• 核心功能

在 Stable Diffusion / ComfyUI 等文生图平台中，通过触发词 shion (overwatch) 稳定生成《守望先锋》2026 年新英雄死怨（Shion）的角色形象

• 使用方式

出图提示词：shion (overwatch) + 场景/动作/穿搭描述（如白色西装、长筒西裤、红色皮鞋、机械手等）

• 使用建议：

1. 正向提示词以触发词开头：shion (overwatch), 1girl, solo, <其他标签…>

2. 推荐 LoRA 权重 0.6 ~ 0.9（按底模微调）。

3. 负面提示词建议：lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, jpeg artifacts, signature, watermark, username, blurry, bad quality

4. 想强调角色时提高权重，想自然融入时用 0.6 左右。

Core Function

A character-consistency LoRA that reliably generates the Overwatch 2026 hero Shion (死怨) in Stable Diffusion / ComfyUI and other text-to-image platforms via the trigger word shion (overwatch)

All visual traits (white hair, red oni horns, hannya mask, red eyes, white suit, mechanical inner suit, gold tiger-head buckle, red-and-blue-and-gold dual pistols, etc.) are bound to the trigger word, so no need to describe appearance details in every prompt

How to Use

Prompt format: shion (overwatch) + scene/action/outfit tags (e.g., white suit jacket, long suit pants, red leather shoes, mechanical hands, etc.)

Usage Tips

Start the positive prompt with the trigger word: shion (overwatch), 1girl, solo, <other tags...>

Recommended LoRA weight: 0.6 – 0.9 (fine-tune based on your base model)

Recommended negative prompt: lowres, bad anatomy, bad hands, text, error, missing fingers, extra digits, fewer digits, cropped, worst quality, low quality, jpeg artifacts, signature, watermark, username, blurry, bad quality

Raise the weight for a stronger character emphasis; use around 0.6 for a more natural blend