CivitAI Info Display

Every CivitAI image you've ever envied has its generation recipe embedded in its metadata - model, sampler, scheduler, steps, CFG, the whole thing. CivitAI is the ecosystem's central hub precisely because sample images ship with that full metadata attached. The CivitAI Info Display node exists to show it to you without you ever opening the browser tab: pick an image in the pack's gallery and this panel updates in place with the settings that made it.

The honest framing first: this node is a read-only display and nothing more. No inputs, no outputs, no wire to anything. ComfyUI's node-plumbing layer is full of nodes that touch no pixels - this is the extreme version, a note-to-self pinned to your canvas that the pack keeps refreshed for you. You use it as reference while you reconstruct a look: it tells you the checkpoint and sampler to load, the steps and CFG to match, and the CivitAI page URL if you want to go read the creator's own notes.

How it works

The mechanism is gloriously low-tech, which is why it works. When you select an image in the CivitAI Gallery, the pack's frontend JS parses the normalized metadata and draws the text straight onto the canvas for every Info Display node it finds in the graph. Nothing travels through the backend execution graph at all - that's why the Python class behind it is a noop() that returns an empty tuple. The display updates the moment you click, before you queue anything, because it never waited on a queue in the first place. It's marked as an output node (is_output_node: true), so ComfyUI treats it as a terminal - you can drop it anywhere and it won't pull your graph's execution down weird channels.

What you actually see is built from the same fields the gallery parses for everything else: model name, steps, CFG, sampler, and scheduler, with the CivitAI page URL on top. The v0.2.1 update cleaned up the sampler/scheduler parsing, and the v0.3.1 fixes stopped stale metadata bleeding in from previously-selected images and local uploads - the earlier version could show you a prompt from the last image you looked at.

Where it fits

This node only earns its place inside a selection workflow from this pack. Drop it next to the CivitAI Prompt Editor and the Image Preview node and you've got the pack's intended arrangement: browse → see the recipe → edit the prompt → generate. Its sibling nodes handle the parts you can actually wire into the graph; this one is the eyeball. Don't bother trying to connect anything to it - there are no sockets to connect.

Installing and troubleshooting

Install via ComfyUI Manager (search CivitAI ImagePlus) or:

cd ComfyUI/custom_nodes
git clone https://github.com/CovertBannana/ComfyUI_CivitAI_ImagePlus.git

Then restart ComfyUI. Dependencies are just aiohttp, numpy, Pillow - no models, no heavy installs.

The one recurring complaint with this pack is display nodes going stale, and the fix is almost always the same two steps: restart ComfyUI and hard-refresh the browser with Ctrl+F5. The frontend drives this node, so a cached old JS file means the panel simply stops updating. If the Info Display shows the wrong image's data, that's the stale-metadata bug class the v0.3.1 changelog fixed - update the pack, then do the restart-and-refresh dance.

Also worth knowing before you invest: the README flags this pack as being superseded by the author's newer unified ComfyUI-ImagePlus pack. If you're wiring a workflow around these nodes today, that's fine - but check the successor if you're starting from zero.