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 HunyuanRefinerLatent node processes conditioning and latent inputs for refinement operations. It applies noise augmentation to both positive and negative conditioning while incorporating latent image data, and generates a new latent output with specific dimensions for further processing.

Inputs

ParameterData TypeRequiredRangeDescription
positiveCONDITIONINGYes-The positive conditioning input to be processed
negativeCONDITIONINGYes-The negative conditioning input to be processed
latentLATENTYes-The latent representation input
noise_augmentationFLOATYes0.0 - 1.0The amount of noise augmentation to apply (default: 0.10)

Outputs

Output NameData TypeDescription
positiveCONDITIONINGThe processed positive conditioning with applied noise augmentation and latent image concatenation
negativeCONDITIONINGThe processed negative conditioning with applied noise augmentation and latent image concatenation
latentLATENTA new latent output with dimensions [batch_size, 32, height, width, channels]