Indexed LoRA Loader 🎯 🅑🅔🅣🅐

The normal way to try three LoRAs in ComfyUI is to build three Load LoRA nodes, wire them all up, and click through them one at a time. Indexed LoRA Loader is for when that gets old: you load up to 20 LoRAs into a single stack of slots, then pick which one gets applied by changing one integer. It's a "batch through my LoRA collection" node - wire the index input to a counter or an API-driven batch and you can generate a comparison grid of the same prompt across every LoRA you own without touching the graph. rgthree's Power Lora Loader is the heavyweight in this space; this one is the lightweight version that trades per-LoRA toggles for dead-simple indexing.

Mechanically it's a thin wrapper around ComfyUI's own load_lora_for_models. You feed it a model and clip, it looks at index, grabs the LoRA sitting in the lora_<index> slot, and applies it with strength_model and strength_clip (both default 1.0, range −100 to 100 - negative strengths are a real thing for style subtraction). The extra it gives you over stock: it extracts a trigger word from the LoRA's filename and hands it to you as a STRING output, so you can wire it straight into your prompt. The extraction rule is simple: everything before _lora in the filename. Snorricam-3434_lora.safetensors → Snorricam-3434. The trigger_suffix input (default _lora) lets you change that marker if your naming differs.

Inputs worth knowing:

• number_of_loras (1–20) - how many slots show up. The node's frontend JS hides unused slots, so you're not staring at 19 empty dropdowns.
• index (1-based) - which slot gets applied.
• lora_1 … lora_20 - the LoRA pickers; set any to "none" to skip it.
• strength_model / strength_clip - strengths, applied to the selected LoRA only.

Outputs: model and clip (the loaded pipeline), plus trigger_word (STRING).

Install is the pack install: ComfyUI Manager → search "BETA" or "Burgstall" → ComfyUI-BETA-Cropnodes (recently renamed ComfyUI-BETA-Helpernodes). Or:

cd ComfyUI/custom_nodes
git clone https://github.com/Burgstall-labs/ComfyUI-BETA-Cropnodes

restart, done. No extra dependencies for this node, and it reads your LoRAs from the standard ComfyUI/models/loras folder.

The failure mode to know is a feature: if index is out of range, a slot is "none", or the LoRA fails to load, the node doesn't crash - it returns your original model and clip unchanged and prints a warning. That's graceful, but it's also easy to miss if you're batch-running. Same for the trigger word: it's only as good as your filenames. If your files don't follow the name_lora.safetensors convention, the extracted "trigger word" is garbage, and the node has no way to know - that's the one thing you'll want to eyeball before trusting the output in a prompt.