跳到正文
DEV Community· Noman Iftekhar·· 12 天前AI 評分30

AuraRead:我打造的 AI 個人風格分析應用

AuraRead

AI 導讀

作者打造 AuraRead,利用 Gemini 的多模態視覺能力分析照片中的儀容、穿搭、肢體語言及環境光線,生成個人風格分析與建議。應用透過伺服器端 Gemini API 使用 gemini-3.5-flash,並以結構化 JSON 格式輸出結果。用戶可上載肖像或使用網絡攝影機拍照,查看五項氣質評分、五色配色方案及造型和職涯建議。

正文

What I Built
I built AuraRead, an AI-powered visual persona, style, and charisma alignment app that decodes a person's aesthetic presence using Gemini's multimodal vision capabilities. With a quick photo upload or live webcam snapshot, AuraRead analyzes grooming, wardrobe structure, body language, and ambient lighting to generate an in-depth style archetype, aura attribute scores (elegance, confidence, warmth, professionalism, serenity), an authentic 5-tone color palette with hex codes, and actionable career and styling advice.
Key Prompts Used
Multimodal Visual & Aura Interpretation:
"Analyze the uploaded photo and interpret this person's vibe, style, charisma, styling selections, body language, facial posture, aura levels, and grooming choice. Generate five custom numeric attributes (elegance, confidence, warmth, professionalism, serenity) scaled from 0 to 100, produce a tailored color palette with hex codes, identify attire structures and body language cues, suggest ideal career paths, and return the data formatted strictly in structured JSON."
Persona Consultation & Lifestyle Advisory:
"Act as an elite luxury visual consultant and persona strategist. Using the analyzed style archetype, energy classification, and aesthetic summary, provide charming, tailored answers to user questions regarding fragrance pairings, watch selections, event attires, and posture refinements."
Key Features Utilized
Server-Side Gemini API (@google/genai): Uses gemini-3.5-flash with guaranteed structured JSON schema output (responseMimeType: "application/json").
Dual Visual Input: Upload any portrait image or snap a photo directly using the built-in browser webcam interface.
Interactive Vibe Meters & Simulation: Dynamic radar and progress bars with live sliders to simulate attribute tweaks.
Sartorial Color Palette: Extracts dominant tones from clothing and environment with color names, hex codes, and styling guidance.
AI Style Advisor Chat: Built-in consultation widget allowing users to ask bespoke styling questions about their active look.
Modern UI Stack: React 19, Tailwind CSS v4, Motion animations, and Lucide icons.
Demo
Experience AuraRead live in your browser:

Walkthrough
Default Persona Showcase: Explore the baseline Neo-Traditional Ambassador profile detailing wardrobe choices, posture analysis, and career synergies.
Scan or Upload: Click Capture Photo to snap a selfie or Upload Portrait to test any outfit.
Explore the Breakdown: View quantitative charisma scores, color swatches with hex values, and lifestyle recommendations.
Ask the Visual Advisor: Use the consultation widget at the bottom to ask questions like "What watch style complements this outfit?" or "What fragrance notes match this vibe?"
My Experience
Participating in the DEV Education Track: Build Apps with Google AI Studio was a great opportunity to explore the newest capabilities of Gemini and rapid application development.
Key Takeaways
Structured Outputs are a Game Changer: By enforcing a strict JSON schema with Type.OBJECT and Type.ARRAY, Gemini returned reliable, parseable data every time—eliminating broken JSON and markdown stripping.
Multimodal Speed with Gemini Flash: The turnaround time from uploading a high-resolution base64 photo to receiving a comprehensive, multi-attribute analysis was nearly instantaneous.
Secure Full-Stack Architecture: Running an Express backend with the official @google/genai TypeScript SDK kept API credentials securely on the server while feeding clean, typed data to the React front end.
What Was Surprising
What surprised me most was how perceptively Gemini analyzes subjective visual cues. Rather than just recognizing "a person in formal attire," it identified nuances in arm posture, facial expressions, fabric textures, and grooming symmetry, translating them into meaningful archetypes and color harmonies that feel genuinely personalized.

來源:DEV Community · dev.to