Files

T

Antigravity Agent 9148c358d0 docs(10_Wiki): 위키 전체 재구성 — Topic_* 폴더를 4개 카테고리로 통합 + 대규모 중복 제거

Topic_Agent/Topic_Blog/Topics/Topics_Biz/Topics_Meeting/Topics_Rag의 마크다운 지식 문서를
Topic_General/Topic_Programming/Topic_Graphic/Topic_Business 4개 카테고리로 재분류.

- 중복 제거: frontmatter의 status:duplicate/merged + duplicate_of/redirect_to 필드로
  자기 자신을 중복으로 선언한 리다이렉트 stub 1032개 제거, 완전 동일 내용 파일 472개 제거,
  동일 파일명·다른 내용 충돌 시 더 큰(완전한) 버전만 유지(162개 제거) — 총 1639개 중복 제거.
- 분류: 폴더 단위로 명확한 항목(AI_and_ML/Coding/Architecture 등 → Programming,
  Comfyui/Visual_Effects → Graphic, Topics_Biz/Topics_Meeting/사업 등 → Business,
  Poetic_Blog_Writing/창의성/Game_Design 등 → General)은 폴더 우선순위로,
  나머지 혼재 폴더(Topic_Agent/Topic_Blog/Topics 루트/Thinking & Reasoning/Other/UI_UX_Assets)는
  title/tags 키워드 스코어링으로 파일 단위 분류(불명확한 경우 General로 폴백).
  원본 폴더명은 "From_*" 서브폴더로 보존해 추적 가능성 유지.
- 최종 배치: Programming 2784 / General 1608 / Graphic 285 / Business 249 = 4926개 문서.
- 에이전트 운영 상태(.astra/.agent/.obsidian/sessions/memory/_company/docs/lessons/_shared/src)는
  지식 콘텐츠가 아니므로 재분류 대상에서 제외하고 원위치 유지.
- Topics/Topic_email(상위 보호 폴더 Topic_email과 파일명 100% 중복) 삭제 — 보호 폴더 자체는 미변경.
- 완전히 비게 된 Topic_Agent/Topic_Blog/Topics_Biz/Topics_Rag 폴더 제거.

2026-07-05 00:33:48 +09:00

5.1 KiB

Raw Blame History

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack

title

Lighting & Composition

매 한 줄

"매 image / video 의 emotional weight 의 90%는 lighting + composition". 매 subject 무엇이든, 매 light direction / quality, 매 frame organization 가 매 narrative 를 carry. 매 2026 의 generative AI (FLUX 1.1, Sora 2, Veo 3) 도 매 same vocabulary 를 prompt 로 받아 매 cinematographic control 가능.

매 핵심

매 lighting 의 axes

Direction: front / side / back / top / bottom (각 emotional valence 다름).
Quality: hard (sharp shadows) vs soft (diffused).
Color temperature: warm (3000K) vs cool (6500K).
Ratio: key:fill ratio (1:1 flat, 4:1 dramatic).

매 composition 의 axes

Rule of thirds: 매 subject 매 third lines 의 intersection.
Leading lines: 매 viewer's eye 의 directing.
Negative space: 매 emptiness 가 carry meaning.
Depth layers: foreground / mid / background.
Aspect ratio: 16:9 (cinematic), 9:16 (mobile), 1:1 (square), 2.39:1 (anamorphic).

매 응용

Photography: portrait / product / landscape lighting setups.
Cinematography: 매 scene mood 의 establishing.
Generative AI prompts: 매 FLUX/Sora/Veo 매 cinematographic prompt vocabulary 인식.
UI design: 매 hero image direction selection.

💻 패턴

Three-point lighting prompt

A portrait of a software engineer, three-point lighting: 
key light from camera-left at 45 degrees (soft, 5500K), 
fill light from camera-right at 1:2 ratio, 
backlight rim from upper-right (warm 3200K), 
shallow depth of field, 85mm lens equivalent, 
shot on Arri Alexa 35.

Rembrandt lighting (FLUX 1.1 prompt)

flux_prompt = """
Rembrandt lighting on subject's face: 
small triangle of light on shadow-side cheek, 
key light camera-left high at 45°, 
deep shadow on right side, 
chiaroscuro mood, oil painting aesthetic, 
photorealistic, 8K detail.
"""

Golden hour video (Sora 2)

sora_prompt = {
    "shot": "tracking shot through wheat field",
    "lighting": "golden hour, sun low at 15° angle camera-back-left, "
                "warm 2800K, long shadows, lens flare",
    "composition": "rule of thirds, horizon on lower third, "
                   "leading lines from wheat rows toward subject",
    "camera": "Steadicam, 24fps, 35mm anamorphic, T2.0",
    "duration_s": 8,
}

Composition checker (CV-based QA)

import cv2
import numpy as np

def rule_of_thirds_score(img: np.ndarray) -> float:
    """매 saliency map peak 매 third-line proximity."""
    h, w = img.shape[:2]
    saliency = cv2.saliency.StaticSaliencyFineGrained_create()
    _, sal = saliency.computeSaliency(img)
    peak_y, peak_x = np.unravel_index(sal.argmax(), sal.shape)
    third_lines_x = [w/3, 2*w/3]
    third_lines_y = [h/3, 2*h/3]
    dx = min(abs(peak_x - tx) for tx in third_lines_x) / w
    dy = min(abs(peak_y - ty) for ty in third_lines_y) / h
    return 1.0 - min(dx, dy) * 2  # higher = better composition

Color temp grading (Veo 3 prompt augmentation)

def grade_prompt(base: str, mood: str) -> str:
    grades = {
        "warm_nostalgic": "teal-orange grade, warm midtones (3200K), cool shadows",
        "cold_clinical": "desaturated blues, 6500K key, high-key flat lighting",
        "noir": "high-contrast B&W, low-key, single hard source, 4:1 ratio",
    }
    return f"{base} | grade: {grades[mood]}"

매 결정 기준

Mood	Lighting	Composition
Heroic	Backlight rim + low-key fill	Low angle, centered
Intimate	Soft key, high ratio	Close-up, off-center
Tense	Hard side light, deep shadow	Dutch tilt, asymmetric
Whimsical	Bright fill, warm tones	Wide, symmetrical
Documentary	Available light	Eye-level, rule of thirds

기본값: 매 three-point + rule of thirds — 매 safe baseline.

🔗 Graph

부모: Cinematography

🤖 LLM 활용

언제: 매 image/video prompt engineering — 매 cinematographic vocabulary 매 quality lift 큼. 언제 X: 매 abstract / non-representational generation — vocabulary 의 X.

❌ 안티패턴

Flat front lighting: 매 amateur look — depth loss.
Centered everything: 매 visual boring.
Mixed color temps unintentional: 매 amateur giveaway.
Over-prompting: 매 50+ tokens 의 lighting → 매 model confusion.

🧪 검증 / 중복

Verified (Brown Cinematography; Block Visual Story; FLUX 1.1 prompt guide; Sora 2 system card 2025).
신뢰도 A.

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — lighting/composition vocabulary + 2026 generative AI prompts

5.1 KiB Raw Blame History