VNCCS LoRA Bypass

This is functionally the same job as core ComfyUI's LoraLoader: take a model and clip, apply a LoRA at some strength, hand back the modified model and clip. The "Bypass" in the name is the whole point - it's built so you can leave the node wired permanently into a graph and toggle whether a LoRA actually applies without deleting a node or rerouting anything.

VNCCS's 3.0 pipeline runs largely on Qwen-Image-Edit with a stack of task-specific LoRAs on top - posing, clothing transfer, and so on, per the pack author's own description of the rebuild. A node that lets you drop a LoRA slot into a workflow and cleanly no-op it when you don't need that particular LoRA for a given run is genuinely useful plumbing for a pipeline built that way, rather than a one-off convenience.

The inputs that matter

• model and clip - the standard pair every LoRA loader needs.
• lora_name - here's the detail worth knowing: it's a plain text field, not the dropdown you'd get from a stock LoraLoader. Default is the literal string "None". That means there's no autocomplete or picker pulling from your loras/ folder - you type the filename yourself, and a typo just silently fails to apply rather than throwing an obvious error. Set it to None and the node is meant to pass model/clip through unchanged, which is the actual bypass behavior.
• strength_model and strength_clip (default 1 each, range −10 to 10) - the usual two-strength split, letting you weight a LoRA's effect on the diffusion model separately from its effect on the text encoder.

Two outputs, model and clip, same as any LoRA loader - chain several of these to stack multiple LoRAs.

Where it fits

Use it anywhere in a VNCCS graph where you want a LoRA slot that's easy to disable for a test run without touching the wiring - comparing a character's output with and without a style LoRA, for instance, or building a workflow template you hand to someone else where they can type None to skip a LoRA they don't have installed.

How to install it

Bundled with the pack:

• ComfyUI Manager - search VNCCS - Visual Novel Character Creation Suite, install, restart.
• Manual - cd ComfyUI/custom_nodes && git clone https://github.com/AHEKOT/ComfyUI_VNCCS.git, then cd ComfyUI_VNCCS_Utils && pip install -r requirements.txt, restart, and click "Install missing custom nodes" in Comfy Manager once.

Common issues

LoRA doesn't seem to apply, no error shown. Because lora_name is a free-text field rather than a dropdown, this is almost always a typo or a mismatched extension - mylora instead of mylora.safetensors, wrong capitalization, or a filename copied from a different folder structure than the one ComfyUI actually scans. Since there's no autocomplete to catch it for you, copy the exact filename from your loras/ directory rather than typing it from memory.

Confusing this with the LoRA picker inside VNCCS Pipe. VNCCS Pipe has its own lora_name field, but that one's a proper dropdown enum populated from your installed LoRAs. If you want the safety of a picker rather than a text field, that's the node to use for a single LoRA slot inside the pipe bus - reach for LoRA Bypass specifically when you need an easily-toggled standalone slot outside the pipe.