Nodes/ComfyUI_agilly1989_motorway/ImageBlend_motorway_edition
ComfyUI Node

ImageBlend_motorway_edition

Blend two images with modes and a factor — from the pipe, not wires

By agilly1989·Created 2 years ago·Updated about a year ago· 7
ImageBlend_motorway_edition
  • MOTORWAY 🚌💨
  • MOTORWAY 🚌💨
INPUT_image1_keyimage1
INPUT_image2_keyimage2
blend_factor0.50
blend_mode
OUTPUT_IMAGE_keyIMAGE

ImageBlend_motorway_edition is the stock ComfyUI ImageBlend node with its plumbing moved into a MOTORWAY 🚌💨 pipe. Instead of two image wires, it reads both images out of the pipe by key, blends them, and drops the result back into the pipe.

It's part of agilly1989/ComfyUI_agilly1989_motorway, the author's key-value "pipe" pack - you park values in a Motorway under named keys and pull them out later, which is how you stop a big workflow from turning into a wall of crossing wires. Every *_motorway_edition node is an auto-generated clone of a stock node wearing the pipe harness; this one sits under agilly1989 Nodes/Motorway-ed/image/postprocessing.

Inputs that matter

  • INPUT_image1_key (default "image1") and INPUT_image2_key (default "image2") - pipe keys for the two images
  • blend_factor (FLOAT, 0 to 1, default 0.5) - how far toward image 2 you go
  • blend_mode - normal, multiply, screen, overlay, soft_light, difference
  • OUTPUT_IMAGE_key (default "IMAGE") - key the blended result lands under

The only visible output is the Motorway pipe. The blended image lives in the pipe under OUTPUT_IMAGE_key until a later node reads that key.

Quick grounding on the modes, because normal is not the only thing people use this for: multiply darkens (great for overlaying a texture or vignette), screen lightens (add glow), difference is the trick for finding what changed between two frames, and overlay/soft_light are the classic composite-style modes. blend_factor at 0.5 with normal is a straight average; dial it up to push toward the second image. Both inputs still need the same dimensions - blending doesn't resize anything.

How the wrapper works

Same pattern as the rest of the family: pop the Motorway, pull each input out via its INPUT_*_key string, run stock ImageBlend with your widget values, store the output under OUTPUT_IMAGE_key, and return a clone of the pipe so the input pipe isn't mutated for anything upstream.

Installing

cd ComfyUI/custom_nodes
git clone https://github.com/agilly1989/ComfyUI_agilly1989_motorway

Restart ComfyUI, or find it via ComfyUI Manager by searching "motorway". No requirements.txt, no model downloads - pure Python.

Gotchas

Keys are exact and case-sensitive; a 'MotorwayClass' object has no attribute 'hash_' error means the key you asked for was never written into the pipe - check spelling and that an earlier node wrote to that exact string. Also note the defaults trap: this node reads lowercase keys ("image1", "image2") but writes an uppercase one ("IMAGE"), so if you chain two motorway-edition nodes the strings won't line up until you set them yourself. And as with the whole pack, it's BETA with the clone feature disabled in the 1.1.7 hotfix - stock ImageBlend remains the reliable way to do this if the pipe version isn't in your menu.

Categoryagilly1989 Nodes/Motorway-ed/image/postprocessing

Inputs (6)

NameTypeDefaultDescription
MOTORWAY 🚌💨MOTORWAY 🚌💨
INPUT_image1_keySTRINGimage1
INPUT_image2_keySTRINGimage2
blend_factorFLOAT0.500–1
blend_modeCOMBO6 options: normal, multiply, screen, overlay, soft_light, difference
OUTPUT_IMAGE_keySTRINGIMAGE

Outputs (1)

NameTypeDescription
MOTORWAY 🚌💨MOTORWAY 🚌💨