LoRA Text Strength Visualizer WordPlot

When you're juggling four or five LoRAs in one prompt, reading the numbers is miserable: <lora:style_a:0.7> <lora:style_b:0.35> <lora:char:0.9> <lora:detail:0.5> - it takes real concentration to see the balance. LoRA Text Strength Visualizer WordPlot renders that balance as an image: each LoRA name drawn at a size and brightness proportional to its strength. The strongest LoRA dominates the plot, the weak ones shrink into the background. You glance, you understand, you move on.

It's a pure visualization node - it changes nothing, loads nothing, applies nothing. What it gives you is an at-a-glance snapshot of a prompt's LoRA hierarchy that's much easier to hold in your head than the raw tags. That's genuinely useful in two places: while you're tuning (watch the plot respond as you edit strengths) and when you're documenting (screenshot the plot into a workflow or a changelog so future-you knows what mix produced a result).

How it works

It parses <lora:name:strength> tags out of your text and renders them as a word plot on a canvas. The rendering rule is simple and effective: each LoRA's name is drawn scaled and brightened by its strength value - strong LoRA = big, bright text; weak LoRA = small, dim text. That turns the numeric balance into a visual one. It renders only the LoRA tags, not the whole prompt, so it's scoped to exactly the thing you're balancing.

Inputs:

• text - your prompt (multiline), containing the LoRA tags.
• width / height - canvas size, 64–2048, defaults to 512×512. Bump it up if you have many LoRAs and the plot gets cramped.

Output is a single IMAGE - preview it, save it, or wire it anywhere an image goes. Because it's an output node, it always runs, so it stays live as you edit.

It's the third member of the pack's LoRA-text trio, and the three compose nicely: LoRA Strength Multiplier on Text changes the strengths, LoRA Randomize Strength on Text re-rolls them, and this node shows you what either one produced. Feed the modified text from either into this node and watch the plot move.

Installing it

Part of CorvaeOboro's ComfyUI_illumorae (CC0, pure code - it's PIL/numpy text rendering, no models). ComfyUI Manager → search illumorae, or:

cd ComfyUI/custom_nodes
git clone https://github.com/CorvaeOboro/ComfyUI_illumorae

Common issues

• The plot is empty - no LoRA tags in your text. It renders tags only; plain prompt words don't appear.
• Everything looks the same size - if all your LoRAs are at similar strengths, the plot is (correctly) boring. That's information too.
• Font looks blocky - it picks a system font (with a Windows-first font path in the code); on Linux it can fall back to something basic. Raise width/height for legibility if it matters.
• "It didn't change my image" - it's a visualizer, not a loader. To actually apply strengths, the tags still have to reach a LoRA loader that reads them from text.