image-create

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

image-create

Use this skill for all pure text-to-image generation requests on Starchild.

Covers: logo design, poster design, illustration, meme creation, game assets, social media content, 3D rendering, educational illustrations, fashion design, food photography, pet illustrations, wedding design, holiday marketing, and artistic style creation.

Core principle: call the provided script. Do not re-implement proxy/billing plumbing.

When to use image-create vs image-portrait:

image-create → user wants to CREATE something from a text description (no face/identity needed)
image-portrait → user wants a portrait with their face/identity preserved from a reference photo

在Starchild上，所有纯text-to-image生成请求都可使用此技能。

涵盖场景：标志设计、海报设计、插画、表情包制作、游戏资产、社交媒体内容、3D渲染、教育插画、时尚设计、美食摄影、宠物插画、婚礼设计、节日营销以及艺术风格创作。

核心原则： 调用提供的脚本，不要重新实现代理/计费相关逻辑。

何时使用image-create vs image-portrait：

image-create → 用户希望通过文本描述创作内容（无需人脸/身份信息）
image-portrait → 用户希望基于参考照片生成保留其人脸/身份的肖像

1. Quick start — basic generation (most common)

1. 快速入门——基础生成（最常用）

python

exec(open('skills/image-create/generate_image.py').read())
result = generate_image(
    prompt="a futuristic city skyline at sunset with flying cars",
)

python

exec(open('skills/image-create/generate_image.py').read())
result = generate_image(
    prompt="a futuristic city skyline at sunset with flying cars",
)

result -> {"success": True, "images": [{"local_path": "output/images/..."}], ...}

undefined

undefined

2. Quick start — with category preset

2. 快速入门——使用分类预设

python

exec(open('skills/image-create/generate_image.py').read())
result = generate_image(
    prompt="StarChild AI platform",
    category="logo",
    style="tech",
)

python

exec(open('skills/image-create/generate_image.py').read())
result = generate_image(
    prompt="StarChild AI platform",
    category="logo",
    style="tech",
)

3. Quick start — category only (no custom prompt)

3. 快速入门——仅使用分类（无自定义提示词）

python

exec(open('skills/image-create/generate_image.py').read())
result = generate_image(
    category="3d",
    style="diorama",
)

python

exec(open('skills/image-create/generate_image.py').read())
result = generate_image(
    category="3d",
    style="diorama",
)

Uses the built-in style template as the full prompt

undefined

undefined

Delivering the result to the user — IMPORTANT

向用户交付结果——重要提示

Never hand the user the raw fal.media URL. fal serves files with restrictive CSP headers. The only reliable delivery path is the already-downloaded local file:

Use each image's
```
local_path
```
(e.g.
```
output/images/xxx.png
```
) — the script always downloads on success.
Tell the user the files are saved to
```
output/images/
```
and viewable in the workspace file panel.
On Web channel, embed inline so the user can preview in chat:
markdown
```
![image](output/images/<filename>.png)
```

On Telegram / WeChat: send via

send_to_telegram(file_path="output/images/...", message_type="image")

send_to_wechat(file_path="output/images/...", message_type="image")

绝不要直接给用户原始的fal.media链接。 fal提供的文件带有严格的CSP头部，唯一可靠的交付方式是已下载的本地文件：

使用每个图像的
```
local_path
```
（例如
```
output/images/xxx.png
```
）——脚本在生成成功后总会自动下载文件。
告知用户文件已保存至
```
output/images/
```
，可在工作区文件面板中查看。
在Web渠道中，将图像嵌入聊天内容以便用户预览：
markdown
```
![image](output/images/<filename>.png)
```

在Telegram/微信渠道：通过

send_to_telegram(file_path="output/images/...", message_type="image")

或

send_to_wechat(file_path="output/images/...", message_type="image")

发送。

4. Parameters

4. 参数

Parameter	Required	Default	Description
`prompt`	yes*	—	Text description of the desired image
`category`	yes*	—	Preset category (see §5)
`style`	no	`"default"`	Sub-style within the category (see §5)
`model`	no	`"nanopro"`	Model: `"nano2"` (fastest ~15s), `"nanopro"` (balanced ~25s, default), or `"gpt"` (best quality ~150s)
`count`	no	`1`	Number of images to generate (1–4)
`aspect_ratio`	no	auto	Output ratio: `1:1` , `2:3` , `3:2` , `3:4` , `4:3` , `4:5` , `5:4` , `9:16` , `16:9` , `21:9` . Auto-selected by category if not set.
`image_path`	no	—	Optional local file path for reference/inspiration image
`image_url`	no	—	Optional public URL for reference/inspiration image

*At least one of

prompt

category

must be provided.

Prompt priority:

prompt + category/style

(enhanced) >

prompt

only >

category + style

category

default.

Aspect ratio auto-selection: When not explicitly set, the script picks the best ratio for the category (e.g.,

1:1

for logos,

3:4

for posters,

16:9

for banners).

参数	必填	默认值	描述
`prompt`	是*	—	所需图像的文本描述
`category`	是*	—	预设分类（见第5节）
`style`	否	`"default"`	分类下的子风格（见第5节）
`model`	否	`"nanopro"`	模型选项： `"nano2"` （最快约15秒）、 `"nanopro"` （平衡型约25秒，默认）或 `"gpt"` （最高质量约150秒）
`count`	否	`1`	生成图像的数量（1–4）
`aspect_ratio`	否	auto	输出比例： `1:1` 、 `2:3` 、 `3:2` 、 `3:4` 、 `4:3` 、 `4:5` 、 `5:4` 、 `9:16` 、 `16:9` 、 `21:9` 。若未设置，将根据分类自动选择。
`image_path`	否	—	用于参考/灵感的可选本地文件路径
`image_url`	否	—	用于参考/灵感的可选公共URL

*必须提供

prompt

或

category

中的至少一个。

提示词优先级：

prompt + category/style

（增强版） > 仅

prompt

category + style

category

默认值。

比例自动选择规则： 若未明确设置，脚本会为分类选择最佳比例（例如标志用

1:1

，海报用

3:4

，横幅用

16:9

）。

5. Category and style presets

5. 分类与风格预设

E: Design — Logo (

category="logo"

)

E: 设计——标志（

category="logo"

）

⚠️ AI cannot reliably render text. Generate the icon/symbol only; add text/wordmark in a design tool (Figma, Canva, Illustrator).

Style	Key	Best for
Abstract geometric	`abstract`	Tech, conceptual brands (Nike swoosh style)
Pictorial icon	`pictorial`	Universal brands, works without text (Apple style)
Mascot character	`mascot`	Friendly brands, food, sports (KFC style)
Tech company	`tech`	SaaS, AI, fintech startups
Food brand	`food`	Restaurant, bakery, organic
Fashion brand	`fashion`	Luxury, apparel, beauty
Gaming	`gaming`	Esports, game studio
General	`default`	Any professional logo

Logo prompt anti-patterns (avoid):

❌
```
"logo with text 'Company Name'"
```
— text will be garbled
❌
```
"photorealistic logo"
```
— logos aren't photos
❌
```
"3D rendered logo"
```
— too complex, won't scale down

✅

"flat vector logo of [subject], minimal geometric style, single color, white background"

⚠️ AI无法可靠渲染文字。 仅生成图标/符号；文字/标志名称需在设计工具（Figma、Canva、Illustrator）中添加。

风格	标识	适用场景
抽象几何	`abstract`	科技类、概念品牌（类似Nike swoosh风格）
具象图标	`pictorial`	通用品牌，无需文字即可识别（类似Apple风格）
吉祥物角色	`mascot`	友好型品牌、美食、运动类（类似KFC风格）
科技公司	`tech`	SaaS、AI、金融科技初创企业
美食品牌	`food`	餐厅、烘焙店、有机食品品牌
时尚品牌	`fashion`	奢侈品、服饰、美妆品牌
游戏类	`gaming`	电竞、游戏工作室
通用风格	`default`	任何专业标志

标志提示词反模式（需避免）：

❌
```
"logo with text 'Company Name'"
```
——文字会出现错乱
❌
```
"photorealistic logo"
```
——标志并非照片
❌
```
"3D rendered logo"
```
——过于复杂，无法缩小使用

✅

"flat vector logo of [subject], minimal geometric style, single color, white background"

E: Design — Poster (

category="poster"

)

E: 设计——海报（

category="poster"

）

Style	Key	Best for	Recommended ratio
Movie poster	`movie`	Film promotion, cinematic	3:4
Music festival	`music_festival`	Concerts, festivals	3:4
Tech conference	`tech_conference`	Tech events, hackathons	3:4
Travel destination	`travel`	Tourism, wanderlust	3:4
Product launch	`product_launch`	Product announcements	3:4
Minimalist art	`minimalist`	Home decor, gallery	3:4
Sports event	`sports`	Athletic events	3:4
General	`default`	Any poster	3:4

风格	标识	适用场景	推荐比例
电影海报	`movie`	电影宣传、影视风格	3:4
音乐节海报	`music_festival`	演唱会、音乐节	3:4
科技会议海报	`tech_conference`	科技活动、黑客松	3:4
旅游目的地海报	`travel`	旅游推广、旅行主题	3:4
产品发布海报	`product_launch`	产品发布公告	3:4
极简艺术海报	`minimalist`	家居装饰、画廊展示	3:4
体育赛事海报	`sports`	体育活动	3:4
通用风格	`default`	任何海报	3:4

E: Design — Illustration (

category="illustration"

)

E: 设计——插画（

category="illustration"

）

Style	Key	Best for
Fantasy	`fantasy`	Fantasy worlds, magic, dragons
Sci-fi	`scifi`	Futuristic scenes, space
Children's book	`children`	Kids content, storybooks (ages 3-5)
Editorial	`editorial`	Magazine, article headers
Botanical	`botanical`	Scientific plant illustrations
General	`default`	Any illustration

风格	标识	适用场景
奇幻风格	`fantasy`	奇幻世界、魔法、龙类主题
科幻风格	`scifi`	未来场景、太空主题
儿童绘本风格	`children`	儿童内容、故事书（适合3-5岁）
社论插画	`editorial`	杂志、文章标题配图
植物插画	`botanical`	科学类植物插画
通用风格	`default`	任何插画

E: Design — Meme (

category="meme"

)

E: 设计——表情包（

category="meme"

）

Style	Key	Best for
Animal meme	`animal`	Cute/funny animal memes
Reaction	`reaction`	Reaction templates
Surreal	`surreal`	Absurdist internet humor
General	`default`	Any meme

风格	标识	适用场景
动物表情包	`animal`	可爱/搞笑动物表情包
反应类表情包	`reaction`	反应模板类
超现实风格	`surreal`	荒诞网络幽默风格
通用风格	`default`	任何表情包

K: Game Assets (

category="game_asset"

)

K: 游戏资产（

category="game_asset"

）

Style	Key	Best for	Common sizes
Character concept	`character`	RPG characters, heroes	1024x1024
Environment	`environment`	Game worlds, levels	1920x1080
Weapon/prop	`weapon`	Items, weapons, artifacts	1024x1024
UI icon	`ui_icon`	Game UI, mobile icons	32x32 to 128x128
Pixel sprite	`pixel_sprite`	Retro game characters	32x32 to 64x64
Tileset	`tileset`	Seamless environment tiles	256x256 to 512x512
General	`default`	Any game asset	1024x1024

风格	标识	适用场景	常见尺寸
角色概念设计	`character`	RPG角色、英雄	1024x1024
游戏环境	`environment`	游戏世界、关卡	1920x1080
武器/道具	`weapon`	物品、武器、艺术品	1024x1024
UI图标	`ui_icon`	游戏UI、移动应用图标	32x32至128x128
像素精灵	`pixel_sprite`	复古游戏角色	32x32至64x64
瓦片集	`tileset`	无缝环境瓦片	256x256至512x512
通用风格	`default`	任何游戏资产	1024x1024

M: Social Media (

category="social_media"

)

M: 社交媒体（

category="social_media"

）

Style	Key	Best for	Recommended ratio	Resolution
Instagram post	`instagram`	IG feed posts	1:1	1080x1080
小红书	`xiaohongshu`	Xiaohongshu posts	3:4	1080x1440
TikTok cover	`tiktok_cover`	TikTok thumbnails	9:16	1080x1920
YouTube thumbnail	`youtube_thumbnail`	YT thumbnails	16:9	1280x720
Banner	`banner`	Twitter/YouTube banners	16:9	1920x1080
Story	`story`	IG/FB stories	9:16	1080x1920
General	`default`	Any social content	1:1	1024x1024

风格	标识	适用场景	推荐比例	分辨率
Instagram帖子	`instagram`	IG动态帖子	1:1	1080x1080
小红书	`xiaohongshu`	小红书帖子	3:4	1080x1440
TikTok封面	`tiktok_cover`	TikTok缩略图	9:16	1080x1920
YouTube缩略图	`youtube_thumbnail`	YT缩略图	16:9	1280x720
横幅	`banner`	Twitter/YouTube横幅	16:9	1920x1080
故事模板	`story`	IG/FB故事	9:16	1080x1920
通用风格	`default`	任何社交媒体内容	1:1	1024x1024

N: 3D (

category="3d"

)

N: 3D（

category="3d"

）

Style	Key	Best for
3D character	`character`	Pixar-style characters
Product render	`product`	Product visualization
Diorama	`diorama`	Miniature scenes, isometric
App icon	`icon`	iOS/Android app icons
3D text	`text`	Chrome/metallic text
3D scene	`scene`	Low-poly environments
General	`default`	Any 3D render

风格	标识	适用场景
3D角色	`character`	皮克斯风格角色
产品渲染	`product`	产品可视化
微缩场景	`diorama`	迷你场景、等距视角
应用图标	`icon`	iOS/Android应用图标
3D文字	`text`	镀铬/金属质感文字
3D场景	`scene`	低多边形环境
通用风格	`default`	任何3D渲染

P: Education (

category="education"

)

P: 教育（

category="education"

）

Style	Key	Best for
Textbook	`textbook`	Textbook illustrations
Infographic	`infographic`	Data visualization
Science	`science`	Scientific diagrams, anatomy
History	`history`	Historical scene reconstruction
Diagram	`diagram`	Technical process diagrams
General	`default`	Any educational content

风格	标识	适用场景
教材插画	`textbook`	教材配图
信息图	`infographic`	数据可视化
科学类	`science`	科学图表、解剖图
历史类	`history`	历史场景还原
流程图	`diagram`	技术流程图表
通用风格	`default`	任何教育内容

Q: Fashion (

category="fashion"

)

Q: 时尚（

category="fashion"

）

Style	Key	Best for
Clothing design	`clothing`	Garment sketches, fashion plates
Accessory	`accessory`	Jewelry, bags, shoes
Nail art	`nail_art`	Nail designs
Textile pattern	`textile`	Fabric patterns, surface design
General	`default`	Any fashion design

风格	标识	适用场景
服装设计	`clothing`	服装草图、时尚版型图
配饰设计	`accessory`	珠宝、包袋、鞋履
美甲设计	`nail_art`	美甲样式
纺织图案	`textile`	面料图案、表面设计
通用风格	`default`	任何时尚设计

R: Food (

category="food"

)

R: 美食（

category="food"

）

Style	Key	Best for
Dish photo	`dish`	Food photography, editorial
Menu design	`menu`	Restaurant menus
Packaging	`packaging`	Food packaging design
Recipe card	`recipe_card`	Recipe illustrations
General	`default`	Any food content

风格	标识	适用场景
菜品摄影	`dish`	美食摄影、社论配图
菜单设计	`menu`	餐厅菜单
包装设计	`packaging`	食品包装设计
食谱卡片	`recipe_card`	食谱插画
通用风格	`default`	任何美食内容

S: Pet (

category="pet"

)

S: 宠物（

category="pet"

）

Style	Key	Best for
Humanized pet	`humanized`	Pets in human clothes
Renaissance	`renaissance`	Royal/regal pet portraits
Cartoon	`cartoon`	Disney/Pixar style pets
Merchandise	`merchandise`	Pet-themed product patterns
Memorial	`memorial`	Pet tribute artwork
General	`default`	Cute pet illustrations

风格	标识	适用场景
拟人化宠物	`humanized`	穿着人类服装的宠物
文艺复兴风格	`renaissance`	皇家/贵族风格宠物肖像
卡通风格	`cartoon`	迪士尼/皮克斯风格宠物
宠物周边图案	`merchandise`	宠物主题产品图案
纪念风格	`memorial`	宠物纪念艺术作品
通用风格	`default`	可爱宠物插画

I: Product Photography (

category="product"

)

I: 产品摄影（

category="product"

）

Derived from product-photography skill best practices.

Style	Key	Best for
Hero shot	`hero`	Primary product image, magazine ads
Packshot	`packshot`	E-commerce listings, Amazon (pure white bg)
Lifestyle	`lifestyle`	Product in context, editorial
Flat lay	`flat_lay`	Instagram, top-down arrangement
General	`default`	Any product photo

Product photography tips:

Hero shot: product fills 80% of frame, slight 15-30° angle for dimension
Packshot (Amazon): pure white background, product fills 85%+, no props/text/watermarks
Always specify lighting: "soft studio lighting", "dramatic rim lighting"
For e-commerce: "sharp focus", "no shadows" or "subtle shadow only"

源自产品摄影技能的最佳实践。

风格	标识	适用场景
主图拍摄	`hero`	产品主图、杂志广告
包装图拍摄	`packshot`	电商列表、亚马逊平台（纯白背景）
场景化拍摄	`lifestyle`	产品实景展示、社论配图
平铺拍摄	`flat_lay`	Instagram、俯视构图
通用风格	`default`	任何产品摄影

产品摄影技巧：

主图：产品占据画面80%，轻微15-30°角度以体现立体感
包装图（亚马逊）：纯白背景，产品占据画面85%以上，无道具/文字/水印
务必指定光线："柔和工作室光线"、"戏剧性轮廓光"
电商场景："清晰对焦"、"无阴影"或"仅轻微阴影"

Y: Wedding (

category="wedding"

)

Y: 婚礼（

category="wedding"

）

Style	Key	Best for	Format
Classic invitation	`invitation`	Floral elegant invitations	5x7 inches
Modern invitation	`invitation_modern`	Minimalist invitations	5x7 inches
Rustic invitation	`invitation_rustic`	Bohemian invitations	5x7 inches
Venue preview	`venue`	Venue decoration preview	—
Save the date	`save_the_date`	Pre-announcement cards	4x6 inches
General	`default`	Any wedding design	—

风格	标识	适用场景	格式
经典请柬	`invitation`	花卉风格优雅请柬	5x7英寸
现代请柬	`invitation_modern`	极简风格请柬	5x7英寸
乡村风格请柬	`invitation_rustic`	波西米亚风格请柬	5x7英寸
场地预览	`venue`	婚礼场地布置预览	—
日期预告卡	`save_the_date`	婚礼预告卡	4x6英寸
通用风格	`default`	任何婚礼设计	—

Z: Holiday Marketing (

category="holiday"

)

Z: 节日营销（

category="holiday"

）

Style	Key	Best for
Christmas card	`christmas_card`	Christmas greetings
Chinese New Year	`chinese_new_year`	春节 designs
New Year	`new_year`	New Year celebrations
Valentine's	`valentines`	Valentine's Day
Halloween	`halloween`	Halloween designs
Promotional	`promotional`	Sale banners, promos
Mid-Autumn	`mid_autumn`	中秋节 designs
General	`default`	Any holiday content

Major holiday calendar for campaign planning:

Holiday	Timing	Best for
Chinese New Year	Jan-Feb	Gifts, family, food
Valentine's Day	Feb 14	Romance, gifts
Women's Day	Mar 8	Empowerment, gifts
520 (5/20)	May 20	Romance (Chinese Valentine's)
618 Shopping	June	E-commerce sales
Qixi (七夕)	Jul-Aug	Romance
Mid-Autumn	Sept	Family, mooncakes
National Day	Oct 1	Travel, shopping
11.11 Singles' Day	Nov 11	Major sales
12.12 Double 12	Dec 12	Year-end sales
Christmas	Dec 25	Gifts, winter

风格	标识	适用场景
圣诞贺卡	`christmas_card`	圣诞祝福
春节	`chinese_new_year`	春节设计
新年	`new_year`	新年庆祝
情人节	`valentines`	情人节
万圣节	`halloween`	万圣节设计
促销风格	`promotional`	促销横幅、推广内容
中秋节	`mid_autumn`	中秋节设计
通用风格	`default`	任何节日内容

营销活动规划主要节日日历：

节日	时间	适用场景
春节	1-2月	礼品、家庭、美食
情人节	2月14日	浪漫、礼品
妇女节	3月8日	女性赋能、礼品
520	5月20日	浪漫（中国情人节）
618购物节	6月	电商促销
七夕	7-8月	浪漫
中秋节	9月	家庭、月饼
国庆节	10月1日	旅游、购物
双11光棍节	11月11日	大型促销
双12	12月12日	年终促销
圣诞节	12月25日	礼品、冬季主题

C: Art Style (

category="art_style"

)

C: 艺术风格（

category="art_style"

）

Style	Key	Best for
Studio Ghibli	`ghibli`	Ghibli-style scenes
American comic	`american_comic`	Marvel/DC style
Japanese manga	`manga`	Manga illustrations
Pixel art	`pixel_art`	Retro game style
Pencil sketch	`pencil_sketch`	Hand-drawn look
3D cartoon	`3d_cartoon`	Pixar/Disney style
Steampunk	`steampunk`	Victorian sci-fi
Fantasy magic	`fantasy_magic`	Magical scenes
Wuxia/Xianxia	`wuxia`	Chinese martial arts
Sci-fi space	`scifi_space`	Space scenes
Pop art	`pop_art`	Warhol style
Ukiyo-e	`ukiyo_e`	Japanese woodblock
Impressionist	`impressionist`	Monet/Renoir style
Art Nouveau	`art_nouveau`	Mucha style
General	`default`	Any artistic style

风格	标识	适用场景
吉卜力风格	`ghibli`	吉卜力风格场景
美式漫画	`american_comic`	漫威/DC风格
日本漫画	`manga`	漫画插画
像素艺术	`pixel_art`	复古游戏风格
铅笔素描	`pencil_sketch`	手绘风格
3D卡通	`3d_cartoon`	皮克斯/迪士尼风格
蒸汽朋克	`steampunk`	维多利亚科幻风格
奇幻魔法	`fantasy_magic`	魔法场景
武侠/仙侠	`wuxia`	中国武侠风格
科幻太空	`scifi_space`	太空场景
波普艺术	`pop_art`	沃霍尔风格
浮世绘	`ukiyo_e`	日本浮世绘风格
印象派	`impressionist`	莫奈/雷诺阿风格
新艺术风格	`art_nouveau`	穆夏风格
通用风格	`default`	任何艺术风格

6. Model selection guide

6. 模型选择指南

Model	Key	Speed	Quality	Best for
NanoPro	`nanopro`	~25s	Good	Default for all requests. Fast iteration.
GPT Image 2	`gpt`	~150s	Best	When user explicitly asks for "highest quality" or "best quality". Text-heavy designs.

Decision rules:

Default: always use
```
nanopro
```
unless the user explicitly requests higher quality.
Use
gpt
when: user says "highest quality", "best quality", "premium", or the design requires precise text rendering (logos with specific text, posters with typography).
Use
nanopro
when: user wants fast results, is iterating on designs, or generating multiple variants.

python

undefined

模型	标识	速度	质量	适用场景
NanoPro	`nanopro`	~25秒	良好	所有请求的默认选项，迭代速度快。
GPT Image 2	`gpt`	~150秒	最佳	用户明确要求“最高质量”或“最佳质量”时使用，适用于文字密集型设计。

决策规则：

默认设置： 除非用户明确要求更高质量，否则始终使用
```
nanopro
```
。
使用
gpt
的场景：用户提及“最高质量”、“最佳质量”、“premium”，或设计需要精确文字渲染（带有特定文字的标志、含排版的海报）。
使用
nanopro
的场景：用户需要快速结果、正在迭代设计，或生成多个变体。

python

undefined

Default (fast)

默认设置（快速）

result = generate_image(prompt="cute cat logo", category="logo")

High quality (user requested)

高质量（用户要求）

result = generate_image(prompt="cute cat logo", category="logo", model="gpt")

---

result = generate_image(prompt="cute cat logo", category="logo", model="gpt")

---

7. Aspect ratio guide

7. 比例指南

Category	Default ratio	Notes
Logo	1:1	Square, scalable
Poster	3:4	Portrait orientation
Illustration	4:3	Landscape, wide scene
Meme	1:1	Square, shareable
Game asset	1:1	Square, consistent
Social media	1:1	Varies by platform
3D	1:1	Square render
Education	4:3	Landscape, readable
Fashion	3:4	Portrait, full garment
Food	4:3	Landscape, appetizing
Pet	1:1	Square, cute
Product	1:1	Square, e-commerce
Wedding	3:4	Portrait, elegant
Holiday	4:3	Landscape, festive
Art style	4:3	Landscape, scenic

Platform-specific overrides (auto-applied):

TikTok cover →
```
9:16
```
(1080x1920)
Instagram story →
```
9:16
```
(1080x1920)
YouTube thumbnail →
```
16:9
```
(1280x720)
Social media banner →
```
16:9
```
(1920x1080)
小红书 →
```
3:4
```
(1080x1440)

The script auto-selects the best ratio when

aspect_ratio

is not explicitly set.

分类	默认比例	说明
标志	1:1	方形，可缩放
海报	3:4	竖版
插画	4:3	横版，宽场景
表情包	1:1	方形，便于分享
游戏资产	1:1	方形，保持一致
社交媒体	1:1	因平台而异
3D	1:1	方形渲染
教育	4:3	横版，易读
时尚	3:4	竖版，展示完整服装
美食	4:3	横版，更有食欲
宠物	1:1	方形，可爱风格
产品	1:1	方形，电商场景
婚礼	3:4	竖版，优雅风格
节日	4:3	横版，节日氛围
艺术风格	4:3	横版，风景场景

平台特定覆盖规则（自动应用）：

TikTok封面 →
```
9:16
```
（1080x1920）
Instagram故事 →
```
9:16
```
（1080x1920）
YouTube缩略图 →
```
16:9
```
（1280x720）
社交媒体横幅 →
```
16:9
```
（1920x1080）
小红书 →
```
3:4
```
（1080x1440）

当未明确设置

aspect_ratio

时，脚本会自动选择最佳比例。

8. Prompt engineering best practices

8. 提示词工程最佳实践

The 5-element prompt structure

五要素提示词结构

Every effective image prompt should include:

[subject/content], [style/aesthetic], [composition/layout], [lighting/atmosphere], [quality modifiers]

每个有效的图像提示词应包含以下要素：

[主体/内容], [风格/美学], [构图/布局], [光线/氛围], [质量修饰词]

Key principles

核心原则

Be specific about the subject:
- ❌ "a logo"
- ✅ "a minimalist cat silhouette logo for a pet grooming business, clean vector style"
Specify the visual style:
- "flat design", "3D render", "watercolor painting", "photorealistic"
- "minimalist", "detailed", "abstract", "geometric"
Include composition guidance:
- "centered composition", "rule of thirds", "symmetrical layout"
- "close-up", "wide shot", "isometric view", "bird's eye view"
Lighting matters:
- Studio: "soft diffused studio lighting", "Rembrandt chiaroscuro"
- Natural: "golden hour warm light", "dappled sunlight through trees"
- Dramatic: "dramatic rim lighting", "volumetric light beams", "neon glow"
- Flat: "even flat lighting with no shadows" (for icons/diagrams)
Quality anchors:
- "professional quality", "print-ready", "4K resolution"
- "octane render", "Unreal Engine quality", "magazine quality"
For text in images (use
gpt
model):
- Explicitly state the text:
```
text reading "SALE 50% OFF"
```
- Specify font style: "bold sans-serif typography", "elegant script font"
- GPT model handles text rendering much better than nanopro

明确主体：
- ❌ "a logo"
- ✅ "a minimalist cat silhouette logo for a pet grooming business, clean vector style"
指定视觉风格：
- "flat design"、"3D render"、"watercolor painting"、"photorealistic"
- "minimalist"、"detailed"、"abstract"、"geometric"
包含构图指导：
- "centered composition"、"rule of thirds"、"symmetrical layout"
- "close-up"、"wide shot"、"isometric view"、"bird's eye view"
光线很重要：
- 工作室光线："soft diffused studio lighting"、"Rembrandt chiaroscuro"
- 自然光："golden hour warm light"、"dappled sunlight through trees"
- 戏剧性光线："dramatic rim lighting"、"volumetric light beams"、"neon glow"
- 平光："even flat lighting with no shadows"（适用于图标/图表）
质量锚点：
- "professional quality"、"print-ready"、"4K resolution"
- "octane render"、"Unreal Engine quality"、"magazine quality"
图像中的文字（使用
gpt
模型）：
- 明确说明文字内容：
```
text reading "SALE 50% OFF"
```
- 指定字体风格："bold sans-serif typography"、"elegant script font"
- GPT模型处理文字渲染的效果远优于nanopro

Logo-specific prompt tips (from logo-design-guide)

标志专属提示词技巧（来自标志设计指南）

Keywords that work:

flat vector logo, simple minimal icon, single color silhouette,
geometric logo mark, clean lines, negative space design,
line art logo, flat design icon, minimalist symbol

Keywords that fail:

❌ photorealistic logo (contradiction — logos aren't photos)
❌ 3D rendered logo (too complex, won't scale down)
❌ gradient logo (inconsistent results, hard to reproduce)
❌ logo with text "Company Name" (text rendering fails)

Prompt structure for logos:

flat vector logo of [subject], [style], [color constraint], [background], [additional detail]

有效关键词：

flat vector logo, simple minimal icon, single color silhouette,
geometric logo mark, clean lines, negative space design,
line art logo, flat design icon, minimalist symbol

无效关键词：

❌ photorealistic logo（矛盾——标志并非照片）
❌ 3D rendered logo（过于复杂，无法缩小使用）
❌ gradient logo（结果不一致，难以复刻）
❌ logo with text "Company Name"（文字渲染失败）

标志提示词结构：

flat vector logo of [subject], [style], [color constraint], [background], [additional detail]

Children's illustration tips (from book-illustrator)

儿童插画技巧（来自书籍插画师指南）

Ages 0-2: Simple, bold, high-contrast, clear shapes
Ages 3-5: Colorful, expressive, engaging characters with movement
Ages 6-8: More detailed scenes with visual storytelling
Ages 9-12: Sophisticated illustrations supporting text
The 3-Color Rule: Limit each character to 3-4 primary colors for visual clarity

0-2岁： 简洁、醒目、高对比度、清晰形状
3-5岁： 色彩丰富、富有表现力、角色生动有动感
6-8岁： 场景更详细，包含视觉叙事
9-12岁： 插画更精致，辅助文字内容
三色原则： 每个角色限制使用3-4种主色调，保证视觉清晰

Game asset tips (from game-asset-generation)

游戏资产技巧（来自游戏资产生成指南）

Always specify pixel dimensions for sprites: "32x32", "64x64", "128x128"
For seamless textures: "must tile perfectly with no visible seams when repeated"
For sprite sheets: specify grid layout "4x2 grid (256x64 total)"
For icons: "clear silhouette readable at 32x32 pixels"

务必为精灵指定像素尺寸："32x32"、"64x64"、"128x128"
无缝纹理："must tile perfectly with no visible seams when repeated"
精灵图：指定网格布局"4x2 grid (256x64 total)"
图标："clear silhouette readable at 32x32 pixels"

Example: building a custom prompt

示例：构建自定义提示词

User: "Design a logo for my coffee shop called Bean Dream"

python

result = generate_image(
    prompt=(
        "flat vector logo of a coffee bean morphing into a crescent moon, "
        "minimalist design, warm brown and cream color palette, "
        "clean lines, white background, "
        "professional branding quality, works at any size"
    ),
    category="logo",
    model="gpt",  # GPT for better detail
)

用户："为我的咖啡店Bean Dream设计一个标志"

python

result = generate_image(
    prompt=(
        "flat vector logo of a coffee bean morphing into a crescent moon, "
        "minimalist design, warm brown and cream color palette, "
        "clean lines, white background, "
        "professional branding quality, works at any size"
    ),
    category="logo",
    model="gpt",  # GPT模型细节更出色
)

Example: game character concept

示例：游戏角色概念设计

User: "Create a warrior character for my RPG game"

python

result = generate_image(
    prompt=(
        "female warrior character, ornate golden armor with dragon motifs, "
        "flowing red cape, wielding a glowing enchanted sword, "
        "determined fierce expression, battle-ready stance, "
        "front view T-pose, clean white background"
    ),
    category="game_asset",
    style="character",
)

用户："为我的RPG游戏创建一个战士角色"

python

result = generate_image(
    prompt=(
        "female warrior character, ornate golden armor with dragon motifs, "
        "flowing red cape, wielding a glowing enchanted sword, "
        "determined fierce expression, battle-ready stance, "
        "front view T-pose, clean white background"
    ),
    category="game_asset",
    style="character",
)

Example: social media content

示例：社交媒体内容

User: "Make a TikTok cover about cooking tips"

python

result = generate_image(
    prompt=(
        "cooking tips video thumbnail, colorful kitchen scene, "
        "fresh ingredients flying in the air, chef's hands visible, "
        "fun energetic vibe, bold visual impact"
    ),
    category="social_media",
    style="tiktok_cover",
    # aspect_ratio auto-set to 9:16
)

用户："制作一个关于烹饪技巧的TikTok封面"

python

result = generate_image(
    prompt=(
        "cooking tips video thumbnail, colorful kitchen scene, "
        "fresh ingredients flying in the air, chef's hands visible, "
        "fun energetic vibe, bold visual impact"
    ),
    category="social_media",
    style="tiktok_cover",
    # 比例自动设置为9:16
)

9. Intent recognition guide

9. 意图识别指南

Use this table to map user requests to the correct category/style/parameters:

User says	Category	Style	Notes
"design a logo", "make a logo"	`logo`	auto-detect	Ask about industry for style
"create a poster", "event poster"	`poster`	auto-detect
"draw an illustration", "illustrate"	`illustration`	auto-detect
"make a meme", "funny image"	`meme`	auto-detect
"game character", "RPG asset"	`game_asset`	`character`
"game environment", "level design"	`game_asset`	`environment`
"weapon design", "sword/shield"	`game_asset`	`weapon`
"game icon", "UI icon"	`game_asset`	`ui_icon`
"pixel art sprite"	`game_asset`	`pixel_sprite`
"tileable texture", "seamless tile"	`game_asset`	`tileset`
"Instagram post", "IG content"	`social_media`	`instagram`
"小红书", "xiaohongshu"	`social_media`	`xiaohongshu`
"TikTok cover", "抖音封面"	`social_media`	`tiktok_cover`	9:16
"YouTube thumbnail"	`social_media`	`youtube_thumbnail`	16:9
"banner", "cover image"	`social_media`	`banner`	16:9
"story template"	`social_media`	`story`	9:16
"3D character", "Pixar style"	`3d`	`character`
"product render", "3D product"	`3d`	`product`
"diorama", "miniature scene"	`3d`	`diorama`
"app icon", "3D icon"	`3d`	`icon`
"3D text", "chrome text"	`3d`	`text`
"textbook illustration"	`education`	`textbook`
"infographic", "data viz"	`education`	`infographic`
"scientific diagram"	`education`	`science`
"historical scene"	`education`	`history`
"fashion sketch", "clothing design"	`fashion`	`clothing`
"accessory design", "jewelry"	`fashion`	`accessory`
"nail art", "nail design"	`fashion`	`nail_art`
"textile pattern", "fabric design"	`fashion`	`textile`
"food photo", "dish"	`food`	`dish`
"menu design"	`food`	`menu`
"food packaging"	`food`	`packaging`
"recipe card"	`food`	`recipe_card`
"product photo", "product shot"	`product`	`hero`
"Amazon listing", "e-commerce photo"	`product`	`packshot`	White bg
"lifestyle product", "product in context"	`product`	`lifestyle`
"flat lay", "top-down product"	`product`	`flat_lay`
"pet in clothes", "humanized pet"	`pet`	`humanized`
"royal pet", "renaissance pet"	`pet`	`renaissance`
"cartoon pet", "Disney pet"	`pet`	`cartoon`
"pet merchandise", "pet pattern"	`pet`	`merchandise`
"pet memorial", "rainbow bridge"	`pet`	`memorial`
"wedding invitation", "请柬"	`wedding`	`invitation`
"modern invitation"	`wedding`	`invitation_modern`
"rustic invitation"	`wedding`	`invitation_rustic`
"wedding venue", "婚礼布置"	`wedding`	`venue`
"save the date"	`wedding`	`save_the_date`
"Christmas card", "圣诞贺卡"	`holiday`	`christmas_card`
"春节", "Chinese New Year"	`holiday`	`chinese_new_year`
"New Year design", "新年"	`holiday`	`new_year`
"Valentine's card", "情人节"	`holiday`	`valentines`
"Halloween", "万圣节"	`holiday`	`halloween`
"sale banner", "promotional"	`holiday`	`promotional`
"中秋节", "Mid-Autumn"	`holiday`	`mid_autumn`
"Ghibli style", "吉卜力"	`art_style`	`ghibli`
"comic style", "漫画风"	`art_style`	`american_comic` or `manga`
"pixel art", "像素风"	`art_style`	`pixel_art`
"pencil sketch", "素描"	`art_style`	`pencil_sketch`
"steampunk", "蒸汽朋克"	`art_style`	`steampunk`
"fantasy", "魔法"	`art_style`	`fantasy_magic`
"wuxia", "武侠", "仙侠"	`art_style`	`wuxia`
"space", "太空", "sci-fi"	`art_style`	`scifi_space`
"pop art", "波普"	`art_style`	`pop_art`
"ukiyo-e", "浮世绘"	`art_style`	`ukiyo_e`
"impressionist", "印象派"	`art_style`	`impressionist`
"Art Nouveau", "新艺术"	`art_style`	`art_nouveau`
"highest quality", "best quality"	any	any + `model="gpt"`
Custom description not in presets	—	—	Use `prompt=` directly

使用下表将用户请求映射到正确的分类/风格/参数：

用户表述	分类	风格	说明
"design a logo"、"make a logo"	`logo`	自动检测	询问行业以确定风格
"create a poster"、"event poster"	`poster`	自动检测
"draw an illustration"、"illustrate"	`illustration`	自动检测
"make a meme"、"funny image"	`meme`	自动检测
"game character"、"RPG asset"	`game_asset`	`character`
"game environment"、"level design"	`game_asset`	`environment`
"weapon design"、"sword/shield"	`game_asset`	`weapon`
"game icon"、"UI icon"	`game_asset`	`ui_icon`
"pixel art sprite"	`game_asset`	`pixel_sprite`
"tileable texture"、"seamless tile"	`game_asset`	`tileset`
"Instagram post"、"IG content"	`social_media`	`instagram`
"小红书"、"xiaohongshu"	`social_media`	`xiaohongshu`
"TikTok cover"、"抖音封面"	`social_media`	`tiktok_cover`	9:16
"YouTube thumbnail"	`social_media`	`youtube_thumbnail`	16:9
"banner"、"cover image"	`social_media`	`banner`	16:9
"story template"	`social_media`	`story`	9:16
"3D character"、"Pixar style"	`3d`	`character`
"product render"、"3D product"	`3d`	`product`
"diorama"、"miniature scene"	`3d`	`diorama`
"app icon"、"3D icon"	`3d`	`icon`
"3D text"、"chrome text"	`3d`	`text`
"textbook illustration"	`education`	`textbook`
"infographic"、"data viz"	`education`	`infographic`
"scientific diagram"	`education`	`science`
"historical scene"	`education`	`history`
"fashion sketch"、"clothing design"	`fashion`	`clothing`
"accessory design"、"jewelry"	`fashion`	`accessory`
"nail art"、"nail design"	`fashion`	`nail_art`
"textile pattern"、"fabric design"	`fashion`	`textile`
"food photo"、"dish"	`food`	`dish`
"menu design"	`food`	`menu`
"food packaging"	`food`	`packaging`
"recipe card"	`food`	`recipe_card`
"product photo"、"product shot"	`product`	`hero`
"Amazon listing"、"e-commerce photo"	`product`	`packshot`	纯白背景
"lifestyle product"、"product in context"	`product`	`lifestyle`
"flat lay"、"top-down product"	`product`	`flat_lay`
"pet in clothes"、"humanized pet"	`pet`	`humanized`
"royal pet"、"renaissance pet"	`pet`	`renaissance`
"cartoon pet"、"Disney pet"	`pet`	`cartoon`
"pet merchandise"、"pet pattern"	`pet`	`merchandise`
"pet memorial"、"rainbow bridge"	`pet`	`memorial`
"wedding invitation"、"请柬"	`wedding`	`invitation`
"modern invitation"	`wedding`	`invitation_modern`
"rustic invitation"	`wedding`	`invitation_rustic`
"wedding venue"、"婚礼布置"	`wedding`	`venue`
"save the date"	`wedding`	`save_the_date`
"Christmas card"、"圣诞贺卡"	`holiday`	`christmas_card`
"春节"、"Chinese New Year"	`holiday`	`chinese_new_year`
"New Year design"、"新年"	`holiday`	`new_year`
"Valentine's card"、"情人节"	`holiday`	`valentines`
"Halloween"、"万圣节"	`holiday`	`halloween`
"sale banner"、"promotional"	`holiday`	`promotional`
"中秋节"、"Mid-Autumn"	`holiday`	`mid_autumn`
"Ghibli style"、"吉卜力"	`art_style`	`ghibli`
"comic style"、"漫画风"	`art_style`	`american_comic` 或 `manga`
"pixel art"、"像素风"	`art_style`	`pixel_art`
"pencil sketch"、"素描"	`art_style`	`pencil_sketch`
"steampunk"、"蒸汽朋克"	`art_style`	`steampunk`
"fantasy"、"魔法"	`art_style`	`fantasy_magic`
"wuxia"、"武侠"、"仙侠"	`art_style`	`wuxia`
"space"、"太空"、"sci-fi"	`art_style`	`scifi_space`
"pop art"、"波普"	`art_style`	`pop_art`
"ukiyo-e"、"浮世绘"	`art_style`	`ukiyo_e`
"impressionist"、"印象派"	`art_style`	`impressionist`
"Art Nouveau"、"新艺术"	`art_style`	`art_nouveau`
"highest quality"、"best quality"	任意	任意 + `model="gpt"`
不在预设中的自定义描述	—	—	直接使用 `prompt=`

10. Using with a reference image (optional)

10. 使用参考图像（可选）

While this skill is primarily text-to-image, you can provide a reference image for design inspiration:

python

undefined

虽然此技能主要用于text-to-image，但你也可以提供参考图像作为设计灵感：

python

undefined

Reference image for design guidance

参考图像用于设计指导

result = generate_image( prompt="redesign this logo in a modern minimalist style", category="logo", image_path="uploads/old_logo.png", )

Reference URL

参考URL

result = generate_image( prompt="create a similar style illustration but with a forest theme", image_url="https://example.com/reference.jpg", )


When a reference image is provided, the script uses the `/edit` endpoint instead of the generate endpoint.

---

result = generate_image( prompt="create a similar style illustration but with a forest theme", image_url="https://example.com/reference.jpg", )


当提供参考图像时，脚本会使用`/edit`端点而非生成端点。

---

11. Multiple images

11. 生成多张图像

python

undefined

python

undefined

Generate 4 logo variants

生成4个标志变体

result = generate_image( prompt="minimalist mountain logo for outdoor brand", category="logo", count=4, )

result["images"] -> list of 4 image dicts

result["images"] -> 包含4个图像字典的列表

---

---

12. Anti-patterns (avoid these)

12. 反模式（需避免）

Avoid	Why	Instead
`"logo with text 'My Brand'"`	AI garbles text	Generate icon only, add text in Figma/Canva
`"photorealistic logo"`	Logos aren't photos	Use "flat vector logo"
`"3D rendered logo"`	Won't scale to favicon	Use "flat minimal icon"
Vague prompts like "nice image"	Poor results	Be specific: subject, style, colors, lighting
Too many concepts in one prompt	Confused output	Focus on one clear concept
Requesting exact pixel dimensions	Not supported	Use `aspect_ratio` parameter
Using `nanopro` for text-heavy designs	Text rendering poor	Use `model="gpt"` for text

需避免的操作	原因	替代方案
`"logo with text 'My Brand'"`	AI会使文字错乱	仅生成图标，在Figma/Canva中添加文字
`"photorealistic logo"`	标志并非照片	使用"flat vector logo"
`"3D rendered logo"`	无法缩小为网站图标	使用"flat minimal icon"
模糊提示词如"nice image"	结果质量差	明确说明：主体、风格、颜色、光线
一个提示词包含过多概念	输出混乱	聚焦一个清晰的概念
请求精确像素尺寸	不支持该功能	使用 `aspect_ratio` 参数
对文字密集型设计使用 `nanopro`	文字渲染效果差	对文字内容使用 `model="gpt"`

13. Provided scripts

13. 提供的脚本

File	Purpose
`generate_image.py`	Core script: prompt building → submit → poll → download. Handles all categories, styles, two models.
`exports.py`	Re-exports `generate_image` , `CATEGORY_STYLES` , `MODELS` for programmatic use.
`_cost_track.py`	Cost tracking helper — records per-call costs via sc-proxy headers. Self-contained, no external dependencies.

文件	用途
`generate_image.py`	核心脚本：提示词构建 → 提交 → 轮询 → 下载。支持所有分类、风格及两种模型。
`exports.py`	重新导出 `generate_image` 、 `CATEGORY_STYLES` 、 `MODELS` ，供程序化调用。
`_cost_track.py`	成本跟踪辅助工具——通过sc-proxy头部记录每次调用的成本。独立封装，无外部依赖。

14. Local testing

14. 本地测试

Set

FAL_KEY

env var to call fal.ai directly (bypasses sc-proxy):

bash

undefined

设置

FAL_KEY

环境变量以直接调用fal.ai（绕过sc-proxy）：

bash

undefined

Basic generation

基础生成

FAL_KEY=your-fal-key python3 skills/image-create/generate_image.py "a cute robot" illustration fantasy 1 nanopro

Args: <prompt> [category] [style] [count] [model]

参数：<prompt> [category] [style] [count] [model]

---

---

15. Troubleshooting

15. 故障排除

Problem	Fix
`Either 'prompt' or 'category' must be provided`	Provide at least a prompt or category
`File not found: ...`	Check the workspace path for reference image
`Unsupported image format`	Use `.jpg` , `.jpeg` , `.png` , `.webp` , or `.bmp`
`Image too large`	Resize reference image to under 10 MB
`HTTP 402 insufficient_credits`	Top up balance; cost is pre-charged on submit
`HTTP 403 endpoint_not_allowed`	sc-proxy only allows approved fal endpoints; contact admin
Generation `FAILED` upstream	Simplify prompt, retry
Job stuck `IN_PROGRESS` >10 min	Save `request_id` , retry later
Text not rendering well	Switch to `model="gpt"` — GPT handles text much better
`gpt` model too slow	Switch to `nanopro` (default) for faster results
Logo too complex to scale	Use "flat vector", "minimal", "single color" in prompt
Seamless texture has visible seams	Add "must tile perfectly with no visible seams" to prompt

问题	解决方法
`Either 'prompt' or 'category' must be provided`	至少提供一个提示词或分类
`File not found: ...`	检查参考图像的工作区路径
`Unsupported image format`	使用 `.jpg` 、 `.jpeg` 、 `.png` 、 `.webp` 或 `.bmp` 格式
`Image too large`	将参考图像调整至10MB以下
`HTTP 402 insufficient_credits`	充值余额；提交时会预先扣费
`HTTP 403 endpoint_not_allowed`	sc-proxy仅允许已批准的fal端点；联系管理员
上游生成 `FAILED`	简化提示词，重试
任务卡在 `IN_PROGRESS` 超过10分钟	保存 `request_id` ，稍后重试
文字渲染效果差	切换至 `model="gpt"` ——GPT处理文字的效果好得多
`gpt` 模型速度太慢	切换至 `nanopro` （默认）以获得更快结果
标志过于复杂无法缩放	在提示词中使用"flat vector"、"minimal"、"single color"
无缝纹理有可见接缝	在提示词中添加"must tile perfectly with no visible seams"

16. Infrastructure (reference)

16. 基础设施（参考）

Caller →
```
sc-proxy
```
→
```
queue.fal.run/{model}
```
→ fal model providers
All requests must include
```
Authorization: Key fake-falai-key-12345
```
(proxy injects the real
```
FAL_KEY
```
)
Pre-charge happens at submit. Poll/result calls are free.
Final images live at
```
https://*.fal.media/...
```
— public CDN, no auth needed for download.
Cost tracking via
```
_cost_track.py
```
— records
```
X-Credits-Used
```
from sc-proxy response headers.
Each skill contains its own
```
_cost_track.py
```
copy (skills are independently deployed).

调用者 →
```
sc-proxy
```
→
```
queue.fal.run/{model}
```
→ fal模型提供商
所有请求必须包含
```
Authorization: Key fake-falai-key-12345
```
（代理会注入真实的
```
FAL_KEY
```
）
提交时预先扣费，轮询/结果查询免费。
最终图像存储在
```
https://*.fal.media/...
```
——公共CDN，下载无需认证。
通过
```
_cost_track.py
```
跟踪成本——记录sc-proxy响应头部中的
```
X-Credits-Used
```
。
每个技能包含独立的
```
_cost_track.py
```
副本（技能独立部署）。

Model endpoints

模型端点

Model	Generate (text only)	Edit (with ref image)
nanopro	`fal-ai/nano-banana-pro`	`fal-ai/nano-banana-pro/edit`
gpt	`openai/gpt-image-2`	`openai/gpt-image-2/edit`

模型	生成（仅文本）	编辑（带参考图像）
nanopro	`fal-ai/nano-banana-pro`	`fal-ai/nano-banana-pro/edit`
gpt	`openai/gpt-image-2`	`openai/gpt-image-2/edit`

image-create

Original

Translation

image-create

image-create

1. Quick start — basic generation (most common)

1. 快速入门——基础生成（最常用）

result -> {"success": True, "images": [{"local_path": "output/images/..."}], ...}

result -> {"success": True, "images": [{"local_path": "output/images/..."}], ...}

2. Quick start — with category preset

2. 快速入门——使用分类预设

3. Quick start — category only (no custom prompt)

3. 快速入门——仅使用分类（无自定义提示词）

Uses the built-in style template as the full prompt

Uses the built-in style template as the full prompt

Delivering the result to the user — IMPORTANT

向用户交付结果——重要提示

4. Parameters

4. 参数

5. Category and style presets

5. 分类与风格预设

E: Design — Logo (category="logo")

E: 设计——标志（category="logo"）

E: Design — Poster (category="poster")

E: 设计——海报（category="poster"）

E: Design — Illustration (category="illustration")

E: 设计——插画（category="illustration"）

E: Design — Meme (category="meme")

E: 设计——表情包（category="meme"）

K: Game Assets (category="game_asset")

K: 游戏资产（category="game_asset"）

M: Social Media (category="social_media")

M: 社交媒体（category="social_media"）

N: 3D (category="3d")

N: 3D（category="3d"）

P: Education (category="education")

P: 教育（category="education"）

Q: Fashion (category="fashion")

Q: 时尚（category="fashion"）

R: Food (category="food")

R: 美食（category="food"）

S: Pet (category="pet")

S: 宠物（category="pet"）

I: Product Photography (category="product")

I: 产品摄影（category="product"）

Y: Wedding (category="wedding")

Y: 婚礼（category="wedding"）

Z: Holiday Marketing (category="holiday")

Z: 节日营销（category="holiday"）

C: Art Style (category="art_style")

C: 艺术风格（category="art_style"）

6. Model selection guide

6. 模型选择指南

Default (fast)

默认设置（快速）

High quality (user requested)

高质量（用户要求）

7. Aspect ratio guide

7. 比例指南

8. Prompt engineering best practices

8. 提示词工程最佳实践

The 5-element prompt structure

五要素提示词结构

Key principles

核心原则

Logo-specific prompt tips (from logo-design-guide)

标志专属提示词技巧（来自标志设计指南）

Children's illustration tips (from book-illustrator)

儿童插画技巧（来自书籍插画师指南）

Game asset tips (from game-asset-generation)

游戏资产技巧（来自游戏资产生成指南）

Example: building a custom prompt

示例：构建自定义提示词

Example: game character concept

示例：游戏角色概念设计

Example: social media content

示例：社交媒体内容

9. Intent recognition guide

9. 意图识别指南

10. Using with a reference image (optional)

E: Design — Logo (
`category="logo"`
)

E: 设计——标志（
`category="logo"`
）

E: Design — Poster (
`category="poster"`
)

E: 设计——海报（
`category="poster"`
）

E: Design — Illustration (
`category="illustration"`
)

E: 设计——插画（
`category="illustration"`
）

E: Design — Meme (
`category="meme"`
)

E: 设计——表情包（
`category="meme"`
）

K: Game Assets (
`category="game_asset"`
)

K: 游戏资产（
`category="game_asset"`
）

M: Social Media (
`category="social_media"`
)

M: 社交媒体（
`category="social_media"`
）

N: 3D (
`category="3d"`
)

N: 3D（
`category="3d"`
）

P: Education (
`category="education"`
)

P: 教育（
`category="education"`
）

Q: Fashion (
`category="fashion"`
)

Q: 时尚（
`category="fashion"`
）

R: Food (
`category="food"`
)

R: 美食（
`category="food"`
）

S: Pet (
`category="pet"`
)

S: 宠物（
`category="pet"`
）

I: Product Photography (
`category="product"`
)

I: 产品摄影（
`category="product"`
）

Y: Wedding (
`category="wedding"`
)

Y: 婚礼（
`category="wedding"`
）

Z: Holiday Marketing (
`category="holiday"`
)

Z: 节日营销（
`category="holiday"`
）

C: Art Style (
`category="art_style"`
)

C: 艺术风格（
`category="art_style"`
）