Noisy Latent Image

Documentation

• Class name: BNK_NoisyLatentImage
• Category: latent/noise
• Output node: False

The Noisy Latent Image node generates a tensor of random noise as latent images, configurable by dimensions and batch size, and intended for use in generative models where noise serves as a foundational element for image synthesis.

Input types

Required

• source
  • Specifies the computational resource (CPU or GPU) to be used for generating the noise, affecting the execution speed and efficiency.
  • Comfy dtype: COMBO[STRING]
  • Python dtype: str
• seed
  • Determines the initialization value for random number generation, ensuring reproducibility of the noise patterns.
  • Comfy dtype: INT
  • Python dtype: int
• width
  • Sets the width of the generated latent images, influencing the dimensions of the output noise tensor.
  • Comfy dtype: INT
  • Python dtype: int
• height
  • Sets the height of the generated latent images, influencing the dimensions of the output noise tensor.
  • Comfy dtype: INT
  • Python dtype: int
• batch_size
  • Defines the number of latent images to generate in a single batch, allowing for batch processing of noise generation.
  • Comfy dtype: INT
  • Python dtype: int

Output types

• latent
  • Comfy dtype: LATENT
  • The generated tensor of random noise, structured as latent images for further processing or model input.
  • Python dtype: torch.Tensor

Usage tips

• Infra type: GPU
• Common nodes: unknown

Source code

class NoisyLatentImage:
    @classmethod
    def INPUT_TYPES(s):
        return {"required": {
            "source":(["CPU", "GPU"], ),
            "seed": ("INT", {"default": 0, "min": 0, "max": 0xffffffffffffffff}),
            "width": ("INT", {"default": 512, "min": 64, "max": MAX_RESOLUTION, "step": 8}),
            "height": ("INT", {"default": 512, "min": 64, "max": MAX_RESOLUTION, "step": 8}),
            "batch_size": ("INT", {"default": 1, "min": 1, "max": 64}),
            }}
    RETURN_TYPES = ("LATENT",)
    FUNCTION = "create_noisy_latents"

    CATEGORY = "latent/noise"

    def create_noisy_latents(self, source, seed, width, height, batch_size):
        torch.manual_seed(seed)
        if source == "CPU":
            device = "cpu"
        else:
            device = comfy.model_management.get_torch_device()
        noise = torch.randn((batch_size,  4, height // 8, width // 8), dtype=torch.float32, device=device).cpu()
        return ({"samples":noise}, )