ComfyUI_Mexx_Poster

ComfyUI_Mexx_Poster doesn't generate anything. It's an output node that takes the image you just made and drops it into a trading-card-game-style frame - the kind with a clan badge, HP pips, a nickname, and a "skill description" in Chinese - then saves the finished poster next to your raw render. Think of it as a one-click "make it look like a card from a mobile CCG" filter sitting at the end of your graph. It's from a small Chinese-authored pack (SoftMeng), and it does exactly one job. If you post anime/waifu gens and want them to read as collectible card art, that job is worth having.

Full honesty before you install: this is a niche, gimmicky node with basically zero English-community footprint - Reddit has nothing on it, so you're on your own for polish. It ships with hardcoded Chinese card text and randomized colors. That's a feature if you like the vibe and a wall if you don't. The saving grace is that it's just HTML/CSS under the hood, so you can edit the templates yourself. We'll get to that.

How it works

The source is refreshingly small. On execution it does two things:

1. Saves your image batch as a PNG (with ComfyUI's workflow metadata embedded, so your graph travels with the file - a nice touch for sharing).
2. Renders an HTML template to PNG using the html2image library, which is a headless Chrome screenshot. It throws your saved image at a CSS background, layers on the card frame, randomizes a few colors and some Chinese card text, and screenshots the result at a fixed size.

Two templates ship in the pack: card (425×585) and card_big (768×1040). It writes the plain image and then <prefix>_NNNNN_Poster.png into your ComfyUI output folder.

Inputs that matter

Only five, and three are worth your attention:

• images (IMAGE) - your render, straight from the VAE decoder. Only the first frame ends up in the poster.
• style - the enum: card or card_big. Bigger template, same frame.
• filename_prefix - base name for both saved files, like the built-in Save Image node.
• fps (default 6) - only matters if you feed a batch; it sets the frame duration when saving as an animated PNG. Doesn't touch the poster.
• compress_level (default 4, 0–9) - PNG compression on the raw save. Leave it.

There are no outputs - it's an output node; it just writes files and shows you the poster in the UI.

Install

ComfyUI Manager will find it if you search the pack title, or:

cd ComfyUI/custom_nodes
git clone https://github.com/SoftMeng/ComfyUI_Mexx_Poster
cd ComfyUI_Mexx_Poster
pip install -r requirements.txt   # just html2image

Then restart ComfyUI. There's a demo workflow in the repo that pairs this with the author's companion ComfyUI_Mexx_Styler pack and an LCM checkpoint (MexxLDI MSDXL LCM 2 on Civitai). Not required - the README just notes that without an LCM model you'll want to bump your sampler steps.

Where people get burned

• It needs Chrome. html2image renders via headless Chrome, and requirements.txt won't install that for you. On a bare Linux box or a RunPod image without a browser you'll hit a CouldNotFindBrowser-style error. Fix: install chromium or google-chrome, or point it at an existing binary with CHROME_BIN (other env candidates: CHROME_EXE, HTML2IMAGE_CHROME_BIN).
• The text is fixed Chinese flavor text. Random clans, nicknames, and skill descriptions come from a hardcoded list in the source - you can't feed it a title through the node. To make it yours, edit card/poster.html and poster.css in the pack directory. That's the real power move here.
• Aspect ratio. The card frame uses background-size: cover, so a landscape or square render gets center-cropped. Generate portrait-ish (the card templates are roughly 3:4) or you'll lose the edges of your composition.

If you want a controllable English card frame, you'll outgrow this fast. But for a zero-effort "post this as a card" end to a workflow, it's charming, tiny, and - once Chrome is installed - dead reliable.