Anima Multi LoRA Loader

LoRA workflow in ComfyUI usually means: go to Civitai, download files, drop them in the right folder, then chain a bunch of LoraLoader nodes together with weights you're tuning by hand. The Anima Multi LoRA Loader collapses that whole loop into one node - a frontend panel that searches Civitai, previews thumbnails, downloads the LoRA into the right place, and applies a stack of them to your model in one shot.

It exists for a real reason: Anima is unusually LoRA-friendly. It stacks multiple LoRAs without quality collapse (three at once is reportedly fine as long as you don't max every weight), and the model's author's framing is "a light touch is all you need." A node that makes stacking frictionless is exactly what that ecosystem wanted.

How it works

The node has two inputs:

• model - the base model (from a checkpoint loader) to apply LoRAs to.
• lora_list_json - a JSON array of entries like {"name": "my_lora.safetensors", "enabled": true, "strength_model": 0.8}, maintained automatically by the visual LoRA selector.

On execution it walks the list in order, resolves each file (ComfyUI's loras folder first, then the pack's own download directory, then a fuzzy basename match as a last resort), and applies it sequentially with its per-LoRA model strength via ComfyUI's standard LoRA loading path. The result is a single MODEL output you can pipe anywhere a model goes - no chain of LoraLoader nodes needed.

The frontend is where the magic lives: local LoRA previews, Civitai search, download progress, favorites, and a persistent thumbnail cache so already-seen LoRAs render instantly on re-open. Downloads land in a directory the pack manages, tracked in a config file under ComfyUI's user directory (anima_lora_config.json).

The inputs that matter

• model - your base checkpoint's MODEL output.
• lora_list_json - leave it alone and let the panel manage it; hand-edit only if you know the format.

Output: MODEL, same type, LoRAs applied.

Gotchas worth knowing

• Missing LoRAs are silently skipped. If a file can't be found anywhere, the node logs [Anima Tools] LoRA file not found: ... skipping. in the console and moves on. Your output will quietly not have that LoRA - check the console if results look wrong.
• Order matters. LoRAs apply sequentially, so the list order changes the final model. The panel gives you reordering controls; use them.
• It uses Civitai's public API (a search endpoint plus the civitai.red API base) - no API key needed for basic search and download, though a key may get you faster/uncapped API use if you're hammering it.

Install

Zero-dependency pack:

cd ComfyUI/custom_nodes
git clone https://github.com/nregret/Comfyui-Anima-Tools

Or ComfyUI Manager → "Anima Tools" → Install, then restart. Under AnimaArt as "Anima Multi LoRA Loader". No pip installs; the search and download features need internet, the thumbnail cache makes repeat visits fast.

The honest verdict: for a single LoRA, the stock LoraLoader is fine and you don't need this. The moment you're running Anima with two or three style/character LoRAs and you're tired of the download-and-move dance, this node is the workflow upgrade. Just keep your weights modest - Anima rewards a light touch.