用结构化JSON来给AI下指令

🎮 最近在AI圈子里，很多创作者开始用结构化的JSON格式来描述设计需求。这种方式不仅让每个参数都被精准表达，还极大提升了AI生成内容的可控性和一致性。与传统的自然语言提示相比，JSON格式能分层定义材质、光效、动作等细节，减少歧义和生成误差

📰 这种“JSON prompting”趋势不仅仅适用于图像也适用于视频。以上的像素游戏设备便是用一下的指令生成的

Prompt: retexture the image attached based on the JSON aesthetic below
{
"style": "voxel 3D",
"material": {
"base": "hard plastic",
"texture": "smooth with subtle gridlines",
"finish": "semi-matte"
},
"lighting": {
"type": "studio lightbox",
"intensity": "medium",
"shadows": "sharp and short",
"highlights": "mild, block edge reflections"
},
"color_scheme": {
"primary": ["#ffcc00", "#ff3c3c", "#000000", "#ffffff"],
"accents": ["#00cfff", "#00cc00", "#ff00cc", "#666666"],
"contrast": "high"
},
"shapes": "voxel blocks, chunky, pixelated geometry",
"composition": {
"background": "plain light gray",
"object_count": "single item per frame",
"layout": "centered, angled slightly for 3D view"
},
"rendering": {
"engine": "3D voxel renderer",
"resolution": "high",
"detail_level": "low-poly, simplified structure"
}
}

📮 这篇文章是转发给你的吗？点击这里加入频道，一起探索更多有趣的内容！