KandinskyV5_VAEDecode

KandinskyV5_VAEDecode

Decodes Kandinsky-5 5D video latents to frames using the VAE's built-in tiling.

Pack: ComfyUI-Kandinsky

custom_nodes.ComfyUI-Kandinsky

Inputs (6)

Name	Type	Required
vae	VAE	required
samples	LATENT	required
enable_tiling	BOOLEAN	required
tile_min_frames	INT	required
tile_min_height	INT	required
tile_min_width	INT	required

Outputs (1)

Name	Type
IMAGE	IMAGE