Nodes/Arctenoxs-Essentials_ComfyUI/VAE Encode + Dimensions (Arctenox's Essentials)
ComfyUI Node

VAE Encode + Dimensions (Arctenox's Essentials)

Encode an IMAGE to a LATENT via VAE, with dimension pass-through. Outputs: • latent – encoded latent (ready for KSampler / Refiner) • width – source image pixel width (INT) • height – source image pixel height (INT) The dimension outputs let you wire resolution directly into nodes like KSampler Refiner (target_width / target_height) or Cost Estimator without adding a separate Image Dimensions node. Tiled encode reduces peak VRAM for large images at the cost of potentially very subtle seam artifacts at extreme resolutions.

By Arctenox·Created 8 months ago·Updated about a month ago· 1
VAE Encode + Dimensions (Arctenox's Essentials)
  • image
  • vae
  • latent
  • width
  • height
tile_encodefalse
CategoryArctenox Essentials/Latent

Inputs (3)

NameTypeDefaultDescription
imageIMAGE
vaeVAE
tile_encodeoptBOOLEANfalseUse tiled VAE encoding to reduce VRAM usage. Useful for very large images. May produce subtle tile seams on high-contrast edges at extreme sizes.

Outputs (3)

NameTypeDescription
latentLATENT
widthINT
heightINT