Mememizator Combine Images

Mememizator Combine Images does exactly what the name says: it takes up to four IMAGE inputs, pastes them onto one canvas, and hands you back a single image you can feed straight into the main Mememizator node. It's the front half of a "before/after" or four-panel meme workflow - think two generated images side by side with a demotivator caption under the whole thing.

Why is this a separate node instead of something you'd do in Photoshop? Because the whole point is that it runs inside the graph, on the batch, deterministically. Combine four images, caption the result, save - all without leaving ComfyUI. It's also genuinely useful outside memes as a quick way to make comparison grids of your own generations, which is honestly its best use.

How it works

The node converts each input tensor to a Pillow image, pastes them onto a canvas, and converts back to a tensor. Layout is the main knob:

• horizontal - a single row.
• vertical - a single column.
• 2x2 - a grid of up to four cells, two per row.

Each image is centered inside its cell, padding sets the gap between and around images, and background (a hex color, default #000000) fills the empty canvas. If you set max_width or max_height above 0, the result is downscaled (LANCZOS) to fit while keeping aspect - set either to 0 to disable that limit. Four inputs into one canvas, no surprises.

The inputs that matter

• image_1 - required. image_2, image_3, image_4 - optional; leave any unused.
• layout - horizontal, vertical, or 2x2.
• max_width / max_height - output size caps, 0 = no cap.
• background - canvas color as a hex string.
• padding - spacing around and between images, in pixels.

The output is a single IMAGE tensor - wire it into the image input on Mememizator, or into a Save Image for a plain comparison grid.

The one gotcha: batches

Batches are supported but they have rules. If image_1 is a batch of four, then every optional input you connect must either match that batch size or be a single image that gets broadcast across the whole batch. Give it two inputs with mismatched batch sizes and the node raises an error rather than guessing - which is the right behavior, but it's the kind of thing that trips people up the first time they feed a batch into a four-input node.

Install

This one ships in the same pack as the other two nodes, so install is identical: ComfyUI Manager, search ComfyUI Mememizator, or clone it by hand:

cd ComfyUI/custom_nodes
git clone https://github.com/ArtemKo7v/ComfyUI-Mememizator
cd ComfyUI-Mememizator
pip install -e .

Dependencies are just numpy and pillow - no models, no torch install, nothing heavy. Restart ComfyUI after installing, and remember the pack needs internet on first meme render to fetch the Impact font. A small, focused utility, and honestly the only one of the three I'd reach for on its own just to make comparison grids.