Multiple Image Blend

A compact compositing node: plug in up to four images, set one opacity value, and get a single blended image back. There's no per-layer opacity, no blend modes to pick from - just one shared strength control across however many images you actually connect.

What it's for

This is the kind of node you reach for when you want to combine a handful of images into one without building out a chain of separate blend/composite nodes for each pair. Averaging a few generations of the same prompt to smooth out per-seed variance, laying a texture or overlay pass on top of a base render, or merging several passes of the same composition - anything where "combine these images at this strength" is the whole ask, rather than needing fine per-layer control.

image_a is the one required image slot; everything else is optional. That means the node works with as few as one image connected (in which case there's nothing to blend against, so it effectively passes through) up to four, with image_b, image_c, and image_d layered in on top as you wire them up. The shared opacity control governs how strongly the additional layers come through against the base.

Inputs and outputs that matter

• image_a (required) - your base image. This is the only image input the node insists on.
• opacity (required, 0–1, default 1, step 0.01) - one shared strength value applied across whatever additional images you've connected. At 1, the blend leans fully toward the combined layers; dial it down to let more of the base show through.
• image_b, image_c, image_d (all optional) - extra layers. Connect as many or as few as you need; leaving them empty just means fewer images get blended in.

The output is a single image - wire it into a Save Image node, a further processing step, or back into another blend node if you're chaining several composites together.

Installing it

ComfyUI Manager: search ComfyUI-ReplenishNodes, install, restart. Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/hben35096/ComfyUI-ReplenishNodes

No model downloads or extra Python dependencies documented for this node - it's plain image-processing logic.

Where people get tripped up

Images of different sizes. This node has no size-handling controls at all - no resize option, no alignment setting. If your four images aren't all the same resolution, expect an error or a broken composite rather than a graceful fallback. If you need to blend images of mismatched dimensions, this pack's own Multiple Image Blend 2 exists specifically to add an alignment option for that case - reach for that one instead if your sources don't already match.

Expecting per-layer control. There's exactly one opacity value shared across every connected image. If you need image_b blended in stronger than image_c, this node can't express that - you'd need to pre-blend pairs individually with separate opacity settings, or chain multiple blend nodes together, one pair at a time.

Only one image connected. With just image_a wired in and nothing else, there's nothing to blend against - you'll get back essentially the base image unchanged. That's expected, not a bug; the node only does work once a second image shows up.