🌌 Tab 1 — Curtains Recreates the folded curtain structure of real auroras. Six glowing ribbons ripple like waves, each particle dancing with its own independent phase and velocity. Visualises how solar particles spiral down along Earth's magnetic field lines, exciting oxygen and nitrogen atoms to emit their characteristic light. 🌀 Tab 2 — Vortex The auroral corona as seen looking straight up at the magnetic pole. Particles spiral inward along converging field lines, and the Arms slider lets you adjust the spiral structure from 1 to 8 arms. Built on Float32Array typed arrays for smooth handling of 1,200 particles with minimal memory overhead. ✨ Tab 3 — Particle Storm Simulates a Kp-9 extreme geomagnetic storm — the most intense class of aurora event. Four modes: Cascade — solar particles raining down from above Explosion — particles bursting outward from a central point River — a flowing stream of particles channelled along magnetic field lines Neural — plasma connecting between nodes like a living neural network 🎵 Tab 4 — Resonance Visualises the Schumann resonance of Earth's ionosphere (7.83 Hz). Aurora particles vibrate in sync with the planet's own electromagnetic frequency. Raising the Harmonics slider adds more complex waveforms in superposition. Pulsing rings and particles move in perfect phase synchronisation. ❄️ Tab 5 — Arctic Scene A complete Arctic night scene. Twinkling starfield, moonlight glow, three-layer aurora, snow-dusted tundra silhouette, pine tree shadows, ice surface shimmer, and wind-driven snowfall — all capturing the atmosphere of Tromsø, Norway or Fairbanks, Alaska on a clear winter night.

📊 탭 1 — 핵심 확률 KPI 4개: 화석화 확률(0.00001%), 지표 노출(1/1천만), 발견·수집(1/10억), 현재 명명 종 수(~1,000종) 단계별 확률 감소 바 차트 (로그 스케일) 도트 매트릭스 2개: 1만 마리 중 화석화 1개 / 화석 1,000개 중 발견 1개 1억 마리 공룡으로 시작하는 퍼널 차트 🔬 탭 2 — 단계별 분석 사망→매몰→광물화→지각 생존→노출→파손 방지→발견까지 7단계 막대 그래프 각 단계 설명 카드 6장 서식 환경별 화석화 성공률 비교 (강·호수 85% ~ 오픈 평원 5%) 🌍 탭 3 — 지역별 발견율 대륙별 누적 발견 종 수 막대 그래프 (중국 350종 1위) 지역 카드 6장 (면적당 밀도 포함) 1824~2024년 신종 기재 누적 꺾은선 그래프 (시대별 에라 배경 포함) ⚖️ 탭 4 — 비교 시각화 공룡 화석 발견 vs 로또 1등·번개·비행기 사고 확률 로그 스케일 비교 화석 유형별 희귀도 버블 차트 (이빨~피부 인상 8종) 지질시대별 화석 발견 밀도 타임라인 (트라이아스기~백악기 후기) 🧮 탭 5 — 확률 계산기 6개 조건(서식 환경·개체 크기·뼈 밀도·퇴적 속도·경과 시간·탐사 접근성) 설정 실시간 확률 계산 및 탐사 접근성 곡선 업데이트 시나리오별 예상 확률 카드 6장 (이상적 조건~깃털 인상) 확률 계산 공식 및 참고 문헌 표시

주요 기능 입력창에 타이핑하면 120ms 디바운스로 즉시 재계산 (별도 버튼 없이 실시간) 게이지: 파랑(무거움/어려움) → 회갈색(보통) → 주황(가벼움/쉬움) 그라데이션 위를 바늘이 이동 점수(0~100), 등급(매우 쉬움~매우 어려움), 설명 문구 함께 표시 문장 수 / 평균 문장 길이 / 평균 어절 길이 / 긴 문장 비율 지표 카드 본문 아래에 유독 길거나 무거운 문장은 밑줄 색으로 표시

주요 기능 문장 자동 분리 + 내장 한국어 감성 어휘 사전으로 점수 계산 (부정어 "안/못/않" 반전, "매우/너무/정말" 등 강조어 가중치 처리) 각 문장에 색상 밑줄 + 점수 배지 표시 상단 감정 스펙트럼 바에 평균 점수 마커 표시 긍정/중립/부정 문장 수 요약 및 분포 막대

The final score is a weighted combination of four algorithms, displayed as a ring gauge with a colour-coded verdict (Very High / High / Moderate / Low / Negligible). Cosine Similarity (45%) — the angle between the two TF-IDF vectors in high-dimensional term space Jaccard Index (20%) — the ratio of shared vocabulary to total unique vocabulary across both documents Bigram Overlap F1 (20%) — the fraction of 2-word phrases shared between the two texts Trigram Overlap F1 (15%) — the fraction of 3-word phrases shared; high trigram overlap is a strong indicator of direct copying

🌌 Tab 1 — Live Black Hole A real-time interactive simulation of Gargantua. Hover over the canvas to see how much light bends at that point, with the deflection angle shown live. Right-side control panel: Spin (a) — Black hole rotation speed (Kerr parameter). Higher values make one side dramatically brighter due to relativistic beaming. Disk brightness / thickness — Visual adjustments for the accretion disk appearance. View angle — Shift the viewing perspective from face-on (90°) to edge-on (0°). Hawking radiation — Visualises particle emission from the event horizon as a quantum effect. 📖 Tab 2 — Step by Step Walks through the six stages that produce Gargantua's appearance, in order: A massive star → 2. Core collapse (supernova) → 3. Schwarzschild radius → 4. Event horizon (sphere) → 5. Photon sphere → 6. Accretion disk + gravitational lensing 🔬 Tab 3 — Anatomy Click any structural component to open a detailed explanation in the side panel; the corresponding layer highlights on the canvas simultaneously. Components covered: Singularity · Event Horizon · Photon Sphere · ISCO (Innermost Stable Circular Orbit) · Accretion Disk · Ergosphere (spinning black holes only) 🌐 Tab 4 — Why a Sphere? Answers the question "it looks like a ring, so why is it called a sphere?" from three angles: Side view (the angle shown in the film) — explains why the back of the disk appears bent over the top through gravitational lensing Top view (looking straight down) — shows the shadow is always circular regardless of orientation 3D sphere structure — compass arrows illustrate that the event horizon sits at the same radius in every direction

Type a password to see how strong it actually is — entropy, estimated crack time under different attack conditions, and what's still missing. Nothing is sent anywhere; everything runs locally in your browser.

A genealogical map of how today's Indo-European languages descend from a shared ancestor. Click any branch to expand or collapse it, or search for a language to trace its lineage back to the root.

Top hero: Displays your local time zone, detected by the browser, as a large digital clock. City cards: New York, London, Dubai, Seoul, Tokyo, and Sydney are shown by default. Each card's background color shifts naturally with day and night (deep navy at midnight, bright sky blue at noon). An analog clock is shown alongside the digital time, weekday, and UTC offset. Add/remove cities: Select a city from the dropdown to add it, or use the × button on a card to remove it.

π (pi) is the ratio of a circle's circumference to its diameter — the same number, no matter how big or small the circle. Below are four different ways people have calculated it, from ancient geometry to modern random sampling. Drag the sliders and press the buttons to see each one work.

This is an HTML page that visualizes an animated Perlin noise field in real time, styled as a terrain (contour) map. Canvas: 3D Perlin noise (x, y, time) is computed per pixel using fBm (octave summation) and mapped to terrain colors (water–sand–grassland–forest–rock–snow). Hovering the mouse over the canvas shows the noise value at that coordinate in real time. Controls: Scale, Octaves, Persistence, and animation speed sliders / Terrain·Thermal·Grayscale palette switcher / Contour bands toggle / a Reseed button to generate a new terrain.

6 Strike Modes 🖱 Click Strike — Fires a bolt from the top of the canvas to your cursor. Every click generates a unique fractal path. ✏️ Drag Path — Draw a route by clicking and dragging; bolts are generated along the entire path when you release. 💥 Multi-Bolt — One click spawns 3–5 bolts simultaneously at slightly offset positions, each with a small staggered delay. 🔗 Chain Lightning — Lightning hops through a midpoint before reaching the target, creating a relay-style chain strike. ⚪ Ball Lightning — Spawns a physics-simulated plasma orb with gravity, wall reflection, and surface sparks. ⬆️ Upward Strike — Simulates a ground-to-cloud return stroke — the actual visible flash that shoots upward in real lightning. ⚙️ Shape Controls Five sliders to fully customise the bolt shape: Jaggedness · Segments · Branching (0–0.7) · Branch length · Core width. ✨ Appearance Controls Core width · Glow radius · Flash duration · Auto interval 8 colour presets — Classic (white/blue), Gold, Purple, Green, Red, Cyan, Pink, Orange Hue shift slider for real-time colour rotation using CSS filter: hue-rotate() 🌩 Storm BG — Activates a dark cloud gradient background with 60 animated rain streaks rendered via Float32Arrays. 🔊 Sound — Web Audio API generates a low-frequency thunder rumble using a noise buffer filtered through a lowpass filter. ⌨️ Keyboard Shortcuts Space — instant strike at a random position A — toggle auto-strike C — clear the canvas G — toggle glow effect 1 – 6 — switch strike mode ↑ / ↓ — adjust jaggedness

🧱 Tab 1 — Anatomy Breaks a real prompt down into six colour-coded blocks. Click any block to open a detailed explanation in the right panel covering the component's purpose, how to write it, common mistakes, and actionable tips. 🎭 Role (purple) — persona definition 📋 Context (blue) — background information 🎯 Task (green) — the core instruction 📐 Format (yellow) — output structure 📸 Examples (orange) — few-shot samples 🚧 Constraints (red) — what to exclude Six quality score bars are shown at the bottom — Clarity, Specificity, Context depth, and more. The six technique cards below the blocks link directly to Tab 3 when clicked. ⚖️ Tab 2 — Before / After Places a weak prompt and a fully engineered prompt side by side for the same request. Four real-world scenarios included: 📧 Sales email writing 💻 Code review 📊 Data analysis ✍️ Creative writing Each pair shows actual AI response examples alongside quality tags (❌ Vague / ✅ Role defined, etc.). 🎯 Tab 3 — Techniques Eight techniques explained using a left-side navigation and right-side detail panel layout: Chain of Thought · Few-Shot · Role Prompting · Output Formatting · Self-Consistency · Tree of Thought · Iterative Refinement · Constraint Prompting Each technique includes a concept explanation, a side-by-side prompt comparison (❌ without the technique / ✅ with it), and four practical tips. 🛠️ Tab 4 — Playground Build your own prompt using four text inputs (Role · Context · Task · Format) and four technique toggles (Chain of Thought · Role Prompting · Few-Shot · Structured Output). Clicking ⚡ Build & Analyse produces: A fully assembled prompt preview with each component highlighted in its own colour A quality score (0–100) plus estimated token and word count A component checklist (✓ / ✗) showing what is present and what is missing Specific improvement suggestions — add a role, prompt too short or too long, consider CoT, and so on

⚖️ Measurement Input Metric / Imperial toggle — instantly converts between kg·cm and lbs·ft·in Slider + number input, two-way sync — moving the slider updates the number box and vice versa Age input and biological sex selection (used as a reference for healthy ranges) 📊 BMI Gauge Semi-circular analogue gauge — the needle rotates smoothly to match the BMI value via CSS transition Five colour zones: Blue (underweight) → Green (normal) → Yellow (overweight) → Orange (obese I) → Red (obese II+) Active category is highlighted; all others are dimmed 📐 Body Metrics Panel Height (m), weight (kg), BMI, and category displayed in real time Ideal weight range bar (WHO standard BMI 18.5–24.9) with a marker showing current weight position Automatically calculates how much weight to lose or gain to reach the ideal range 💡 Health Guidance (4 tips per category) Underweight — increase caloric intake, strength training, protein at every meal, rule out underlying causes Normal — maintenance strategies, exercise routine, balanced diet, sleep Overweight — 500 kcal daily deficit, walking, reduce ultra-processed food, hydration Obese I — see a doctor, 5–10% weight loss impact, portion control, stress management Obese II+ — medical supervision, metabolic assessment, GLP-1 medications, build a support network 📋 Reference Information WHO BMI Classification (2004) explained Separate guidance for Asian populations (overweight ≥ 23, obese ≥ 27.5) Medical disclaimer noting BMI is a screening tool, not a diagnostic measure

🖱 Drag to Flick mode Grab the wheel directly with your mouse or finger, drag it, then let go — the wheel flicks like a wrist snap. The faster you drag, the faster it spins. A physics engine applies friction to decelerate the wheel naturally until it stops. The moment it settles, the pointer reads whichever segment it points to and announces the result automatically. 🔘 Hold to Spin mode (default) Hold the button down and the wheel continuously accelerates; release it and the wheel decelerates to a stop. The result is entirely down to your instinct — when you press and when you let go. Full mobile touch support included. ⚙️ Physics parameters Friction slider — five steps from Very slow (spins for a long time) to Very fast (stops quickly) Hold speed slider — controls how fast the wheel accelerates while the button is held Common features Switch between modes at any time using the mode tabs at the top A purple glow ring appears around the wheel edge when spinning at high speed Tick sound on each segment crossing; victory melody + 🎉 confetti when the wheel stops History, probability pie, and statistics panel all carry over unchanged

☁️ 8 Cloud Types Cloud Altitude Description ⛅ Cumulus 600–2,000 m Multiple cloud clusters drifting with the wind 🌫 Stratus 0–600 m Uniform layer blanketing the entire sky 🌤 Cirrus 6,000–12,000 m Ice-crystal streaks in wispy mare's-tail shapes ⛈ Cumulonimbus All levels 8-layer towering cell with anvil top at tropopause 🌥 Altocumulus 2,000–6,000 m Grid-like pattern of rounded white-grey puffs 🌧 Nimbostratus Low–mid level Thick dark sheet producing continuous precipitation 🌁 Fog / Mist Ground level 5-layer flowing fog rendered at surface level 🔮 Mammatus Underside of Cb Hanging pouch formations (3 rows × multiple columns) 🎮 Interactive Features ☀️ Day / 🌙 Night — Toggles the sky gradient; night mode renders stars and a crescent moon 💨 Wind — 80 wind-streak particles linked to the Wind speed slider 🌧 Rain — 400-raindrop physics simulation with wind-angle deflection ⚡ Lightning — Fractal lightning bolts generated at random intervals, with higher frequency during Cumulonimbus and high-instability conditions ⏸ Pause — Freezes the animation ⚙️ Atmosphere Sliders — Temperature · Humidity · Wind speed · Atmospheric instability · Time of day (sunrise / midday / sunset / night auto-transition) 📊 Live Weather Readouts — Dew point · Cloud base altitude (LCL) · CAPE (Convective Available Potential Energy) · Visibility 🖼 Background Elements — Rolling hill silhouette + 18-tree treeline + surface fog layer tied to the humidity slider 📋 Right Panel — Per-cloud description · Step-by-step formation process · Weather forecast significance · Atmospheric layer altitude reference

🏗️ Architecture Presets (5 layouts) XOR — 2→4→1. The simplest non-linear classification example. Classifier — 4→6→6→3. Iris dataset-style multi-class classification. Deep — 3→5→5→5→5→2. A deep network for observing vanishing gradient behaviour. Autoencoder — 6→3→3→6. Bottleneck structure for dimensionality reduction. CNN-like — 6→8→8→4. Wide hidden layers resembling a convolutional network. 📐 Layer Editor — Use the Input / Hidden layers / Nodes per layer / Output sliders to build any custom architecture on the fly. ⚡ 6 Activation Functions — ReLU · Sigmoid · Tanh · Leaky ReLU · ELU · Softmax. Switching between them instantly updates the mini function plot in the top-left corner and the description in the right panel. 🎬 3 Modes Training — SGD simulation. Blue circular particles (forward pass) and red triangular particles (backpropagation) flow simultaneously while Loss, Accuracy, and Epoch update in real time. Forward pass — Only the forward-pass particles flow through the network. Backprop — Only the backward gradient particles flow toward the input. 🎨 Node colour meaning 🟢 Green — High activation (> 0.7) 🔵 Cyan — Mid-range activation 🔴 Red — Negative activation ⭕ Dashed circle × — Node deactivated by Dropout 🖱 Interactive Hover a node → tooltip showing its activation and bias values Hover a connection → weight value displayed at the midpoint Click an input node → randomises its activation and propagates the new values forward 📊 Right panel Live statistics: Epoch · Loss · Accuracy · Parameters Dual-line chart: Loss (red) and Accuracy (green) over training time Weight distribution histogram with a rainbow colour gradient Four-step Forward Pass walkthrough and activation function concept guide

🌀 Rotating Rings — Stare at the centre and the concentric ring pattern appears to physically rotate. High-contrast luminance steps misfiring motion-detector cells in your peripheral vision — a Peripheral Drift effect. 🧱 Café Wall — Perfectly parallel horizontal lines appear to converge like trapezoids. The grey mortar joints between offset black-and-white tiles confuse orientation-selective neurons. Row offset and mortar thickness are both adjustable. ⬛ Hermann Grid — Grey ghost dots appear at the intersections between white squares, then vanish when you look directly at them. Caused by lateral inhibition in the retinal ganglion cells. 💃 Spinning Dancer — The silhouette dancer can be seen spinning either clockwise or counterclockwise. Focus on the foot or the shadow to flip the perceived direction of rotation. ⭕ Ebbinghaus Illusion — Two identical circles look different in size depending on the size of the surrounding circles. A slider controls the surround size so you can strengthen or weaken the effect. ↔️ Müller-Lyer — Identical line segments look different in length depending on the direction of the arrowheads. Arrowhead size and line length are both adjustable. An "EQUAL" bracket on the right confirms the lines are the same. 🏺 Rubin's Vase — The image can be read as either a white vase or two faces looking at each other. A classic figure–ground illusion where both interpretations compete inside the brain. 💧 Motion Aftereffect — Stare at the flowing stripes on the left for 20 seconds, then shift your gaze to the static grid on the right — the grid will appear to drift in the opposite direction. A static demonstration of the Waterfall Illusion. 🔲 Simultaneous Contrast — Two squares of identical grey (RGB 128) look dramatically different in brightness depending on the background behind them. A BG contrast slider lets you tune the strength of the effect. ⚡ Scintillating Grid — White dots at intersections seem to flicker and disappear as your eyes move across the pattern. An enhanced version of the Hermann Grid, with adjustable square size, gap, and dot radius. 📦 Necker Cube — A wire-frame cube appears to project outward in two different directions, and your perception flips as you stare at it. The brain alternates between two equally valid 3D interpretations in the absence of depth cues. 👁 Troxler's Fading — Fix your gaze on the central cross without moving your eyes and the surrounding coloured ring slowly fades and disappears. Neural adaptation reduces the brain's response to unchanging stimuli — revealing how much of your visual field is actively ignored.

10 interactive Network Effects visualizers! 📡 Metcalfe's Law — Animates every edge formed as n users connect in real time. Adjust the user count slider and the number of connections (n(n-1)/2) and network value (n²) update instantly. A comparison chart overlays the Linear, Metcalfe, and Reed's Law growth curves side by side. 📈 Adoption Curve — Rogers' technology adoption S-curve unfolds along a time axis. Adjust the tipping point position and growth rate k to visualise the exact moment of "crossing the chasm," with a five-segment adoption band (Innovators → Laggards) displayed below the chart. 🦠 Viral Spreading (SIR) — 200 particles move between S (blue), I (red), and R (grey) states as the epidemic spreads. An R₀ = β/γ gauge updates in real time — above 1 the infection spreads, below 1 it dies out. Includes an explanation of how this maps to the product K-factor. 🗂️ Network Types — Five types (Direct, Indirect, Data, Social, Local) are laid out as interactive cards, each containing a live mini network graph of its own structure. A bar chart at the bottom compares the competitive moat strength of each type. 💥 Critical Mass — A live level gauge and time-series chart show whether a network self-sustains or collapses depending on whether it sits above or below the threshold. Adjust the threshold and seed percentage to catch the exact moment a tipping point is crossed. 🏗️ Platform Competition — A pie chart and time-series graph together show how two platforms' market shares converge toward natural monopoly under network effects. Tune initial size and product quality to experiment with the conditions that trigger Winner-Take-All dominance. 🌀 Churn & Retention — A spinning flywheel diagram visualises how retention rate, viral coefficient, and new user acquisition interlock. The long-run user count trend and LTV estimate update in real time as you adjust the sliders. 🌍 Geographic Spread — Diffusion waves (expanding rings) radiate from the launch city and claim neighbouring cities one by one. A cascade effect is implemented — an adopted city shortens the adoption timer of nearby cities — letting you experience the city-by-city rollout strategy used by Uber and Airbnb. 📱 Social Network — Nodes are added in real time through invite loops, and nodes with more connections are rendered larger. The right-hand panel covers social graph growth insights including Facebook's north-star metric of "7 friends in 10 days." 🛒 Marketplace (Two-Sided) — Live transactions (yellow $ particles) flow back and forth between the buyer zone (blue) and seller zone (green), and both sides grow automatically as transactions accumulate. The right panel explains liquidity score, the cold-start problem, and the concept of rake rate.