Apply LoRA Hooks to Conditioning

The name is slightly lying to you, and that's the whole point. "Hooks" suggests per-step hook patching, but this node deliberately does the opposite: it applies the LoRA stack to the model the same boring, reliable way ComfyUI's core LoRA Loader does - comfy.sd.load_lora_for_models, an add_patches-style patch. The source code says why in plain terms: per-step hook weight patching causes ~15-minute stalls on 14B offloaded WAN models. This node exists to apply your stack fast, and to tell you exactly what it applied.

What it does

Four inputs, four outputs:

• model - your base WAN/flow model.
• positive / negative - conditioning.
• lora - the bundle from IAMCCS_WanLoRAStack (or a chained stack, or a scheduled stack from IAMCCS_WanLoRASchedule).

Outputs are the patched model (goes to the sampler's model input), positive and negative passed through unchanged, and a report string. The conditioning pass-through is the honest part of the design - the LoRA work happens on the model; the conditioning just rides along so the node fits cleanly in the middle of a graph.

The report is genuinely useful: it tells you whether the active stack is default or scheduled, what default LoRAs got applied, and what scheduled LoRAs got injected. In long loop workflows where different generations get different LoRAs, that report is how you verify the right LoRA was on the right generation instead of assuming it.

How it fits

In the pack's two-node LoRA system, this is the "apply" half. Wire IAMCCS_WanLoRAStack (up to 4 LoRAs, WAN-style key remapping) into lora, and this node patches the model. Because it uses the same loader as core ComfyUI, it's stable where fancier hook systems stall - the source docstring explicitly calls out that hook weight patching was the thing to avoid on 14B offloaded models. One caveat: the node forces a re-execute every queue run (IS_CHANGED returns NaN), so don't expect caching to skip it.

Install

Part of IAMCCS-nodes:

• ComfyUI Manager → search "IAMCCS" → install → restart.
• Or:

cd ComfyUI/custom_nodes
git clone https://github.com/IAMCCS/IAMCCS-nodes.git

Restart ComfyUI. Requirements: ComfyUI ≥ 0.3.0, Python ≥ 3.12, Torch ≥ 2.8. No model downloads - the LoRAs come from your own loras folder via the stack node.