Noct Q NSFW Workflow

A ComfyUI workflow for Noct Q (Qwen Image 2.1, uncensored) that writes a new nude scene every time you press Run. It uses stock ComfyUI nodes only, so there is nothing to install.

What it does

• A prompt template with seven random groups: photo look, hair, build, one of six lit scenes, expression, camera angle and a skin detail. ComfyUI draws each group at every run; edit a group to lock it or add your own options.

• Noct Q's quality setting: 25 steps, euler, simple, cfg 3 with a negative prompt.

• A detail pass that enlarges the image x1.3 and redraws it lightly (8 steps, cfg 1, denoise 0.3). It sharpens skin, hair, hands and faces without doubling edges and adds about 40 seconds on an 8 GB card. Can be bypassed for speed.

• Four explicit scene prompts in the notes, ready to paste, and a text file with 30 drawn prompts.

What you need

• ComfyUI 0.37 or newer.

• Noct Q V3 Base int8 in models/diffusion_models, from the Noct Q page, with its text encoder and VAE.

Notes

• The notes list the words that work best.

• The showcase images carry their prompts and settings: drag one into ComfyUI to load the workflow with that prompt.

中文：Noct Q（Qwen Image 2.1 无审查模型）的 ComfyUI 工作流：每次运行自动生成一个新的裸体场景，只用 ComfyUI 原生节点。25 步、cfg 3 加负面提示词，附带细节增强（放大 1.3 倍，8 步，cfg 1，去噪 0.3）。说明里有四个成人场景提示词。需要 Noct Q V3 Base int8。