Extensions/Z-Image-Turbo-Lora-Stack-V4
ComfyUI Extension

Z-Image-Turbo-Lora-Stack-V4

A ComfyUI extension with 1 custom node.

By aistudynow·Created 6 months ago·Updated 6 months ago· 22
aistudynow/Z-Image-Turbo-Lora-Stack-V4
Nodes1
On cloudLocal install
Categoryloaders/lora
Stars22
Updated6 months ago

Nodes (1)

ZImage Turbo LoRA Stack V4
loaders/lora
Readme

ZImageTurboLoraStackV4 (No Nunchaku)

Standalone ComfyUI custom node that applies LoRAs to MODEL only, with key translation for Z-Image style naming.

What it does

  • Accepts MODEL + CLIP input and returns MODEL + CLIP.
  • Applies LoRA only to MODEL (CLIP is passed through unchanged).
  • Supports up to 10 LoRA slots (lora_count, enabled_i, lora_name_i, strength_i).
  • Skips processing when strength is 0.
  • Extracts the core transformer (model.model.diffusion_model) for mapping checks.
  • Rewrites to_qkv -> qkv automatically.
  • Attempts fusion remapping for:
    • to_q/to_k/to_v -> qkv
    • w1/w3 -> w13
  • Falls back to ComfyUI default loader if custom translation fails.

Install

  1. Copy this folder into ComfyUI custom nodes:
    • ComfyUI/custom_nodes/ZImageTurboLoraStackV4
  2. Restart ComfyUI.
  3. Add node: ZImage Turbo LoRA Stack V4 (No Nunchaku).

Files

  • __init__.py
  • zimage_turbo_lora_stack_v4.py