Gigachad Power Lora Loader

Gigachad Power Lora Loader is the pack's answer to the multi-LoRA problem: one node that stacks any number of LoRAs, each with its own on/off toggle, strength, reorder controls, and a search box - all wrapped in a canvas skin that renders gold sparkle particles around the node. It commits to the bit, and honestly, it works.

The concept isn't new. The community standard is rgthree's Power Lora Loader, which does the same job and shows up in hundreds of workflow threads - it's the node people actually name when they ask "what utilities do I install." This is the Gigachad take on that idea: functionally similar, visually loud, and tied into the rest of this pack's "one node per concern" pipeline.

How it works

The Python side is clever about inputs. The node declares a FlexibleOptionalInputType, which accepts any key as a valid optional input, so each LoRA row you add in the UI becomes a lora_N argument at execution time. Each of those can arrive two ways:

• as a dict - the normal widget row: {on, lora, strength, ...},
• as a string - a filename piped in from another node's connection, with the row's strength and toggle still applied from its companion widgets.

The node then loops over every lora_N, resolves the file against your models/loras folder, and applies it to the model and CLIP. The JS side gives you the row UI: add/remove rows, move them up and down (order matters with LoRAs), toggle them individually, and search your library by name - handy if you've got thousands stashed away like the author clearly does.

Inputs are model and clip (both optional); outputs are MODEL and CLIP, so it slots straight into the standard loader → LoRA → sampler chain.

Where it fits

Stacking LoRAs in one node saves you a tower of chained LoraLoader nodes, and the per-row toggle is the killer feature for iterating: flip one off, re-run, flip it back, no graph surgery. If you keep a style LoRA, a character LoRA, and a detail LoRA active together, this is exactly the ergonomics you want.

Install and gotchas

cd ComfyUI/custom_nodes
git clone https://github.com/Winnougan/comfyui-gigachad.git

or search comfyui-gigachad in ComfyUI Manager, then restart. No extra Python dependencies - the "power" is all in the front-end JS.

Two honest notes. First, if you already use rgthree, there's no strong reason to switch - same workflow, and rgthree is the one with community support and years of battle-testing behind it. Second, LoRA application order is left-to-right/top-to-bottom and it does change results, so when you're chasing a specific look, keep your rows in a deliberate order and reorder rather than deleting and re-adding. This is a nice node for a LoRA-heavy workflow; it's just not the one the internet will tell you to install first.