Rebel Load LoRA 💊

Almost nobody runs a stock model in ComfyUI these days - half the fun is stacking LoRAs for a character, a style, or a look. Rebel Load LoRA 💊 is the pack's animated take on the core LoraLoader: feed it a model, a CLIP, and a LoRA file, and it returns the patched versions of both, with your chosen animation playing behind the sliders. It's a one-to-one stand-in for the stock loader - same sockets, same behavior - which means you can drop it into an existing chain without touching anything else.

How it works

It's the standard wrapper pattern: pull the native LoraLoader inputs, add theme and color_scheme, delegate to core ComfyUI. The LoRA patching itself - merging those trained weight deltas into the model and CLIP at the strengths you set - is done by ComfyUI's own code. The animation is pure frontend. Nothing about the finetune changes because the box is prettier, and that's the point: this is for people who want the Rebel look and their LoRAs, without learning a new loader.

The inputs that matter

Four of them do real work:

• model - the diffusion model the LoRA patches, wired from your checkpoint or UNET loader.
• clip - the CLIP model the LoRA patches, from your CLIP source. Many LoRAs touch both; the standard wiring feeds both sockets.
• lora_name - the LoRA file, from models/loras. The dropdown fills with whatever's in your LoRA folder.
• strength_model and strength_clip - how hard the patch applies, both defaulting to 1.0, both going negative if you want to invert or reduce a LoRA's influence. The model and CLIP strengths can be set independently, which matters for LoRAs trained to affect one side more than the other.

Outputs are the patched MODEL and CLIP, feeding downstream into your sampler chain - and if you're stacking multiple LoRAs, chain them in series (each loader's MODEL/CLIP into the next), the same way you would with the stock node.

Installing it

One clone, no dependencies, no model downloads from the pack:

cd ComfyUI/custom_nodes
git clone https://github.com/RealRebelAI/Rebels_Animated_Nodes.git

Restart ComfyUI, then hard refresh the browser (Ctrl + F5 / Cmd + Shift + R) so the cached JavaScript reloads and the animation actually draws. It lives under Rebel AI → Animated Core.

Where people get burned

• "The LoRA does nothing." Check strength_model/strength_clip - a 0 there is a silent off-switch. Also confirm the right sockets are wired; patching the model without the CLIP (or vice versa) is a classic way to get a LoRA that "loads" but barely shows.
• "New LoRA isn't in the dropdown." It reads models/loras; drop the file there and restart ComfyUI so the folder re-scans. That's a ComfyUI-ism, not a Rebel-ism.
• Animations flat on Nodes 2.0. The pack draws on the legacy LiteGraph canvas; the patching works fine but the backdrop can go missing under the newer frontend.
• Missing pack → red node. If you're loading someone's Rebel workflow without the pack installed, every Rebel node shows red; install the pack and restart.

If you've got a stack of LoRAs you load in every workflow, this is the loader that keeps the whole graph on-theme. It won't give you trigger words or multi-LoRA convenience like the rgthree Power Lora Loader does - that's a different (and genuinely powerful) beast. This is the same old reliable LoRA loader, reskinned, and there's nothing wrong with that.