CreateHookLoraModelOnly_motorway_edition

Most LoRA files carry two sets of weights: the model weights that change how the UNet/DiT draws, and the CLIP weights that nudge the text encoder. Sometimes you want one without the other. CreateHookLoraModelOnly builds a scheduled LoRA using only the model-side weights - no CLIP - and the _motorway_edition is the same node reading its inputs from agilly1989's Motorway bus.

When you'd reach for this

Model-only LoRA application is the right call when a LoRA's trigger changes too much in the CLIP side, or when you're applying a style/LoRA purely as image-space influence and want the prompt untouched. In the hooks world specifically, CreateHookLora (full) and CreateHookLoraModelOnly (model side only) let you schedule either flavor over the sampling timeline via keyframes. If you never use strength_clip, this is the node that drops the confusing extra dial.

Inputs and outputs

• MOTORWAY 🚌💨 - the bus, required.
• INPUT_prev_hooks_key - key of an existing HOOKS group to append to, default prev_hooks.
• lora_name - the LoRA dropdown (files in ComfyUI/models/loras), set directly on the node.
• strength_model - model strength, -20 to 20, default 1.
• OUTPUT_HOOKS_key - key for the HOOKS object, default HOOKS.

Only output: the Motorway. Notice what's missing next to the full variant: no strength_clip, because the CLIP side isn't touched at all.

Installing

cd ComfyUI/custom_nodes
git clone https://github.com/agilly1989/ComfyUI_agilly1989_motorway

Restart, or ComfyUI Manager → "ComfyUI_agilly1989_motorway". No pip deps; you supply the LoRA files.

Caveats

Pack-wide: active beta, "BIG BROKEN WITH ASYNC" README banner, _motorway_edition clones commented out of __init__.py in the checked version (re-enable ClonedNodeMapping if missing), keys exact and case-sensitive. And the same upstream caveat as every hook node: hooks are EXPERIMENTAL in core ComfyUI, and a sampler that doesn't support hook scheduling turns this into a static model-only LoRA load - check your sampler before building a whole timeline around it. For a plain unscheduled model-only LoRA, the regular LoraLoaderModelOnly is the simpler tool.