Composite Images ♾️Mixlab

This is the plain, sturdy compositor in Mixlab's Layer group: drop a foreground onto a background at a spot you pick, scale it, and optionally blend it in with multiply instead of a flat paste. No layout math, no manual x/y - you tell it a corner or edge, it does the placing. It's the node behind the pack's poster and layer-composite example workflows, where a piece of art gets pasted onto a mockup, a logo gets dropped onto a product shot, that kind of thing.

How it works

Straight from the maintainer's own description: it "overlays a foreground image onto a background image at specified positions and scales, with optional blending modes and masking capabilities." You're not picking pixel coordinates - you're picking one of eight named anchors, and the node figures out the placement from there.

The inputs and outputs that matter

• background (IMAGE, required) - the base image everything gets pasted onto.
• foreground (accepts anything, required) - the thing being pasted. Loose typing here means it'll take output from more than just a plain Load Image node.
• mask (MASK, required) - shapes the foreground before it lands. This one's not optional, so even a plain rectangular paste needs a mask wired in; a solid white mask matching your foreground's shape gets you a plain rectangle.
• position (enum, optional) - eight choices: overall plus the four corners and the three top/center/bottom-center anchors (center_center, center_top, center_bottom, and the four corner variants). overall stretches to fill the whole background; everything else anchors the foreground at that point.
• scale (FLOAT, default 0.35, range 0.01–1) - how big the foreground renders relative to the background. Default is a third of the frame - easy to forget and end up wondering why your logo is tiny.
• is_multiply_blend (BOOLEAN, default false) - flip it on for a multiply blend (good for shadows, watermarks, anything that should darken rather than sit flat on top) instead of a plain overlay.

The single output is IMAGE - the composited result, ready for a Save Image or another round of compositing.

How to install it

Ships with the Mixlab pack.

• ComfyUI Manager: search comfyui-mixlab-nodes, install, restart.
• Manual: cd ComfyUI/custom_nodes && git clone https://github.com/shadowcz007/comfyui-mixlab-nodes.git, then install requirements - install.bat on Windows portable, or ../../../python_embeded/python.exe -s -m pip install -r requirements.txt otherwise - and restart.

No model downloads for this specific node. Worth knowing before you install, though: this pack is a big grab-bag - LLM chat nodes, text-to-speech, screen sharing, 3D viewers - bundled behind one requirements.txt. You're pulling in a chunk of dependencies for the whole pack just to get one compositor. That's a known friction point with ComfyUI generally: custom nodes share one Python environment with no isolation, so a big pack's requirements can occasionally step on something else you've already installed.

Common issues & troubleshooting

"My paste came out full-frame or oddly stretched." Check position first - overall fills the background regardless of scale. If you wanted a positioned, scaled paste, pick one of the anchor options instead.

"Nothing pastes / it errors on a missing input." mask is required, not optional - this trips people up because a lot of composite nodes elsewhere make masking optional. Feed it a plain white mask if you just want the whole foreground to show.

"The foreground is way smaller than expected." That's the 0.35 default scale. Bump it toward 1 for something closer to full-size.

Install fails or drags in unrelated errors. This is a large pack with a lot of separate features stacked into one requirements.txt - if install.bat throws an error tied to something you'll never use (TTS, LLM clients, screen capture), it's usually safe to run the plain pip install -r requirements.txt manually and watch which specific line fails, rather than assuming the whole pack is broken.