Nano Banana Pro JSON 提示词完全指南:掌握结构化 AI 图像生成
Nano Banana Pro JSON 提示词完全指南:掌握结构化 AI 图像生成
原文来源:fofr.ai | 作者:fofr (Google DeepMind 提示词工程师)
本指南基于 Nano Banana Pro 官方最佳实践,帮助你从初学者进阶为 JSON 提示词专家
目录
什么是 JSON 提示词?
JSON 提示词是一种结构化信息传递方式,它使用 JSON(JavaScript Object Notation)格式来组织图像生成指令。与传统的文本提示词不同,JSON 提示词将各种视觉元素分门别类地组织成层次化的数据结构。
核心概念
{
“subject”: {
“demographics”: { “age”: “25”, “gender”: “Female” },
“pose”: { “head_position”: “Front-facing”, “energy”: “Confident” }
},
“lighting”: { “type”: “Soft studio lighting” },
“camera”: { “lens”: “85mm”, “aperture”: “f/1.8” }
}
这种结构化的方式让 AI 模型能够:
-
更精确地理解每个视觉元素的细节
-
保持输出的一致性和可重复性
-
更容易地进行局部修改而不影响整体风格
为什么选择 JSON?
虽然理论上任何结构化数据格式都可以(作者个人偏好 YAML),但社区已经统一使用 JSON作为标准。JSON 的优势包括:
-
广泛的支持和工具生态
-
清晰的语法结构
-
易于阅读和编辑
-
与大多数编程语言兼容
为什么要使用 JSON 提示词?
Nano Banana Pro 的默认行为
当你使用简单的文本提示词(如"a photo of a dog")时,Nano Banana Pro 会:
-
生成美丽、温暖、细节丰富的图像
-
带有乡村风格和抓拍感
-
所有结果相似但又不完全相同
这种默认风格很棒,但当你需要精确控制时,就需要 JSON 提示词了。
JSON 提示词的优势
| 特性 | 简单文本提示词 | JSON 结构化提示词 |
|---|---|---|
| 控制精度 | 低,依赖模型解释 | 高,明确定义每个元素 |
| 输出一致性 | 变化较大 | 可重复、可预测 |
| 修改灵活性 | 需要重写整个提示词 | 可单独修改任意字段 |
| 复杂场景支持 | 容易遗漏细节 | 系统化覆盖所有方面 |
| 团队协作 | 难以标准化 | 可共享模板和组件 |
实际对比
简单提示词:
a photo of a dog
结果:温暖、美丽,但难以精确控制品种、姿势、光线等。
JSON 提示词:
{
“subject”: {
“type”: “Golden Retriever”,
“age”: “3 years old”,
“pose”: { “position”: “Sitting”, “gaze”: “Looking at camera” }
},
“lighting”: { “type”: “Golden hour sunlight” },
“camera”: { “lens”: “50mm”, “aperture”: “f/2.8” }
}
结果:精确匹配你的每一个要求。
何时应该使用 JSON 提示词
推荐使用场景
-
长提示词或复杂提示词
-
当需要描述多个对象、复杂的场景关系时
-
避免遗漏重要细节
-
-
需要在保持整体风格的同时改变细节
-
例如:同一角色的不同服装、不同表情
-
只需要修改特定字段,其他保持不变
-
-
想要摆脱默认风格
-
通过明确定义大多数细节来获得你想要的外观
-
避免模型的"默认美学"
-
-
已有 JSON 模板作为基础
-
使用社区共享的模板
-
在现有模板上进行修改
-
-
需要模仿现有图像或输出
-
将参考图像转换为结构化描述
-
创建可复用的视觉配方
-
不需要使用 JSON 的情况
-
简单的概念探索
-
快速原型制作
-
当你想要模型的创意发挥时
-
时间紧迫且不需要精确控制时
JSON 提示词的核心结构
一个完整的 Nano Banana Pro JSON 提示词通常包含以下核心部分:
1. Subject(主体)
定义图像的主要对象,通常是最详细的部分。
{
“subject”: {
“demographics”: {
“age”: “Early 20s”,
“gender”: “Female”,
“ethnicity”: “Asian”
},
“face”: {
“skin”: {
“tone”: “Fair, porcelain complexion”,
“texture”: “Smooth, high-end commercial retouch finish”,
“details”: “Subtle nose contour, soft highlights”
},
“eyes”: {
“color”: “Striking blue-grey”,
“gaze_direction”: “Looking upwards and slightly to the left”,
“makeup”: “Defined upper lashes, subtle eyeliner”,
“eyebrows”: “Thick, dark brown, well-groomed natural arch”
},
“mouth”: {
“shape”: “Soft, relaxed”,
“color”: “Natural pinkish-rose”,
“expression”: “Neutral to slightly contemplative”
}
},
“hair”: {
“style”: “Messy high bun/updo”,
“color”: “Dark brown/brunette”,
“texture”: “Fine but voluminous”,
“details”: “Numerous loose flyaways framing the face”
},
“pose”: {
“head_position”: “Front-facing”,
“body_position”: “Shoulders slightly angled, arms crossed”,
“energy”: “Casual, thinking, daydreaming”
}
}
}
2. Attire(服装)
详细描述主体的穿着。
{
“attire”: {
“top”: {
“item”: “Chunky knit sweater”,
“color”: “Heather grey”,
“texture”: “Heavy wool or cotton yarn, visible ribbed collar”,
“fit”: “Oversized, cozy”
},
“bottom”: {
“item”: “High-waisted jeans”,
“color”: “Vintage blue”,
“texture”: “Washed denim with slight distressing”
},
“accessories”: {
“jewelry”: “Minimal gold hoop earrings”,
“other”: “Leather watch on left wrist”
}
}
}
3. Photography(摄影技术)
控制图像的技术方面。
{
“photography”: {
“style”: “High-key studio portrait”,
“shot_scale”: “Medium close-up (chest up)”,
“lighting”: {
“type”: “Soft, diffuse studio lighting”,
“source”: “Large softbox frontal/overhead”,
“details”: “Prominent catchlights in upper pupils, soft shadowing under chin”
},
“camera_gear”: {
“lens”: “85mm Portrait Lens”,
“aperture”: “f/2.8”,
“focus”: “Sharp focus on eyes, slight fall-off on edges”
},
“post_processing”: {
“look”: “Commercial clean aesthetic”,
“grading”: “Neutral cool tones, true-to-life colors”
}
}
}
4. Background(背景)
定义环境和背景元素。
{
“background”: {
“type”: “Studio backdrop”,
“color”: “Solid light grey”,
“texture”: “Smooth, featureless”,
“depth”: “Flat, non-distracting”,
“elements”: [“Soft gradient”, “No visible seams”]
}
}
5. Aesthetic Fidelity(美学保真度)
设定整体视觉风格和氛围。
{
“aesthetic_fidelity”: {
“medium”: “Digital Photography”,
“vibe”: “Minimalist, clean, cozy, introspection”,
“visual_qualities”: [
“High resolution”,
“Sharp details”,
“Soft color palette”,
“Textural contrast”
],
“references”: [“Vogue editorial”, “Apple product photography”]
}
}
6. Constraints(约束条件)
明确指定必须保持和必须避免的内容。
{
“constraints”: {
“must_keep”: [
“Upward gaze”,
“Messy hair flyaways”,
“Grey knit texture”,
“Blue eye color”,
“Studio grey background”
],
“avoid”: [
“Smiling with teeth”,
“Direct eye contact with camera”,
“Complex background”,
“Harsh shadows”,
“Jewelry”,
“Glasses”
]
}
}
7. Negative Prompt(负面提示词)
列出不希望出现的元素。
{
“negative_prompt”: [
“teeth”,
“smile”,
“looking at camera”,
“dark background”,
“patterned background”,
“jewelry”,
“earrings”,
“glasses”,
“low resolution”,
“blurry eyes”,
“overexposed”,
“heavy makeup”
]
}
实战教程:创建你的第一个 JSON 提示词
步骤 1:确定你的视觉目标
在开始编写 JSON 之前,先明确你想要什么:
-
主体是什么?(人物、产品、风景等)
-
整体氛围?(温馨、专业、神秘、活力等)
-
技术风格?(摄影、绘画、3D 渲染等)
步骤 2:使用基础模板
从以下通用模板开始:
{
“subject”: {
“main”: “[主要对象描述]”,
“details”: {
“appearance”: “[外观特征]”,
“pose”: “[姿势/位置]”
}
},
“environment”: {
“setting”: “[场景设置]”,
“lighting”: “[光线条件]”,
“atmosphere”: “[氛围描述]”
},
“technical”: {
“medium”: “[媒介类型:摄影/绘画/3D等]”,
“style”: “[艺术风格]”,
“quality”: “[质量要求]”
},
“constraints”: {
“must_keep”: [“[必须包含的元素]”],
“avoid”: [“[必须避免的元素]”]
},
“negative_prompt”: [“[负面关键词]”]
}
步骤 3:逐步填充细节
示例:创建一个"咖啡馆工作的自由职业者"场景
{
“subject”: {
“main”: “Young professional woman working on laptop”,
“demographics”: {
“age”: “Late 20s”,
“gender”: “Female”
},
“appearance”: {
“hair”: “Shoulder-length wavy brown hair”,
“clothing”: “Casual smart: white blouse, light cardigan”,
“expression”: “Focused, slightly smiling”
},
“pose”: {
“position”: “Sitting at wooden table”,
“action”: “Typing on MacBook Pro”,
“posture”: “Relaxed but professional”
}
},
“environment”: {
“setting”: “Modern specialty coffee shop”,
“elements”: [
“Exposed brick wall”,
“Large windows with natural light”,
“Potted plants on shelves”,
“Minimalist furniture”
],
“time_of_day”: “Late morning”,
“lighting”: “Soft natural light from windows, warm ambient lighting”
},
“technical”: {
“medium”: “Lifestyle photography”,
“camera”: {
“lens”: “35mm”,
“aperture”: “f/2.0”
},
“style”: “Clean, authentic, Instagram-worthy”,
“quality”: “High resolution, sharp focus on subject”
},
“constraints”: {
“must_keep”: [
“MacBook visible”,
“Coffee cup on table”,
“Natural window light”,
“Warm color tones”
],
“avoid”: [
“Other people in focus”,
“Cluttered background”,
“Harsh shadows”,
“Artificial poses”
]
},
“negative_prompt”: [
“crowded”,
“noisy”,
“dark”,
“blurry”,
“overexposed”,
“stock photo look”
]
}
步骤 4:测试和迭代
-
首次生成:使用完整的 JSON 生成图像
-
分析结果:检查哪些方面符合预期,哪些需要调整
-
精确修改:只修改需要调整的字段
-
重复测试:直到获得满意结果
步骤 5:保存为可复用模板
将成功的 JSON 保存为模板,方便未来使用:
{
“template_name”: “Coffee Shop Professional”,
“version”: “1.0”,
“description”: “Lifestyle photography of professional working in cafe”,
“tags”: [“lifestyle”, “professional”, “cafe”, “remote work”],
“json”: { /* 完整的 JSON 内容 */ }
}
高级技巧:从图片生成 JSON 提示词
方法:使用 Gemini 3 Pro 进行图像分析
这是获取高质量 JSON 提示词的最简单方法。Gemini 3 Pro 具有出色的视觉理解能力。
系统提示词模板
You are an expert prompt engineer for Nano Banana Pro.
Your task is to convert the user’s description into a sophisticated, EXTREMELY DETAILED JSON prompt.
You must output a single valid JSON object.
JSON STRUCTURE GUIDELINES:
- Dynamic Fields: You are encouraged to ADD new fields that capture specific details about the subject (e.g., “plating_style” for food, “architecture_era” for buildings, “glitch_patterns” for abstract art).
- Remove Irrelevant Fields: Do NOT include fields that don’t apply. If the subject is a stove, do not include “hair”, “skin”, or “pose”. Remove them entirely rather than setting them to “N/A”.
- Subject Specificity:
- For People: The example structure (subject, face, skin, hair, clothing) is excellent. Keep it.
- For Non-Humans: Create a structure that fits the object. For example, a car might have “chassis”, “paint_finish”, “wheels”.
- Standard Fields: Always include “constraints” (with “must_keep” and “avoid” lists) and “negative_prompt”.
AESTHETIC GOALS:
- Medium Specificity: If the user asks for a specific style (e.g. “oil painting”), describe the brushwork, canvas texture, and drying cracks.
- Lighting: Be precise (soft, hard, volumetric, golden hour, studio, rim lighting).
- Camera: (focal length, depth of field) - ONLY if the style requires photorealism.
Return ONLY the raw JSON string.
工作流程
-
上传参考图片到 Gemini 3 Pro
-
使用上述系统提示词
-
让 Gemini 分析图像并生成 JSON
-
审查和修改生成的 JSON
-
在 Nano Banana Pro 中使用
实际案例
输入图像:一只手拿着 iPhone,屏幕上显示照片库,背景是温馨的客厅。
生成的 JSON(简化版):
{
“subject”: {
“main”: “First-person POV hand holding a smartphone”,
“hand_details”: {
“appearance”: “Male hand, light skin tone”,
“grip”: “Holding an iPhone vertically, thumb hovering near bottom bezel”,
“lighting”: “Illuminated by screen’s glow and warm ambient room light”
},
“device”: {
“type”: “Modern smartphone with notch (iPhone style)”,
“case”: “Black slim case”,
“screen_state”: “On, displaying photo gallery app”
}
},
“screen_content”: {
“interface”: {
“app”: “iOS Photos App”,
“header”: “Time ‘22:22’, Title ‘For You’”
},
“image_grid”: {
“layout”: “3-column grid of thumbnails”,
“subject_matter”: “Photos of woman with curly dark hair in various poses”
}
},
“environment”: {
“setting”: “Cozy living room at night”,
“elements”: {
“left”: “Tall bookshelf with colorful books, mushroom lamp”,
“center”: “Wall-mounted TV showing cityscape”,
“decor”: “Snorlax and Eevee plush toys on side table”
}
},
“photography”: {
“style”: “POV Lifestyle Snapshot”,
“focus”: “Sharp on smartphone screen, shallow depth of field on background”,
“lighting”: {
“type”: “Low-light indoor ambience”,
“sources”: [“Warm tungsten lamp”, “Cool TV glow”, “Bright phone display”]
}
},
“constraints”: {
“must_keep”: [
“Snorlax plushie with sunglasses”,
“Time on phone 22:22”,
“Grid of photos on screen”,
“Warm bookshelf lighting”
],
“avoid”: [“Daylight”, “Empty screen”, “Blurry phone screen”]
},
“negative_prompt”: [“daylight”, “empty room”, “cartoon filter”, “illustration”]
}
修改和变体
一旦有了基础 JSON,你可以轻松创建变体:
2000年代早期版本:
-
将 iPhone 改为 BlackBerry
-
将平板电视改为 CRT 电视
-
添加胶片颗粒感
-
改变光线为钨丝灯
{
“subject”: {
“device”: {
“type”: “Early 2000s BlackBerry (model 7230)”,
“body”: “Bulky blue/black plastic casing with side trackwheel”
}
},
“environment”: {
“center_right”: {
“electronics”: “Bulky CRT TV displaying music video channel with scanlines”
}
},
“photography”: {
“camera_gear”: {
“film_stock”: “Kodak Gold 400 (grainy, warm tones)”
}
}
}
批量生成变体的方法
技术 1:使用前缀指令
在 JSON 提示词前添加变体生成指令:
Generate a new image with SIGNIFICANTLY different nouns, objects, color palette and pose compared to the JSON below. CRITICAL: Strictly preserve the original ‘vibe’, ‘aesthetic’, and ‘mood’. The result should look like a distinct image from the same artistic series.
[INSERT YOUR JSON PROMPT HERE]
技术 2:添加特定修改指令
Additional Instruction: Make it night time, change the season to winter, and add snow.
[INSERT YOUR JSON PROMPT HERE]
技术 3:使用变量替换
创建一个基础模板,然后系统地替换变量:
基础模板:
{
“subject”: { “type”: “[ANIMAL]”, “color”: “[COLOR]” },
“environment”: { “setting”: “[LOCATION]” },
“lighting”: { “type”: “[LIGHTING_TYPE]” }
}
变体 1:[ANIMAL]=fox, [COLOR]=red, [LOCATION]=forest, [LIGHTING_TYPE]=golden hour
变体 2:[ANIMAL]=wolf, [COLOR]=grey, [LOCATION]=snowy mountain, [LIGHTING_TYPE]=moonlight
注意事项
⚠️ 权衡:使用自动变体生成时,提示词准确性会下降。生成的图像可能不再精确匹配你的 JSON 描述,但会保持整体美学风格。
JSON vs 散文提示词
对比示例
JSON 版本(结构化):
{
“subject”: {
“name”: “David Duchovny”,
“clothing”: “PSG football kit”,
“accessories”: {
“shoulder”: “Parthenos sylvia butterfly”,
“hand”: “Lily of the valley bouquet”,
“arm_tattoo”: “Nissan Qashqai”,
“earring”: “Dangling monstera leaf”
},
“pose”: “Making OK gesture with left hand, wearing gold ring”
},
“background”: {
“posters”: [“‘Last Year at Marienbad’ movie”, “Garchomp Pokemon”],
“wall_decor”: “Terminator 2 cross-stitch”,
“window_view”: “Lenticular clouds”,
“furniture”: “High-end decor with NEFF hob”,
“objects”: [
“Jeff Koons sculpture on plinth”,
“Orange Lego Christmas tree”,
“Purple and yellow origami chair”
]
},
“pet”: “Marine iguana”,
“vibe”: “Hastily taken everyday photo”,
“style”: “Casual, authentic snapshot”
}
散文版本(自然语言):
A portrait photo of david duchovny, he is wearing a PSG kit, on his shoulder is a parthenos sylvia, in his hand he is holding lily of the valley, behind him there is a poster for the movie ‘Last Year at Marienbad’ and a Garchomp poster. On a plinth there is a jeff koons sculpture. The sky through the window shows lenticular clouds. There is also a Terminator 2 cross-stitch on the wall. On his arm there is a tattoo of a Nissan Qashqai. There’s an orange lego Christmas tree. The room has a high end decor, a pot is boiling on the NEFF hob. Next to the pot is a little origami chair made from purple and yellow paper. He has a dangling monstera leaf earring. There is a pet marine iguana. In his other hand he is making the ok shape. On that hand there is a gold ring. The whole image gives the impression of a hastily taken photo, an everyday scene.
两种格式的优缺点
| 方面 | JSON | 散文 |
|---|---|---|
| 可读性 | 结构化,易于扫描 | 自然流畅 |
| 编辑性 | 精确修改特定部分 | 需要重写句子 |
| 复杂度管理 | 优秀的层次组织 | 长文本容易混乱 |
| 工具支持 | 易于程序化操作 | 需要 NLP 解析 |
| 学习曲线 | 需要理解 JSON 语法 | 直观自然 |
| 团队协作 | 标准化,可版本控制 | 自由灵活 |
转换方法
你可以使用语言模型在两种格式之间转换:
JSON → 散文:
Given this JSON, keep all the details and convert it to prose. Use only paragraphs. Be concise.
散文 → JSON:
Convert this description into a detailed JSON prompt following Nano Banana Pro best practices. Include all relevant sections: subject, environment, technical, constraints, and negative_prompt.
混合提示词策略
概念
你不需要在 JSON 和散文之间二选一。混合提示词结合了两者的优势:
-
开始用散文描述整体概念和创意方向
-
添加 JSON 片段来控制特定的技术细节
-
不需要有效的 JSON - Nano Banana Pro 会处理混合格式
实际示例
A beautiful portrait of a young woman in a sunflower field during golden hour. She has long flowing hair and is wearing a summer dress. The mood is peaceful and dreamy.
{
“photography”: {
“camera”: “Canon EOS R5”,
“lens”: “85mm f/1.2”,
“aperture”: “f/1.8”,
“lighting”: “Golden hour backlight with reflector fill”
},
“post_processing”: {
“style”: “Warm film emulation”,
“grain”: “Subtle 35mm grain”
}
}
The background should have bokeh circles from the sunlight filtering through the flowers.
使用场景
-
快速原型:先用散文快速迭代概念
-
精确控制:对关键元素使用 JSON
-
团队协作:散文部分易于讨论,JSON 部分精确执行
-
模板复用:将常用的 JSON 片段(如相机设置)复用到不同项目
最佳实践
-
散文描述整体,JSON 控制细节
-
保持 JSON 片段独立,不要与散文句子混合
-
使用注释(如果平台支持)来标记不同部分的目的
-
测试组合效果,确保两部分协调一致
实用模板库
模板 1:专业人像摄影
{
“template_name”: “Professional Portrait”,
“subject”: {
“demographics”: { “age”: “30s”, “gender”: “Any” },
“attire”: {
“style”: “Business casual”,
“colors”: “Neutral tones”
},
“expression”: “Confident, approachable”,
“pose”: “Three-quarter angle, relaxed shoulders”
},
“photography”: {
“style”: “Corporate headshot”,
“lighting”: {
“type”: “Three-point lighting”,
“key”: “Soft 45-degree angle”,
“fill”: “Gentle shadow reduction”,
“backlight”: “Subtle rim light”
},
“camera”: {
“lens”: “85mm”,
“aperture”: “f/4.0”,
“distance”: “6 feet”
},
“background”: {
“type”: “Seamless paper”,
“color”: “Neutral grey or white”,
“blur”: “Slight gaussian blur”
}
},
“constraints”: {
“must_keep”: [“Professional appearance”, “Sharp eyes”],
“avoid”: [“Harsh shadows”, “Distracting background”, “Overly casual pose”]
}
}
模板 2:产品摄影
{
“template_name”: “Product Photography”,
“subject”: {
“type”: “[PRODUCT_TYPE]”,
“material”: “[MATERIAL]”,
“finish”: “[FINISH]”,
“position”: “Centered, slight angle for depth”
},
“lighting”: {
“type”: “Studio product lighting”,
“main”: “Large softbox overhead”,
“fill”: “White reflector cards”,
“accent”: “Small spotlight for highlights”
},
“background”: {
“type”: “Gradient”,
“style”: “Clean, minimal”,
“colors”: [“White to light grey”]
},
“technical”: {
“camera”: “Medium format”,
“aperture”: “f/8 to f/11”,
“focus”: “Stacked focus for full sharpness”
},
“post_processing”: {
“retouching”: “Dust and imperfection removal”,
“color”: “True to life, slight saturation boost”
}
}
模板 3:风景摄影
{
“template_name”: “Landscape Photography”,
“scene”: {
“type”: “[LANDSCAPE_TYPE]”,
“elements”: [“[ELEMENT_1]”, “[ELEMENT_2]”, “[ELEMENT_3]”],
“time”: “[TIME_OF_DAY]”,
“weather”: “[WEATHER_CONDITION]”
},
“composition”: {
“rule”: “Rule of thirds”,
“foreground”: “[FOREGROUND_ELEMENT]”,
“midground”: “[MIDGROUND_ELEMENT]”,
“background”: “[BACKGROUND_ELEMENT]”,
“leading_lines”: “[LEADING_LINES]”
},
“lighting”: {
“type”: “Natural”,
“quality”: “[LIGHT_QUALITY]”,
“direction”: “[LIGHT_DIRECTION]”
},
“camera”: {
“lens”: “Wide angle 16-35mm”,
“aperture”: “f/8 to f/16”,
“filter”: “ND or polarizing if needed”
},
“mood”: “[DESIRED_MOOD]”
}
模板 4:美食摄影
{
“template_name”: “Food Photography”,
“subject”: {
“dish”: “[DISH_NAME]”,
“cuisine”: “[CUISINE_TYPE]”,
“presentation”: “[PLATING_STYLE]”,
“garnish”: “[GARNISH_ELEMENTS]”
},
“props”: {
“surface”: “[TABLE_SURFACE]”,
“utensils”: “[UTENSILS]”,
“linens”: “[NAPKINS/CLOTH]”,
“additional”: [“[PROP_1]”, “[PROP_2]”]
},
“lighting”: {
“type”: “Natural window light”,
“direction”: “Side or back lighting”,
“quality”: “Soft, diffused”,
“shadows”: “Gentle, not harsh”
},
“camera”: {
“angle”: “45-degree or overhead”,
“lens”: “50mm or macro”,
“aperture”: “f/2.8 to f/5.6”
},
“style”: {
“aesthetic”: “[RUSTIC/MODERN/MINIMAL]”,
“color_palette”: “[COLOR_SCHEME]”,
“mood”: “Appetizing, fresh”
}
}
常见问题与最佳实践
常见问题
Q: JSON 提示词必须严格有效吗? A: 不需要。Nano Banana Pro 会解析结构,即使 JSON 有轻微错误也能工作。但保持有效 JSON 有助于工具处理和团队协作。
Q: 可以省略某些字段吗? A: 可以。只包含与你的图像相关的字段。例如,拍摄风景时不需要 “face” 或 “hair” 字段。
Q: 字段名称必须完全匹配吗? A: 不需要。模型理解语义,你可以根据需要使用描述性的字段名。
Q: JSON 提示词比文本提示词长很多,值得吗? A: 当你需要精确控制时,绝对值得。对于快速探索,简单文本提示词仍然有效。
最佳实践清单
- [ ] 从模板开始,不要从零开始写
- [ ] 使用具体描述,避免模糊词汇
- [ ] 保持一致性,同类对象使用相似结构
- [ ] 包含约束条件,明确指定必须保持和避免的内容
- [ ] 使用负面提示词,排除不想要的元素
- [ ] 测试和迭代,不要期望第一次就完美
- [ ] 保存成功的模板,建立个人库
- [ ] 记录版本,跟踪哪些修改有效
- [ ] 分享和学习,从社区获取灵感
- [ ] 混合使用,根据场景选择 JSON 或散文
进阶技巧
-
使用变量系统:在 JSON 中使用占位符如
[COLOR]、[MOOD],便于批量生成 -
创建组件库:将常用的 lighting、camera 设置保存为可复用片段
-
建立风格指南:为品牌或项目创建一致的 JSON 结构
-
版本控制:使用 Git 管理提示词模板
-
A/B 测试:系统地测试不同字段对结果的影响
总结
JSON 提示词不是每个图像生成的必需品,但当你:
-
有特定的视觉愿景
-
需要精确控制输出
-
使用现有模板
-
进行团队协作
-
追求一致性和可重复性
时,它是一个非常强大的工具。
通过本指南,你已经掌握了:
-
✅ JSON 提示词的核心概念和优势
-
✅ 完整的 JSON 结构和工作原理
-
✅ 从创建到优化的完整工作流程
-
✅ 从图片生成 JSON 的高级技巧
-
✅ 批量生成变体的方法
-
✅ JSON 与散文的灵活运用
-
✅ 实用的模板库
现在,开始创建你的第一个 JSON 提示词吧!记住:实验是关键,不断测试、迭代、学习,你会发现 JSON 提示词的无限可能。
资源与参考
-
JSON 提示词原文:Prompting Nano Banana Pro with JSON
-
AI Studio:Google 的 AI 开发平台,支持 Nano Banana Pro
本指南由 AI 助手基于 fofr.ai 官方文档整理编写,旨在帮助中文用户更好地理解和使用 Nano Banana Pro 的 JSON 提示词功能。