图像类 prompt 编写指南
与文本生成的 prompt 不同，图像生成更强调图片的属性，如图片风格、质量、目标数量等，一般图片生成的 prompt 包含以下内容
样式修饰符
能够产生特定样式的描述符（例如，“带有红色色调”、“玻璃制成”、“用 Unity 渲染”）
| 提示词 | 效果图 |
|---|---|
| A pyramid made of glass, rendered in Unity and tinted red |  |
| photorealistic, by greg rutkowski, by christopher nolan, painting, digital painting, concept art, octane render, wide lens, 3D render, cinematic lighting, trending on ArtStation, trending on CGSociety, hyper realist, photo, natural light, film grain |  |
质量增强器
提高生成图像的某些非特定样式质量的术语。例如，“amazing”、“beautiful” 和 “good quality” 都是质量增强器
| 提示词 | 效果图 |
|---|---|
| A pyramid made of glass, rendered in Unity and tinted red | |
| A beautiful, majestic, incredible pyramid, 4K |  |
重复
在提示中重复相同的词语或者类似短语会导致模型在生成的图片中强调该词语
| 提示词 | 效果图 |
|---|---|
| A beautiful painting of a mountain next to a waterfall. |  |
| A very very very very very very very very very very very very very very very very very very very very very very beautiful painting of a mountain next to a waterfall. |  |
加权
一些模型（如 Stable Diffusion、Midjourney 等）允许你对提示中的词语进行加权。这可以用于强调生成图片中的某些词语或短语
| 提示词 | 效果图 |
|---|---|
| mountain |  |
| mountain | tree:-10 |  |
修复变形生成
变形生成在许多模型中都是一个常见问题，特别是在人体部位（如手、脚）上。通过良好的反向提示语，可以在一定程度上解决这个问题
| 提示词 | 效果图 |
|---|---|
| studio medium portrait of Brad Pitt waving his hands, detailed, film, studio lighting, 90mm lens, by Martin Schoeller:6 |  |
| studio medium portrait of Brad Pitt waving his hands, detailed, film, studio lighting, 90mm lens, by Martin Schoeller:6 | disfigured, deformed hands, blurry, grainy, broken, cross-eyed, undead, photoshopped, overexposed, underexposed, lowres, bad anatomy, bad hands, extra digits, fewer digits, bad digit, bad ears, bad eyes, bad face, cropped: -5 |  |