Blend

Every ComfyUI user eventually needs to merge two images, and core ComfyUI makes you build the math yourself or reach for a heavyweight suite. Blend is the simple version: two images in, one blend mode, one factor, one image out. It's the layer-opacity tool from Photoshop, condensed into a single node, and it's quietly one of the most useful things in this pack because it just works.

How it works

It computes the selected mode as a per-pixel operation, then crossfades: image1 * (1 - blend_factor) + blended * blend_factor. So blend_factor is really "how much of the blend result shows over image1" - 0 is pure image1, 1 is the full blend mode. The result is clamped to 0–1.

• blend_mode - five real modes, and they behave exactly like you'd expect from any image editor:
  • normal - image2 straight on top of image1.
  • multiply - darkens; great for dropping texture or shadows onto a render.
  • screen - the opposite; brightens, good for light leaks and glows.
  • overlay - contrast-boosting hybrid; the "make it pop" mode.
  • soft_light - gentler, filmic contrast. My pick for a quick grade.
• blend_factor (0–1, default 0.5) - the crossfade amount.

The nicest touch: if the two images differ in size, the node center-crops image2 to image1's aspect ratio and bilinearly resizes it before blending. No manual resizing step, no dimension-mismatch errors. Output is a single IMAGE.

Install

Blend is one of ~22 nodes in ComfyUI-post-processing-nodes.

• ComfyUI Manager: search "post processing nodes", install, restart.
• Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/EllangoK/ComfyUI-post-processing-nodes

Restart after. No requirements.txt, no downloads - it's plain torch tensor math. It's under postprocessing → Blends in the node menu.

Using it well

The pattern people land on: generate the same image twice (or use one render plus a heavily edited version), then soft_light or overlay at factor 0.2–0.4 to push contrast and texture back into an over-smoothed render. Or use screen with a subtly brightened copy to add haze and light. Because the auto-resize makes it forgiving, it's also a fast way to composite two different-size outputs without thinking about dimensions - just don't expect it to handle completely different compositions gracefully. That's what masks are for.