samples1The first latent sample to blendLATENTYes-
samples2The second latent sample to blendLATENTYes-
blend_factorControls the blending ratio between the two samples (default: 0.5)FLOATYes0 to 1
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 NameDescriptionData Type
latentThe blended latent sample combining both input samplesLATENT
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): a19808c5b606a8c05f2685fcd78d9f08c1ba51613a4029b36cf0ce5305618c2f