π Easy LoRA Basic Loader
The catch-all LoRA loader with four spare prompt slots
- model
- clip
- MODEL
- CLIP
- prompt_1
- prompt_2
- prompt_3
- prompt_4
Not every LoRA is a character or an outfit. Most of your loras folder is probably style transfers, concept adapters, props, vehicles, animals, tools - the stuff that has a trigger word and a preferred strength but no natural "character slot." Easy LoRA Basic Loader, from Comfy Cabinet, is the pack's catch-all node for exactly that.
The pack ships five specialized LoRA loaders, and four of them are opinionated about what category of LoRA they load. The Basic one is the grab bag. It covers everything the others don't - style, concept, tool, assets, vehicle, objects, animal, and base-model LoRAs - and instead of pretending those map onto character or location slots, it gives you four neutral prompt outputs: prompt_1 through prompt_4.
How it works
Same underlying mechanics as its siblings: model and clip in, LoRA applied via ComfyUI's core LoraLoader, patched MODEL and CLIP out. strength_model and strength_clip default to 1.0, span -20 to 20, and 0/0 passes the model through untouched as a clean bypass.
The four prompt strings follow the same override rule as every Easy LoRA node: type something on the canvas and it wins; leave a field empty and it pulls the saved value from the config database. The lora_name dropdown is filtered to the Basic node's category set - and unlike the character or clothing loaders, that's a wide net, so most of your library will show up here. Select a LoRA and the strengths and prompt blocks auto-populate from the saved config.
What you set
Required: model, clip, lora_name, strength_model, strength_clip. Optional: prompt_1 through prompt_4, each a multiline string. Outputs: MODEL, CLIP, and the four strings. For a style LoRA, you'll usually just fill prompt_1 with the trigger and maybe prompt_2 with a quality qualifier, then stitch them into your text encoder. The four-slot design is really there for concept stacks - a vehicle LoRA's trigger plus a material tag plus an era tag - where each fragment deserves to be a labeled field instead of a wall of text.
Install and config
Standard for the pack:
cd ComfyUI/custom_nodes
git clone https://github.com/Zed93/Comfy-Cabinet.git
Restart ComfyUI, or install via ComfyUI Manager by searching "Comfy Cabinet". The only pip dependency is aiohttp, which ComfyUI already ships. Config lives in the dashboard behind the ποΈ Easy Config button in the top menu - browse the gallery, sync Civitai metadata with your own API key (both civitai.com and civitai.red work), tune strengths and prompts, and send the config to the node on canvas.
Where people get burned
The familiar three: configs are keyed by exact LoRA filename, so renaming a file orphans its saved settings; the Civitai sync needs a key but generation doesn't; and it's release-candidate software, so restart before you debug. One practical note: because the Basic dropdown covers a huge category set, your filtered list is long, so the dashboard's search box is the fastest way to find a LoRA if you're not sure it synced. And if a LoRA is categorized wrong on Civitai (it happens - a character LoRA tagged as a concept), this is the node that will still show it to you, which makes it a decent fallback when a specialized loader's filter hides something you're sure you own.
Inputs (9)
| Name | Type | Default | Description |
|---|---|---|---|
| model | MODEL | β | |
| clip | CLIP | β | |
| lora_name | COMBO | 0 options: | |
| strength_model | FLOAT | 1.00-20β20 | β |
| strength_clip | FLOAT | 1.00-20β20 | β |
| prompt_1opt | STRING | β | |
| prompt_2opt | STRING | β | |
| prompt_3opt | STRING | β | |
| prompt_4opt | STRING | β |
Outputs (6)
| Name | Type | Description |
|---|---|---|
| MODEL | MODEL | β |
| CLIP | CLIP | β |
| prompt_1 | STRING | β |
| prompt_2 | STRING | β |
| prompt_3 | STRING | β |
| prompt_4 | STRING | β |