Nodes/ComfyUI-FLOAT_Optimized/Apply FLOAT Encoder (VA)
ComfyUI Node

Apply FLOAT Encoder (VA)

Applies the loaded FLOAT Encoder to a batch of reference images. It preprocesses the images and passes them through the encoder to extract the core appearance latent and multi-scale feature maps, which are bundled into a single Appearance Pipe.

By set-soft·Created about a year ago·Updated 12 months ago· 33
Apply FLOAT Encoder (VA)
  • ref_image
  • float_encoder
  • appearance_pipe (Ws→r)
  • r_s_lambda_latent
  • float_encoder_out
CategoryFLOAT/Very Advanced

Inputs (2)

NameTypeDefaultDescription
ref_imageIMAGEA batch of reference images, correctly sized (e.g., 512x512) for the encoder.
float_encoderFLOAT_ENCODER_MODELThe loaded FLOAT Encoder model module.

Outputs (3)

NameTypeDescription
appearance_pipe (Ws→r)FLOAT_APPEARANCE_PIPE
r_s_lambda_latentTORCH_TENSOR
float_encoder_outFLOAT_ENCODER_MODEL