💊 Easy LoRA Pose & Action Loader

Pose and action LoRAs are the unsung workhorses of anime and illustration workflows - the small files that make a character actually do something instead of standing there. They're also easy to misuse, because their trigger words are often long descriptive phrases ("sitting on a chair, legs crossed") that you can never quite remember from the Civitai page. Easy LoRA Pose & Action Loader, from Comfy Cabinet, is the specialized node that keeps those triggers saved next to the LoRA itself.

In the pack's prompt-slot system, pose and action LoRAs map to two of the eight category slots: expression and situation. That's a sensible pairing - a pose LoRA and an action LoRA both describe what's happening, not what's being worn or where the scene is. So this node exposes exactly those two prompt blocks, and no more.

How it works

The familiar Easy LoRA skeleton: model and clip in, LoRA applied via ComfyUI's core LoraLoader, patched MODEL and CLIP out, plus the two STRING outputs. strength_model and strength_clip default to 1.0, span -20 to 20, and 0/0 passes the model through unpatched if you want to silence the LoRA without touching the graph.

Two convenience details carry over from the other specialized loaders. The lora_name dropdown is filtered to pose/action-type LoRAs (with a fallback to the full list if nothing matches), and picking a LoRA auto-populates the strength widgets and prompt blocks from the config you saved in the dashboard. On the prompt outputs, anything you type on the canvas overrides the saved value; an empty field falls back to it.

What you set

Required: model, clip, lora_name, strength_model, strength_clip. Optional: the expression and situation multiline fields. Outputs: MODEL, CLIP, expression, situation. Feed the two strings into your text encoder - you'll typically stitch them together with the character and location blocks from other Easy nodes, since a pose is only meaningful once you've named who's posing and where.

Install and config

Same as every node in this pack:

cd ComfyUI/custom_nodes
git clone https://github.com/Zed93/Comfy-Cabinet.git

Restart ComfyUI (or install via ComfyUI Manager, searching "Comfy Cabinet"). The only dependency is aiohttp, which ComfyUI already ships. The dashboard - reachable through the 🗄️ Easy Config button in the top menu - handles the gallery, Civitai metadata sync (bring your own API key; both civitai.com and civitai.red work), and pushing saved configs to the node on canvas.

Where people get burned

Configs are keyed to the exact LoRA filename - rename the file, lose the saved trigger. The Civitai sync is optional; without a key there are no previews or trigger words but the loader still works. And it's release-candidate software, so when a button or node misbehaves after an update, restart before you assume it's broken. One trap specific to pose LoRAs: because this node splits prompts into expression and situation, don't forget the LoRA's own trigger phrase belongs in one of those fields - a pose LoRA with empty prompt blocks is just a silent strength boost and a confused-looking character.