Tiled VAE Decode¶
Documentation¶
- Class name:
VAEDecodeTiled_TiledDiffusion
- Category:
_for_testing
- Output node:
False
This node specializes in decoding latent representations into images using a tiled approach, enhancing the efficiency and quality of the decoding process, especially for large images. It leverages a variable tile size to optimize the decoding process, accommodating different computational constraints and image dimensions.
Input types¶
Required¶
samples
- The latent representation to be decoded into an image. This input is crucial for determining the content and quality of the output image.
- Comfy dtype:
LATENT
- Python dtype:
torch.Tensor
vae
- The variational autoencoder model used for the decoding process. This model is essential for transforming the latent representation back into pixel data.
- Comfy dtype:
VAE
- Python dtype:
torch.nn.Module
tile_size
- Specifies the size of the tiles used in the decoding process. Adjusting this parameter can optimize decoding performance and output quality.
- Comfy dtype:
INT
- Python dtype:
int
fast
- A boolean flag that, when enabled, accelerates the decoding process by potentially compromising on the output quality.
- Comfy dtype:
BOOLEAN
- Python dtype:
bool
Output types¶
image
- Comfy dtype:
IMAGE
- The decoded image, reconstructed from the latent representation using the specified VAE model and tile size.
- Python dtype:
torch.Tensor
- Comfy dtype:
Usage tips¶
- Infra type:
GPU
- Common nodes: unknown
Source code¶
class VAEDecodeTiled_TiledDiffusion(TiledVAE):
@classmethod
def INPUT_TYPES(s):
tile_size = get_rcmd_dec_tsize() * opt_f
return {"required": {"samples": ("LATENT", ),
"vae": ("VAE", ),
"tile_size": ("INT", {"default": tile_size, "min": 48*opt_f, "max": 4096, "step": 16}),
"fast": ("BOOLEAN", {"default": True}),
}}
RETURN_TYPES = ("IMAGE",)
FUNCTION = "process"
CATEGORY = "_for_testing"
def __init__(self):
self.is_decoder = True
super().__init__()