CaiguangAI Fashion Studio
裁光 · 服饰 AI 影棚CAIGUANG · AI FASHION STUDIO
从单件服饰或整套搭配照片出发，
选择风格，生成 AI 模特展示图。Start with a garment or outfit photo.
Choose a style. Create AI model images.
单件 / 套装识别 · 24 种风格可选
单张测试 / 六张成片Garment & outfit recognition · 24 styles
Single test / Six-image set
在 Codex 中使用的服饰 AI 创作插件A fashion image creation plugin for Codex
参考素材REFERENCE
先识别服饰、推荐风格，再确认方案；生成后对照源图检查，按需授权修正。Identify the garment, explore styles and confirm a plan. Review generated images against the source, then approve any retries.
从 24 种风格中选择主风格，确认棚拍、场景或混合模式，以及构图与画幅。Choose a main style from 24 options, then set studio, location or mixed scenes, composition and aspect ratio.
默认 AI 模特展示；不露脸局部、平铺、挂拍、人台可明确请求，已有流程定义，逐形式完整公开验证待补。AI model portraits are the default. Faceless detail, flat-lay, hanger and mannequin workflows are defined; complete public validation for each remains pending.
公开案例以 AI 模特图为主，不代表所有功能组合均已验证。Public cases focus on AI model images; they do not validate every feature combination.
STYLING CASES · 6 × 6
6 组 AI 六图案例。皮衣同款对照：外部试用与通勤窗光，服饰细节保留复核。Six six-image AI examples. Compare an external trial and Office Commute daylight from the same jacket input; garment-detail review remains separate.
BEFORE商品参考Product references
AFTERAI 场景穿搭AI scene styling
展示本次实际使用的两张主要服装参考；另有结构与人物参考。裤装照片中的灰色上衣不参与搭配。AI效果用于穿搭展示，细节与实物合身仍需核对。Two primary garment references actually used; additional construction and identity references were supplied. The gray top in the pants photo is not part of the generated outfit. AI styling previews still require detail and physical-fit review.
同一输入 · 外部试用与通勤窗光One input · external trial and Office Commute
AI 生成 · 六张独立图AI-generated · six separate images
AI 生成 · 六张独立图AI-generated · six separate images
AI 生成 · 六张独立图AI-generated · six separate images
AI 生成 · 六张独立图AI-generated · six separate images
01
同一件白马甲，六张独立成片。The same white vest, six individual portraits.
01侧身回转站姿Side-turn standing
02侧身倚靠墙面Side-leaning against a wall
03端正半身坐姿Upright half-body seated
04正面轻步迈步Front-facing gentle step
05微前倾俯身Slight forward lean
06背身回眸Back-turn glance
同一件服饰
六张独立图片One garment
Six individual images
AI 生成 · 已人工验收的历史案例。每张均可放大查看；不代表当前版本的全风格验证。AI-generated · a historical set accepted after human review. Open any image to inspect it; this is not an all-style validation of the current version. 案例记录与媒体条款 ↗Case record & media terms ↗
看见细节A CLOSER LOOK
对照源图检查颜色与结构；
补充搭配在生成前确认。Check color and construction against the source.
Confirm complementary styling before generation.
局部对照用于检查，不代表所有细节完全一致。These close-ups support review; they do not certify every detail.
02
以同一套服饰为参考，展示 24 种可选风格中的 3 种生成效果。The same outfit, shown in 3 of the 24 available styles.
既有公开 AI 样例，每种风格展示一张；主要呈现场景与光线变化，不是同套服饰的六图交付案例。Published AI examples, one per style, showing changes mainly in setting and light; not a complete six-image outfit set. 源图条款 ↗Source terms ↗ · 结果来源与权利 ↗Result provenance & rights ↗
03
从棚拍到街头，从日常到编辑大片。全部在这里。From studio to street, everyday to editorial. All here.
不同风格使用各自的参考素材。24 种风格供选择，每次选择一个主风格；并非一次生成 24 组。Styles use different source materials. Choose one main style from 24 options; one request does not generate 24 sets.
街头环境，自然随性的姿态。Street settings and an easy, unposed attitude.
运动搭配，轻松有活力。Sport-inspired styling with relaxed energy.
柔和光线，轻盈的身体线条。Soft light and graceful, fluid lines.
经典外套，沉稳的复古氛围。Classic outerwear with a heritage mood.
宽松廓形，城市日常感。Relaxed silhouettes in everyday city scenes.
简洁搭配，清楚呈现廓形。Simple styling that gives the silhouette space.
柔美色彩，细腻的装饰感。Soft colors and delicate decorative details.
干净背景，让服饰更清楚。A clean backdrop that brings the garment forward.
自然光线，松弛的生活场景。Natural light and an effortless everyday setting.
户外环境，突出机能层次。Outdoor settings and practical, layered styling.
街巷场景，鲜明的都市气氛。Street scenes with a distinctive urban mood.
温暖光影，丰富的场景质感。Warm light and richly textured settings.
日常空间，柔和自然光。Everyday spaces and soft natural light.
冷调光线，安静的杂志构图。Cool light and quiet editorial framing.
利落线条，简洁的城市穿搭。Clean lines and understated urban styling.
克制留白，现代东方意境。Considered space and a modern Chinese aesthetic.
明净空间，简约的轮廓关系。Airy spaces and pared-back silhouettes.
通勤场景，清晰的搭配层次。Workday settings with considered outfit layers.
经典搭配，温润的室内光影。Classic styling in warm, atmospheric interiors.
学院元素，整洁有层次。Collegiate references and neatly layered looks.
低调色彩，重视光影与材质。Understated colors with attention to light and texture.
开阔场景，轻松的度假气息。Open settings and an easy holiday mood.
工装层次，质朴的复古表达。Utility layers and a grounded vintage mood.
鲜明搭配，带有千禧年代气息。Expressive styling with a turn-of-the-millennium mood.
样例不等于当前版本全风格验证，也不代表图片可自由复用。Examples do not verify every style in the current version or grant unrestricted media reuse.
美式街头、英伦传承、都市宽松男装、韩式男装、美式学院和复古工装使用 Eyüpcan Timur 的摄影作品作人物参考；其余 18 组使用既有公开 AI 女性案例。服饰参考包括署名 Pexels 摄影与项目已授权的米色套装。它们是已公开的历史、补测和替换输入结果，不是 beta.8 新生成结果。American Street, British Heritage, Cityboy, Korean Menswear, Preppy and Workwear Vintage use an Eyüpcan Timur photograph as an identity reference; the other 18 use the existing public AI female case. Garment references include credited Pexels photographs and the project-authorized beige outfit. These are published historical, supplementary and replaced-input results, not new beta.8 outputs.
人物摄影参考：Eyüpcan Timur ↗Identity photograph: Eyüpcan Timur ↗ · AI 人物案例 ↗AI identity case ↗ · 逐组来源与权利 ↗Per-case provenance & rights ↗
04
先识别服饰与选择方向，再决定是否生图。Identify the garment and choose a direction before generating.
尚未安装NEW TO CAIGUANG
插件安装在 Codex 中。按指南完成下载、校验、注册与启用，再开始第一次识别。Install the plugin in Codex. Follow the guide to download, verify, register and enable it, then run your first recognition.
查看安装指南 ↗Open installation guide ↗ 下载裁光插件 ↗Download Caiguang plugin ↗beta.11 在 Codex 中显示为裁光 · Caiguang，调用名仍为 $threadtruth-studio；新环境实测待完成。In Codex, beta.11 is listed as 裁光 · Caiguang and invoked with $threadtruth-studio. Fresh-host testing remains pending.
当前验证环境：维护者 macOS，Python 3，以及支持插件安装的 Codex CLI。新机器与其他系统尚未验证。Verified on the maintainer’s Mac with Python 3 and a Codex CLI supporting plugin installation. Fresh machines and other systems remain unverified.
已经安装ALREADY INSTALLED
在新的 Codex 任务中上传清晰、有权使用的服饰照片，再复制下方指令。Upload a clear, authorized garment photo in a new Codex task, then copy the prompt below.
一张清晰正面照即可开始识别；
其他角度按需要补充。Start recognition with one clear front view.
Add other views as needed.
请用 $threadtruth-studio 识别并推荐风格，不要生图
Use $threadtruth-studio to identify this garment and recommend styles. Do not generate images.
也可选中指令手动复制。这一步不生成图片。You can also select and copy the prompt manually. This step does not generate images.
单件上装需要补全的裤装、鞋包会在方案中确认，不当作源图事实。完整套装应清楚展示各单品及搭配关系。For a single top, complementary bottoms, shoes and accessories are confirmed in the plan, not treated as source facts. An outfit photo should clearly show each item and its layering.
ChatGPT 网页是从 Codex 发起的可选转交路线；插件不会安装到 ChatGPT。生成与重试需要明确授权，失败不自动继续。ChatGPT web is an optional handoff route from Codex; the plugin is not installed in ChatGPT. Generation and retries require explicit approval; failures do not automatically continue.