Repeat Into Grid (latent)

Documentation

• Class name: Repeat Into Grid (latent)
• Category: Bmad/latent
• Output node: False

This node is designed to replicate and arrange input latent samples into a grid format based on specified dimensions. It enables the creation of a structured layout of latent representations, facilitating operations that require uniform spatial arrangements, such as visualization or further processing in grid-based models.

Input types

Required

• samples
  • The latent samples to be tiled into a grid. This input is crucial for determining the content that will be replicated across the grid.
  • Comfy dtype: LATENT
  • Python dtype: torch.Tensor
• columns
  • Specifies the number of columns in the grid. This parameter directly influences the grid's width and the arrangement of the replicated samples.
  • Comfy dtype: INT
  • Python dtype: int
• rows
  • Determines the number of rows in the grid. It affects the grid's height and how the samples are distributed vertically.
  • Comfy dtype: INT
  • Python dtype: int

Output types

• latent
  • Comfy dtype: LATENT
  • The output is a modified version of the input latent samples, now arranged into a grid as specified by the input dimensions.
  • Python dtype: torch.Tensor

Usage tips

• Infra type: GPU
• Common nodes: unknown

Source code

class RepeatIntoGridLatent:
    """
    Tiles the input samples into a grid of configurable dimensions.
    """

    @classmethod
    def INPUT_TYPES(cls):
        return {"required": {"samples": ("LATENT",),
                             "columns": grid_len_INPUT,
                             "rows": grid_len_INPUT,
                             }}

    RETURN_TYPES = ("LATENT",)
    FUNCTION = "repeat_into_grid"
    CATEGORY = latent_category_path

    def repeat_into_grid(self, samples, columns, rows):
        s = samples.copy()
        samples = samples['samples']
        tiled_samples = samples.repeat(1, 1, rows, columns)
        s['samples'] = tiled_samples
        return (s,)