LoRA Visualizer

Paste a prompt with <lora:name:0.8> tags into the LoRA Visualizer and it draws the receipt right on the canvas: a color-coded breakdown of every LoRA referenced, with its thumbnail, strength, trigger words, base model, and example images. Standard LoRAs get a blue section, <wanlora:...> video LoRAs get orange, and hovering a thumbnail pops up the trigger words and example prompts. It's an inspection node - you stick it at the end of a workflow to sanity-check what a prompt actually references before burning a render, or to finally remember what that one LoRA you downloaded and never used actually does.

The name undersells the mechanism slightly. All the parsing happens in Python on the backend - a regex that handles messy names like <lora:Detail Enhancer v2.0: Professional Edition:0.8> - and the visualization is a custom canvas widget rendered by the JavaScript frontend. That split is why it can show you real metadata and not just the tag text: for each detected LoRA it loads the {name}.metadata.json file from your models/loras folder, which is exactly the file format ComfyUI LoRA Manager writes when it fetches a model's page data. No metadata file, no thumbnail - just a line saying "Preview: Not available."

It's a genuinely thin node, and that's fine. One input: prompt_text, the full prompt you want inspected. Two outputs:

• raw_lora_info - a structured text report: name, strength, tag, trigger words, base model, preview/examples availability, and the Civitai URL when the metadata has a model ID.
• original_prompt - the prompt passed through unchanged, so you can tee the string up for whatever comes next.

Because it's marked as an output node, it lives at the end of a chain rather than feeding forward - in the README's example pipeline it sits between the Prompt Composer and the Prompt Splitter so you can eyeball what the composer picked before you split and render.

Install is the same for every node in this pack: ComfyUI Manager, search "LoRA Visualizer", restart; or

cd ComfyUI/custom_nodes
git clone https://github.com/oliverswitzer/ComfyUI-Lora-Visualizer.git

and restart ComfyUI. The dependency you can't skip is ComfyUI LoRA Manager - the pack doesn't call it directly, it just reads the metadata files that Manager generates, so "everything shows Not available" is your first and loudest clue that Manager isn't installed or hasn't refreshed that model yet. That's the fix, not a settings hunt.

Two things to keep in mind. First, <wanlora:...> is not standard ComfyUI syntax - it's this pack's own convention for Wan video LoRAs, so the node doubles as a validator that your tags are at least parseable. Second, and this trips people up: it displays, it doesn't apply. Dropping the Visualizer into your graph won't load any LoRAs or change your output; it's a dashboard, not a loader. The strength values it shows you are the ones in your prompt text - if you want them to do anything, the actual LoRA loading still happens in whatever loader your workflow uses.