Fantastic Plotter Global Lora 🌐

The Fantastic Plotter tests LoRAs against each other - and sometimes that's the wrong question. Say you're comparing three character LoRAs but you always want your style LoRA active underneath. If the style LoRA goes into the sweep, every cell changes two variables at once and you can't tell which one moved the output. The Fantastic Plotter Global Lora exists to take LoRAs out of the sweep: whatever you put here applies to every image the Plotter generates, at a fixed strength, on top of each swept cell's own LoRA.

It's a single-purpose node with exactly one job, and it does it without ceremony.

How to use it

Add it via the Plotter's 🌐 Add Global Lora button, which spawns the node and wires it for you - or add it manually and connect its output to the Plotter's global_loras input. Either way, the wiring is one connection.

Inside, there's a plain row list rather than the loader's slot grid: ➕ Add Lora, then one row per LoRA with an enable box, the name (click to swap), a strength field, and arrows to reorder. The FOLDERS bar is there too, so you get the same folder filtering as the rest of the pack. No randomizer - the list stays fixed for a whole run, which is the point.

The node also carries two control-image toggles, and they're worth understanding because they replace the Plotter's own baseline:

• Control Image (no loras applied) - adds a baseline cell with zero LoRAs, the pure base model.
• Control Image (global loras applied) - adds a second baseline with only the globals and none of the swept LoRAs, so you can see what the globals alone contribute.

Enable both and your grid gains two control rows at the top, repeated across every column. While a Global Lora node is connected, the Plotter's own control toggle is disabled and relabeled "Control via Global node" - control is driven entirely from here. Disconnect it and the Plotter's toggle comes back.

Mechanism, briefly

Each enabled LoRA runs at its own fixed strength on every cell, applied after the swept cell's own LoRA, so the globals stack on top rather than competing. Concretely: if the Plotter sweeps character_v2 at 0.5 and 1.0 and this node holds style_painterly at 0.8, each swept cell gets character + painterly, and the control toggles add pure-base and painterly-only rows. The node's input is a hidden lora_data blob (UI-managed) and its single output is global_loras (type FL_GLOBAL_LORAS), which only the Plotter understands.

Install

Same pack, same one-liner:

cd ComfyUI/custom_nodes
git clone https://github.com/Adudeguyman/comfyui_fantastic-loras

or ComfyUI Manager → search Fantastic Loras, restart, hard-refresh. No dependencies - it's frontend-heavy and the Python side is a handful of lines that just repackages the payload.

Common issues

The usual suspects: if the node renders blank, hard-refresh the browser, then restart ComfyUI if it persists. If your baseline images unexpectedly disappeared, check whether a Global Lora node is connected with both control toggles off - that combination disables the Plotter's own control image, and the grid silently loses its reference row. And remember the strength math: globals stack on top of the swept LoRA, so at high strengths both, you're double-applying. That's often fine - it's the whole point of holding a style constant - but a 1.0 global plus a 1.0 swept character can oversaturate. It's the node you'll use once you start running real comparisons, and the difference between a legible grid and a muddle.