AI 도구 비교 카드 (ChatGPT·Claude·Gemini)

提示詞原文

[Input]
- language: Korean
- output_size: 1024x1536

[Purpose]
Korean AI tool comparison card for social media carousel, blog thumbnail, or newsletter feature image, presented as a polished vertical editorial infographic.
3-column comparison layout for ChatGPT, Claude, and Gemini with quick at-a-glance evaluation, practical notes, update metadata, and shareable card-news details.

[Composition]
- Vertical poster composition: three equal columns across the page, each column occupying about 30% width with 5% gutters, framed by a refined thin outer rule and realistic printed-card edge treatment.
- Top headline band takes 14% of height, main comparison grid takes 72%, bottom recommendation strip takes 14%.
- Use {비교 도구}, {비교 항목}, {강점·약점}, {추천}, {비주얼}, {컬러}, {비율} as editable text content.
- Background should be crisp white with subtle paper grain, faint off-white editorial texture, charcoal dividers, and restrained accent colors per tool.
- Add realistic social-card metadata: top-left small label “AI TOOL GUIDE”, top-right “VOL.03 · 2026.05 업데이트”, a tiny handle “@ai.notes.kr”, and a slide marker “01 / 05” near the lower corner.
- Include small save/share/bookmark micro icons and hashtag-like footer text such as “#AI도구 #업무자동화 #생산성팁” while keeping the 3-column comparison dominant.

[3. CORE STRUCTURE]
- Header area (top 14%): set the main title as a bold Korean masthead, centered, with compact line spacing and strong hierarchy. Main headline must read “AI 도구 비교”; add a smaller subtitle line: “대화·코딩·이미지·긴 문맥까지 한 장으로 정리”.
- Comparison grid (middle 72%): build exactly 3 vertical cards side by side; each card contains one tool name at the top, a small circular or rounded-square icon area beneath, then 3–5 short bullet lines for strengths, weaknesses, and quick use-case notes.
- Above the cards, place {비교 도구} and {비교 항목} as readable supporting labels, styled like small category tabs: “비교 도구: ChatGPT · Claude · Gemini” and “비교 항목: 대화 / 코딩 / 이미지 / 가격 / 언어”.
- Tool 1 card: place ChatGPT content in the left column with green accent bars, checklist dots, a compact score row, and small Korean tags such as “다재다능”, “일상 질문”, “빠른 초안”; keep the card balanced and evenly padded.
- Tool 2 card: place Claude content in the center column with beige accent bars, document-style marks, a compact score row, and small Korean tags such as “긴 문맥”, “코딩”, “문서 분석”; match the same text hierarchy as the other cards.
- Tool 3 card: place Gemini content in the right column with blue accent bars, subtle sparkle/geometric marks, a compact score row, and small Korean tags such as “멀티모달”, “이미지”, “구글 연동”; maintain equal visual weight.
- In each card, add a tiny “BEST FOR” ribbon and short Korean line: ChatGPT “일상 업무·아이디어”, Claude “코딩·긴 문서”, Gemini “이미지·검색 연동”.
- Add a small row of evaluation chips inside each card for “대화”, “코딩”, “이미지”, “가격”, “언어” using dots, short bars, or 5-step markers rather than long paragraphs.
- Recommendation strip (bottom 14%): show {추천} inside a high-contrast bar or rounded label row, with the exact conclusion sentence “코딩 = Claude / 일상 = ChatGPT / 멀티모달 = Gemini”.
- Below the recommendation sentence, add a subtle CTA line: “저장해두고 상황별로 골라 쓰기” plus a tiny source/date note “업데이트 기준: 2026.05 · 개인 사용 시나리오 기준”.
- Use {비주얼}, {컬러}, and {비율} as small production-note labels in the lower metadata area or side note, integrated naturally without creating extra panels.

[4. STYLE RULES]
- Typography: use Pretendard or Noto Sans KR for all Korean text; headline weight 800–900, body weight 500–600, small labels 400–500. Keep letter spacing tight and clean.
- Color palette: white #FFFFFF, charcoal #1F1F1F, ChatGPT green #10A37F, Claude beige #D8C3A5, Gemini blue #4285F4, light gray #EAECEF for separators.
- Border/shadow: thin 1px charcoal rules, soft drop shadow 4px y-offset, 12px blur, 12% opacity for the cards; no heavy neon glow.
- Add editorial details such as hairline dividers, tiny page-number text, small corner brackets, check icons, rating dots, and lightly textured paper fibers to make the card feel like a finished magazine-style infographic.
- Icons should be simple original geometric marks inspired by each tool category, not official certification marks; brand names “ChatGPT”, “Claude”, and “Gemini” must remain clearly readable.

[5. VISUAL STYLE]
- Camera / lens: clean editorial flat graphic layout, top-down poster design feel, simulated with crisp vector-like rendering rather than photographic depth.
- Lighting: even studio-like illumination with gentle contact shadows around the printed card edges, giving a realistic premium social-media graphic finish while keeping all text readable.
- Color grade: neutral white base with controlled accent saturation in green, beige, and blue; overall balanced and modern.
- Rendering quality: high-resolution digital editorial design, sharp typography, realistic paper texture, precise grid alignment, tactile card layering, and polished UI-thumbnail clarity.

[6. MOOD]
Practical editorial comparison sheet — like a polished Notion-style infographic mixed with a sharp product roundup from The Verge, upgraded into a saved-worthy Korean Instagram carousel card with useful real-world labels and decision cues.

[7. CRITICAL]
- Korean text accuracy: 반드시 정확히 보여야 하는 문자열 — “AI 도구 비교”, “ChatGPT”, “Claude”, “Gemini”, “대화”, “코딩”, “이미지”, “가격”, “언어”, “다재다능”, “긴 문맥”, “멀티모달”, “코딩 = Claude / 일상 = ChatGPT / 멀티모달 = Gemini”.
- Additional Korean microcopy must be clean and legible: “VOL.03 · 2026.05 업데이트”, “AI TOOL GUIDE”, “@ai.notes.kr”, “저장해두고 상황별로 골라 쓰기”, “#AI도구 #업무자동화 #생산성팁”.
- Layout invariant: exactly 3 columns, one tool per column, with equal visual weight and no extra fourth panel.
- All Korean text must be clean, correctly spelled, and fully legible.

[8. 🚫 PROHIBITED]
- Fake official seals / certifications
- Broken Korean characters
- Celebrity composites / unauthorized likeness