Image ProcessingFull Capture

VAE Decode

class_type: VAEDecode

Source repo

Decodes a latent representation into a visible pixel image using a VAE (Variational Autoencoder). This is the final conversion step that transforms the sampler output into a viewable image.

What It Does

VAEDecode is the bridge between the latent space (where diffusion models operate) and pixel space (where humans see images). After the sampler produces a denoised latent, VAEDecode converts it into an RGB image tensor.

The VAE quality directly affects the final image—a poor VAE can introduce color shifts, blurriness, or artifacts even from a perfect latent. This is why many workflows use a separate VAELoader to override the checkpoint default.

Inputs

samplesLATENT

Latent image to decode.

vaeVAE

VAE model for decoding.

Outputs

IMAGEIMAGE

Decoded pixel image.

What Numonic Captures

  • VAE source (checkpoint default or external loader) traceable in workflow graph

Known Gaps

  • Decode quality metrics
  • Whether tiled decoding was used (handled by VAEDecodeTiled instead)

Related Nodes

VAE EncodeImage Processing
Load VAELoaders
KSamplerSampling
Save ImageI/O
Back to ComfyUI Nodes

Capture ComfyUI metadata automatically

Numonic extracts workflow metadata from every ComfyUI generation — models, samplers, seeds, prompts, and custom nodes. Track provenance, maintain compliance, and never lose a workflow.

Return to the Guide