Reencode Latent (pipe)

Documentation

• Class name: ReencodeLatentPipe
• Category: ImpactPack/Util
• Output node: False

This node is designed to re-encode latent representations by transforming them through a specified input and output basic pipe. It facilitates the modification of latent spaces, enabling the transition of samples from one latent representation to another, potentially enhancing or altering their characteristics.

Input types

Required

• samples
  • The latent samples to be re-encoded. These samples are the starting point for the transformation process.
  • Comfy dtype: LATENT
  • Python dtype: torch.Tensor
• tile_mode
  • Specifies the mode of tiling to be used during the re-encoding process, affecting how the samples are decoded and encoded.
  • Comfy dtype: COMBO[STRING]
  • Python dtype: str
• input_basic_pipe
  • The basic pipe through which the samples are initially passed for decoding or transformation.
  • Comfy dtype: BASIC_PIPE
  • Python dtype: tuple
• output_basic_pipe
  • The basic pipe used for the final encoding or transformation of the samples, determining their new latent representation.
  • Comfy dtype: BASIC_PIPE
  • Python dtype: tuple

Output types

• latent
  • Comfy dtype: LATENT
  • The re-encoded latent samples, representing the transformed latent space.
  • Python dtype: torch.Tensor

Usage tips

• Infra type: GPU
• Common nodes: unknown

Source code

class ReencodeLatentPipe:
    @classmethod
    def INPUT_TYPES(s):
        return {"required": {
                        "samples": ("LATENT", ),
                        "tile_mode": (["None", "Both", "Decode(input) only", "Encode(output) only"],),
                        "input_basic_pipe": ("BASIC_PIPE", ),
                        "output_basic_pipe": ("BASIC_PIPE", ),
                    },
                }

    CATEGORY = "ImpactPack/Util"

    RETURN_TYPES = ("LATENT", )
    FUNCTION = "doit"

    def doit(self, samples, tile_mode, input_basic_pipe, output_basic_pipe):
        _, _, input_vae, _, _ = input_basic_pipe
        _, _, output_vae, _, _ = output_basic_pipe
        return ReencodeLatent().doit(samples, tile_mode, input_vae, output_vae)