SentAI / index.html
Solar-Prince's picture
Upload 2 files
7c3874c verified
Raw History Blame Contribute Delete
52.8 kB
<!doctype html> <html lang="en"> <head> <meta charset="utf-8" /> <meta name="viewport" content="width=device-width, initial-scale=1, viewport-fit=cover" /> <meta name="theme-color" content="#050816" /> <title>SentAI</title> <script defer src="https://cdn.jsdelivr.net/npm/face-api.js@0.22.2/dist/face-api.min.js"></script> <script defer src="https://cdn.jsdelivr.net/npm/onnxruntime-web@1.21.0/dist/ort.min.js"></script> <style> :root { --bg-a: #050816; --bg-b: #0d1b2f; --card: rgba(15, 23, 42, 0.76); --card-strong: rgba(15, 23, 42, 0.94); --stroke: rgba(148, 163, 184, 0.24); --text: #f8fafc; --muted: #a7b4c8; --soft: rgba(255,255,255,0.08); --accent: #22d3ee; --accent-2: #a78bfa; --good: #2dd4bf; --warn: #fbbf24; --bad: #fb7185; --shadow: 0 24px 90px rgba(0,0,0,0.42); --radius: 28px; font-family: Inter, ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif; } * { box-sizing: border-box; } body { margin: 0; min-height: 100vh; color: var(--text); background: radial-gradient(circle at 12% 10%, rgba(34, 211, 238, 0.26), transparent 34%), radial-gradient(circle at 92% 2%, rgba(167, 139, 250, 0.23), transparent 34%), linear-gradient(135deg, var(--bg-a), var(--bg-b) 52%, #112236); overflow-x: hidden; } body::before { content: ""; position: fixed; inset: 0; pointer-events: none; background-image: linear-gradient(rgba(255,255,255,0.035) 1px, transparent 1px), linear-gradient(90deg, rgba(255,255,255,0.035) 1px, transparent 1px); background-size: 44px 44px; mask-image: radial-gradient(circle at 50% 20%, black 0%, transparent 75%); } .app-shell { width: min(1460px, calc(100% - 32px)); margin: 0 auto; padding: 34px 0 42px; position: relative; z-index: 1; } header.hero { display: grid; grid-template-columns: 1fr auto; align-items: center; gap: 22px; margin-bottom: 26px; } .brand h1 { margin: 0; font-size: clamp(4.2rem, 9vw, 8rem); line-height: 0.88; letter-spacing: -0.08em; font-weight: 950; text-shadow: 0 24px 70px rgba(34, 211, 238, 0.10); } .brand p { margin: 18px 0 0; max-width: 960px; color: var(--muted); font-size: clamp(1rem, 1.7vw, 1.22rem); line-height: 1.55; } .status-stack { display: grid; gap: 10px; justify-items: end; } .status-pill { display: inline-flex; align-items: center; gap: 10px; border: 1px solid var(--stroke); background: rgba(15, 23, 42, 0.64); backdrop-filter: blur(18px); border-radius: 999px; padding: 12px 16px; color: var(--muted); box-shadow: var(--shadow); white-space: nowrap; font-weight: 850; } .dot { width: 10px; height: 10px; border-radius: 50%; background: var(--warn); box-shadow: 0 0 18px var(--warn); } .dot.ready { background: var(--good); box-shadow: 0 0 18px var(--good); } .dot.error { background: var(--bad); box-shadow: 0 0 18px var(--bad); } .toolbar { display: flex; flex-wrap: wrap; gap: 12px; align-items: center; margin-bottom: 18px; } button { appearance: none; border: 1px solid var(--stroke); background: rgba(15, 23, 42, 0.76); color: var(--text); border-radius: 999px; padding: 13px 18px; font-size: 0.98rem; font-weight: 850; letter-spacing: -0.01em; cursor: pointer; transition: transform 160ms ease, border-color 160ms ease, background 160ms ease, opacity 160ms ease; min-height: 48px; } button:hover { transform: translateY(-1px); border-color: rgba(34, 211, 238, 0.55); } button:active { transform: translateY(0); } button.primary { background: linear-gradient(135deg, rgba(34,211,238,0.96), rgba(167,139,250,0.94)); color: #06111f; border-color: transparent; } button.danger { color: #fecdd3; } button:disabled { opacity: 0.46; cursor: not-allowed; transform: none; } .camera-tag { padding: 12px 16px; min-height: 48px; display: inline-flex; align-items: center; border-radius: 999px; border: 1px solid var(--stroke); background: rgba(255,255,255,0.055); color: var(--muted); font-weight: 850; } .grid { display: grid; grid-template-columns: minmax(0, 1.42fr) minmax(360px, 0.76fr); gap: 18px; align-items: stretch; } .panel { background: var(--card); border: 1px solid var(--stroke); border-radius: var(--radius); box-shadow: var(--shadow); backdrop-filter: blur(22px); overflow: hidden; } .stage-panel { padding: 14px; } .stage { position: relative; width: 100%; aspect-ratio: 4 / 3; border-radius: calc(var(--radius) - 10px); overflow: hidden; background: radial-gradient(circle at 50% 30%, rgba(34,211,238,0.09), transparent 36%), #020617; border: 1px solid rgba(148, 163, 184, 0.18); } video, canvas#overlay { position: absolute; inset: 0; width: 100%; height: 100%; object-fit: contain; } video { background: #020617; } canvas#overlay { pointer-events: none; } .empty-state { position: absolute; inset: 0; display: grid; place-items: center; padding: 28px; text-align: center; color: var(--muted); pointer-events: none; } .empty-card { max-width: 620px; border: 1px solid var(--stroke); background: rgba(15,23,42,0.70); border-radius: 24px; padding: 28px; } .empty-card strong { display: block; color: var(--text); font-size: clamp(1.35rem, 3vw, 2.2rem); margin-bottom: 12px; } .side { padding: 18px; display: grid; gap: 14px; align-content: start; } .side h2 { margin: 2px 0 4px; font-size: 1.35rem; letter-spacing: -0.04em; } .metric-grid { display: grid; grid-template-columns: repeat(2, minmax(0, 1fr)); gap: 12px; } .metric { border: 1px solid var(--stroke); background: rgba(255,255,255,0.055); border-radius: 22px; padding: 16px; min-height: 112px; } .metric span { display: block; color: var(--muted); font-size: 0.82rem; font-weight: 850; text-transform: uppercase; letter-spacing: 0.08em; } .metric strong { display: block; margin-top: 10px; font-size: clamp(1.55rem, 3.5vw, 2.25rem); letter-spacing: -0.06em; line-height: 1; } .metric small { display: block; margin-top: 7px; color: var(--muted); font-weight: 750; } .wide { grid-column: 1 / -1; } .bars, .note { border: 1px solid var(--stroke); background: rgba(255,255,255,0.055); border-radius: 22px; padding: 16px; } .bars h3 { margin: 0 0 14px; font-size: 1rem; color: var(--text); letter-spacing: -0.02em; } .bar-row { display: grid; grid-template-columns: 86px 1fr 48px; gap: 10px; align-items: center; margin: 11px 0; color: var(--muted); font-size: 0.9rem; font-weight: 850; } .bar-track { height: 9px; border-radius: 99px; background: rgba(148,163,184,0.2); overflow: hidden; } .bar-fill { width: 0%; height: 100%; border-radius: inherit; background: linear-gradient(90deg, var(--accent), var(--accent-2)); transition: width 180ms ease; } .note { color: #cffafe; background: rgba(34,211,238,0.08); border-color: rgba(34,211,238,0.24); line-height: 1.5; font-size: 0.94rem; } .toast { position: fixed; left: 50%; bottom: 22px; transform: translateX(-50%) translateY(30px); opacity: 0; z-index: 10; background: rgba(2, 6, 23, 0.94); border: 1px solid var(--stroke); border-radius: 999px; padding: 13px 18px; box-shadow: var(--shadow); color: var(--text); font-weight: 850; transition: opacity 180ms ease, transform 180ms ease; max-width: min(92vw, 760px); text-align: center; } .toast.show { opacity: 1; transform: translateX(-50%) translateY(0); } @media (max-width: 980px) { .app-shell { width: min(100% - 20px, 760px); padding-top: 22px; } header.hero { grid-template-columns: 1fr; } .status-stack { justify-items: start; } .status-pill { width: fit-content; } .grid { grid-template-columns: 1fr; } .stage-panel { padding: 10px; } .side { padding: 14px; } .toolbar { gap: 9px; } button, .camera-tag { flex: 1 1 150px; justify-content: center; } } @media (max-width: 560px) { .brand h1 { font-size: clamp(4rem, 22vw, 5.4rem); } .brand p { font-size: 0.98rem; } .metric-grid { grid-template-columns: 1fr; } .bar-row { grid-template-columns: 76px 1fr 44px; font-size: 0.82rem; } .empty-card { padding: 20px; } .app-shell { width: min(100% - 14px, 760px); } body { background-attachment: fixed; } } </style> </head> <body> <main class="app-shell"> <header class="hero" aria-label="SentAI heading"> <div class="brand"> <h1>SentAI</h1> <p>Live facial analysis with automatic high-precision models for expression, apparent age range, and male/female estimate. No manual model switches.</p> </div> <div class="status-stack"> <div class="status-pill" aria-live="polite"><span id="aiDot" class="dot"></span><span id="aiStatus">Preparing AI models...</span></div> <div class="status-pill" aria-live="polite"><span id="precisionDot" class="dot"></span><span id="precisionStatus">Precision models loading automatically</span></div> </div> </header> <section class="toolbar" aria-label="Camera controls"> <button id="startBtn" class="primary" disabled>Start camera</button> <button id="switchBtn" disabled>Switch front/rear</button> <button id="stopBtn" class="danger" disabled>Stop</button> <span id="cameraTag" class="camera-tag">Camera: not started</span> </section> <section class="grid"> <div class="panel stage-panel"> <div id="stage" class="stage"> <video id="video" autoplay muted playsinline></video> <canvas id="overlay"></canvas> <div id="emptyState" class="empty-state"> <div class="empty-card"> <strong>Ready for live analysis</strong> Tap <b>Start camera</b>. Use bright front lighting, keep one face centered, and hold each expression for a moment so the neural ensemble can stabilize. </div> </div> </div> </div> <aside class="panel side" aria-label="Live analysis panel"> <h2>Live details</h2> <div class="metric-grid"> <div class="metric"> <span>Possible feeling</span> <strong id="feelingValue">-</strong> <small id="feelingConfidence">Waiting</small> </div> <div class="metric"> <span>Gender</span> <strong id="genderValue">-</strong> <small id="genderConfidence">Waiting</small> </div> <div class="metric"> <span>Apparent age</span> <strong id="ageValue">-</strong> <small id="ageSource">Waiting</small> </div> <div class="metric"> <span>Faces</span> <strong id="facesValue">0</strong> <small id="faceScore">No face yet</small> </div> <div class="metric"> <span>FPS</span> <strong id="fpsValue">-</strong> <small>Live loop</small> </div> <div class="metric"> <span>Latency</span> <strong id="latencyValue">-</strong> <small id="latencySource">Model time</small> </div> </div> <div class="bars wide"> <h3>Feeling scores</h3> <div id="emotionBars"></div> </div> <div class="note wide"> SentAI estimates visible facial expression and apparent age from camera frames. It cannot know a person's true internal feeling, but this version uses an automatic multi-model deep-learning ensemble to improve sad, fear, and disgust detection. </div> </aside> </section> </main> <div id="toast" class="toast" role="status" aria-live="polite"></div> <script type="module"> import { pipeline as xenovaPipeline, env as xenovaEnv } from "https://cdn.jsdelivr.net/npm/@xenova/transformers@2.17.2"; import { AutoModel, AutoProcessor, load_image, env as hfEnv } from "https://cdn.jsdelivr.net/npm/@huggingface/transformers@3.8.1"; const FACE_API_MODEL_URL = "https://cdn.jsdelivr.net/gh/justadudewhohacks/face-api.js@0.22.2/weights"; const EMOTION_TRANSFORMER_MODEL_ID = "Xenova/facial_emotions_image_detection"; const AGE_GENDER_MODEL_ID = "onnx-community/age-gender-prediction-ONNX"; const OPENCV_FER_MODEL_URL = "https://huggingface.co/opencv/facial_expression_recognition/resolve/main/facial_expression_recognition_mobilefacenet_2022july.onnx"; const els = { aiDot: document.getElementById("aiDot"), aiStatus: document.getElementById("aiStatus"), precisionDot: document.getElementById("precisionDot"), precisionStatus: document.getElementById("precisionStatus"), startBtn: document.getElementById("startBtn"), switchBtn: document.getElementById("switchBtn"), stopBtn: document.getElementById("stopBtn"), cameraTag: document.getElementById("cameraTag"), video: document.getElementById("video"), overlay: document.getElementById("overlay"), stage: document.getElementById("stage"), emptyState: document.getElementById("emptyState"), feelingValue: document.getElementById("feelingValue"), feelingConfidence: document.getElementById("feelingConfidence"), genderValue: document.getElementById("genderValue"), genderConfidence: document.getElementById("genderConfidence"), ageValue: document.getElementById("ageValue"), ageSource: document.getElementById("ageSource"), facesValue: document.getElementById("facesValue"), faceScore: document.getElementById("faceScore"), fpsValue: document.getElementById("fpsValue"), latencyValue: document.getElementById("latencyValue"), latencySource: document.getElementById("latencySource"), emotionBars: document.getElementById("emotionBars"), toast: document.getElementById("toast"), }; const emotionLabels = ["Happy", "Sad", "Fear", "Anger", "Confused", "Disgust"]; const ferLabels = ["Anger", "Disgust", "Fear", "Happy", "Confused", "Sad", "Confused"]; const detectorConfig = { inputSize: 416, scoreThreshold: 0.32, intervalMs: 65, precisionIntervalMs: 650, smoothing: 0.26 }; const ctx = els.overlay.getContext("2d"); let coreReady = false; let stream = null; let running = false; let detecting = false; let currentFacing = "user"; let lastLoopTime = performance.now(); let fpsSmooth = 0; let lastDetections = []; let emotionEma = null; let genderMaleEma = null; let lastBaseAgeSample = 0; let latestPrimary = null; const ageHistory = []; const precision = { loading: false, readyCount: 0, ferSession: null, emotionPipe: null, ageModel: null, ageProcessor: null, lastRun: 0, busy: false, ferScores: null, ferAt: 0, transformerScores: null, transformerAt: 0, age: null, ageAt: 0, gender: null, genderScore: 0, genderAt: 0, latency: 0, }; function setPill(dot, label, text, kind = "loading") { label.textContent = text; dot.classList.remove("ready", "error"); if (kind === "ready") dot.classList.add("ready"); if (kind === "error") dot.classList.add("error"); } function showToast(message) { els.toast.textContent = message; els.toast.classList.add("show"); clearTimeout(showToast.timer); showToast.timer = setTimeout(() => els.toast.classList.remove("show"), 3400); } function clamp01(value) { return Math.max(0, Math.min(1, Number.isFinite(value) ? value : 0)); } function percent(value) { return `${Math.round(clamp01(value) * 100)}%`; } function sigmoid(x) { return 1 / (1 + Math.exp(-x)); } function softmax(values) { const list = Array.from(values || []).map(v => Number(v)); if (!list.length) return []; const max = Math.max(...list); const exps = list.map(v => Math.exp(v - max)); const sum = exps.reduce((a, b) => a + b, 0) || 1; return exps.map(v => v / sum); } function median(values) { const list = values.filter(Number.isFinite).slice().sort((a, b) => a - b); if (!list.length) return NaN; const mid = Math.floor(list.length / 2); return list.length % 2 ? list[mid] : (list[mid - 1] + list[mid]) / 2; } function blankScores() { return Object.fromEntries(emotionLabels.map(label => [label, 0])); } function normalizeScores(scores) { const out = blankScores(); for (const label of emotionLabels) out[label] = clamp01(scores[label] || 0); return out; } function renormalize(scores) { const out = normalizeScores(scores); const sum = emotionLabels.reduce((a, label) => a + out[label], 0); if (sum <= 0) return out; for (const label of emotionLabels) out[label] = out[label] / sum; return out; } function ageRange(age, source = "core", samples = 1) { if (!Number.isFinite(age)) return "-"; const corrected = Math.max(0, Math.min(100, age)); const half = source === "precision model" ? (samples >= 3 ? 4 : 5) : (samples >= 6 ? 6 : 8); const lo = Math.max(0, Math.round(corrected - half)); const hi = Math.min(100, Math.round(corrected + half)); if (hi <= 12) return "0-12"; return `${lo}-${hi}`; } function calibrateFaceApiExpressions(expressions = {}) { const raw = { happy: clamp01(expressions.happy), sad: clamp01(expressions.sad), fearful: clamp01(expressions.fearful), angry: clamp01(expressions.angry), disgusted: clamp01(expressions.disgusted), surprised: clamp01(expressions.surprised), neutral: clamp01(expressions.neutral), }; const nonNeutralTop = Math.max(raw.happy, raw.sad, raw.fearful, raw.angry, raw.disgusted, raw.surprised); const confused = Math.min(0.38, raw.neutral * 0.24 + raw.surprised * 0.42 + Math.max(0, 1 - nonNeutralTop) * 0.05); return normalizeScores({ Happy: Math.pow(raw.happy, 1.10), Sad: Math.pow(raw.sad, 0.96), Fear: Math.max(Math.pow(raw.fearful, 0.98), raw.surprised * raw.fearful * 0.55), Anger: Math.pow(raw.angry, 1.02), Disgust: Math.pow(raw.disgusted, 0.96), Confused: confused, }); } function normalizeExternalEmotion(outputs) { const scores = blankScores(); const list = Array.isArray(outputs) ? outputs : [outputs]; for (const item of list) { const label = String(item.label || item.class || "").toLowerCase(); const score = clamp01(item.score || item.probability || 0); if (label.includes("happy") || label.includes("happiness") || label.includes("joy")) scores.Happy = Math.max(scores.Happy, score); else if (label.includes("sad") || label.includes("sadness")) scores.Sad = Math.max(scores.Sad, score); else if (label.includes("fear") || label.includes("fearful")) scores.Fear = Math.max(scores.Fear, score); else if (label.includes("angry") || label.includes("anger")) scores.Anger = Math.max(scores.Anger, score); else if (label.includes("disgust") || label.includes("disgusted")) scores.Disgust = Math.max(scores.Disgust, score); else if (label.includes("surprise") || label.includes("neutral")) { const scaled = label.includes("neutral") ? score * 0.45 : score * 0.52; scores.Confused = Math.max(scores.Confused, Math.min(0.58, scaled)); } } return normalizeScores(scores); } function calibrateFinalEmotion(scores) { const priors = { Happy: 0.86, Sad: 1.18, Fear: 1.24, Anger: 0.96, Confused: 0.72, Disgust: 1.34 }; const out = blankScores(); for (const label of emotionLabels) { let v = clamp01(scores[label] || 0); if (["Sad", "Fear", "Disgust"].includes(label)) v = Math.pow(v, 0.92); if (label === "Happy") v = Math.pow(v, 1.08); if (label === "Confused") v = Math.pow(v, 1.04); out[label] = clamp01(v * priors[label]); } return renormalize(out); } function combineEmotionScores(faceScores) { const now = performance.now(); const ferFresh = precision.ferScores && (now - precision.ferAt < 3400); const transformerFresh = precision.transformerScores && (now - precision.transformerAt < 5000); const combined = blankScores(); let totalWeight = 0; const addWeighted = (scores, weight) => { if (!scores || weight <= 0) return; totalWeight += weight; for (const label of emotionLabels) combined[label] += (scores[label] || 0) * weight; }; if (ferFresh && transformerFresh) { addWeighted(precision.ferScores, 0.55); addWeighted(precision.transformerScores, 0.34); addWeighted(faceScores, 0.11); } else if (ferFresh) { addWeighted(precision.ferScores, 0.76); addWeighted(faceScores, 0.24); } else if (transformerFresh) { addWeighted(precision.transformerScores, 0.78); addWeighted(faceScores, 0.22); } else { addWeighted(faceScores, 1.0); } if (totalWeight > 0) { for (const label of emotionLabels) combined[label] /= totalWeight; } const calibrated = calibrateFinalEmotion(combined); if (!emotionEma) { emotionEma = calibrated; } else { for (const label of emotionLabels) { emotionEma[label] = emotionEma[label] * (1 - detectorConfig.smoothing) + calibrated[label] * detectorConfig.smoothing; } } return renormalize(emotionEma); } function topEmotion(scores) { const entries = Object.entries(scores).sort((a, b) => b[1] - a[1]); let [label, score] = entries[0] || ["Confused", 0]; const nonConfused = entries.filter(([name]) => name !== "Confused"); const [altLabel, altScore] = nonConfused[0] || [label, score]; if (label === "Confused" && altScore >= 0.22 && (score - altScore) < 0.10) { label = altLabel; score = altScore; } if (score < 0.18) { label = "Confused"; score = Math.max(scores.Confused || 0.16, 0.16); } return { label, score: clamp01(score) }; } function pushAgeSample(age, source, weight = 1) { if (!Number.isFinite(age) || age < 0 || age > 100) return; const now = performance.now(); ageHistory.push({ age, source, weight, t: now }); while (ageHistory.length > 50) ageHistory.shift(); const cutoff = now - 45000; while (ageHistory.length && ageHistory[0].t < cutoff) ageHistory.shift(); } function stableAgeEstimate() { const precisionSamples = ageHistory.filter(s => s.source === "precision model"); const usable = precisionSamples.length >= 2 ? precisionSamples : ageHistory; if (!usable.length) return { age: NaN, source: "Waiting", samples: 0 }; const expanded = []; for (const sample of usable) { const repeats = Math.max(1, Math.round(sample.weight || 1)); for (let i = 0; i < repeats; i += 1) expanded.push(sample.age); } return { age: median(expanded), source: precisionSamples.length >= 2 ? "precision model" : "core model", samples: usable.length, }; } function updateGenderEstimate(label, confidence, weight = 1) { if (!label) return; const pMale = label.toLowerCase() === "male" ? clamp01(confidence) : 1 - clamp01(confidence); const alpha = Math.min(0.68, 0.15 * weight); genderMaleEma = genderMaleEma === null ? pMale : genderMaleEma * (1 - alpha) + pMale * alpha; } function currentGender() { if (genderMaleEma === null) return { label: "-", confidence: 0 }; const label = genderMaleEma >= 0.5 ? "Male" : "Female"; const confidence = Math.max(genderMaleEma, 1 - genderMaleEma); return { label, confidence }; } function makeInsight(det) { const faceScores = calibrateFaceApiExpressions(det.expressions || {}); const scores = combineEmotionScores(faceScores); const emotion = topEmotion(scores); const now = performance.now(); if (Number.isFinite(det.age) && now - lastBaseAgeSample > 700) { pushAgeSample(det.age, "core model", 1); lastBaseAgeSample = now; } const coreGender = (det.gender || "").toLowerCase() === "female" ? "Female" : "Male"; updateGenderEstimate(coreGender, det.genderProbability || 0, 1); if (precision.gender && (now - precision.genderAt < 9000)) updateGenderEstimate(precision.gender, precision.genderScore, 3.8); const age = stableAgeEstimate(); const gender = currentGender(); return { emotionLabel: emotion.label, emotionScore: emotion.score, emotionScores: scores, gender: gender.label, genderScore: gender.confidence, ageRange: ageRange(age.age, age.source, age.samples), ageSource: age.samples ? `${age.source}, ${age.samples} samples` : "Waiting", faceScore: det.detection && det.detection.score ? det.detection.score : 0, }; } function renderBars(scores = {}) { els.emotionBars.innerHTML = emotionLabels.map(label => { const value = scores[label] || 0; return ` <div class="bar-row"> <span>${label}</span> <div class="bar-track"><div class="bar-fill" style="width:${Math.round(value * 100)}%"></div></div> <span>${percent(value)}</span> </div>`; }).join(""); } function resetDetails() { emotionEma = null; genderMaleEma = null; ageHistory.length = 0; els.feelingValue.textContent = "-"; els.feelingConfidence.textContent = "Waiting"; els.genderValue.textContent = "-"; els.genderConfidence.textContent = "Waiting"; els.ageValue.textContent = "-"; els.ageSource.textContent = "Waiting"; els.facesValue.textContent = "0"; els.faceScore.textContent = "No face yet"; els.latencyValue.textContent = "-"; els.latencySource.textContent = "Model time"; renderBars({}); } function choosePrimary(detections) { if (!detections || !detections.length) return null; return detections.slice().sort((a, b) => { const ab = a.detection.box; const bb = b.detection.box; return (bb.width * bb.height) - (ab.width * ab.height); })[0]; } function updatePrecisionStatus() { const parts = []; if (precision.ferSession) parts.push("MobileFaceNet emotion"); if (precision.emotionPipe) parts.push("ViT emotion"); if (precision.ageModel && precision.ageProcessor) parts.push("ViT age/gender"); if (parts.length >= 2) setPill(els.precisionDot, els.precisionStatus, `Precision ready: ${parts.join(" + ")}`, "ready"); else if (parts.length === 1) setPill(els.precisionDot, els.precisionStatus, `Precision partially ready: ${parts[0]}`, "ready"); else if (precision.loading) setPill(els.precisionDot, els.precisionStatus, "Precision models loading automatically", "loading"); else setPill(els.precisionDot, els.precisionStatus, "Precision models unavailable; using core AI", "error"); } function updateDetails(detections, elapsedMs) { els.facesValue.textContent = String(detections.length); els.latencyValue.textContent = `${Math.round(elapsedMs)}ms`; els.latencySource.textContent = precision.latency ? `Core + precision ${Math.round(precision.latency)}ms` : "Core model time"; const now = performance.now(); const instantFps = 1000 / Math.max(1, now - lastLoopTime); lastLoopTime = now; fpsSmooth = fpsSmooth ? (fpsSmooth * 0.82 + instantFps * 0.18) : instantFps; els.fpsValue.textContent = fpsSmooth.toFixed(1); const primary = choosePrimary(detections); latestPrimary = primary; if (!primary) { els.feelingValue.textContent = "-"; els.feelingConfidence.textContent = "No face"; els.genderValue.textContent = "-"; els.genderConfidence.textContent = "No face"; els.ageValue.textContent = "-"; els.ageSource.textContent = "No face"; els.faceScore.textContent = "No face yet"; renderBars({}); return; } maybeRunPrecision(primary); const insight = makeInsight(primary); els.feelingValue.textContent = insight.emotionLabel; els.feelingConfidence.textContent = `${percent(insight.emotionScore)} confidence`; els.genderValue.textContent = insight.gender; els.genderConfidence.textContent = insight.gender === "-" ? "Waiting" : `${percent(insight.genderScore)} confidence`; els.ageValue.textContent = insight.ageRange; els.ageSource.textContent = insight.ageSource; els.faceScore.textContent = `${percent(insight.faceScore)} face score`; renderBars(insight.emotionScores); } function drawRoundRect(context, x, y, w, h, r) { const radius = Math.min(r, w / 2, h / 2); context.beginPath(); context.moveTo(x + radius, y); context.arcTo(x + w, y, x + w, y + h, radius); context.arcTo(x + w, y + h, x, y + h, radius); context.arcTo(x, y + h, x, y, radius); context.arcTo(x, y, x + w, y, radius); context.closePath(); } function drawCornerBox(x, y, w, h) { const line = Math.max(4, els.overlay.width / 180); const len = Math.max(26, Math.min(w, h) * 0.22); ctx.lineWidth = line; ctx.strokeStyle = "#22d3ee"; ctx.shadowColor = "rgba(34,211,238,0.85)"; ctx.shadowBlur = 14; ctx.beginPath(); ctx.moveTo(x, y + len); ctx.lineTo(x, y); ctx.lineTo(x + len, y); ctx.moveTo(x + w - len, y); ctx.lineTo(x + w, y); ctx.lineTo(x + w, y + len); ctx.moveTo(x + w, y + h - len); ctx.lineTo(x + w, y + h); ctx.lineTo(x + w - len, y + h); ctx.moveTo(x + len, y + h); ctx.lineTo(x, y + h); ctx.lineTo(x, y + h - len); ctx.stroke(); ctx.shadowBlur = 0; ctx.lineWidth = Math.max(2, line * 0.55); ctx.strokeStyle = "rgba(255,255,255,0.95)"; drawRoundRect(ctx, x, y, w, h, 16); ctx.stroke(); } function drawLabel(x, y, w, insight) { const scale = Math.max(1, els.overlay.width / 850); const padX = 12 * scale; const padY = 8 * scale; const font1 = Math.round(18 * scale); const font2 = Math.round(14 * scale); const lineH = font1 + 9 * scale; const label1 = `${insight.emotionLabel} ${percent(insight.emotionScore)}`; const label2 = `${insight.gender} ${percent(insight.genderScore)} | Age ${insight.ageRange}`; ctx.font = `900 ${font1}px Inter, system-ui, sans-serif`; const textW = Math.max(ctx.measureText(label1).width, ctx.measureText(label2).width); const boxW = Math.min(Math.max(textW + padX * 2, w * 0.72), els.overlay.width - 16); const boxH = lineH * 2 + padY * 1.6; const bx = Math.max(8, Math.min(x, els.overlay.width - boxW - 8)); const by = Math.max(8, y - boxH - 8); drawRoundRect(ctx, bx, by, boxW, boxH, 14 * scale); ctx.fillStyle = "rgba(2, 6, 23, 0.84)"; ctx.fill(); ctx.strokeStyle = "rgba(34, 211, 238, 0.65)"; ctx.lineWidth = 1.5 * scale; ctx.stroke(); ctx.fillStyle = "#ecfeff"; ctx.font = `900 ${font1}px Inter, system-ui, sans-serif`; ctx.fillText(label1, bx + padX, by + padY + font1); ctx.fillStyle = "#a7b4c8"; ctx.font = `800 ${font2}px Inter, system-ui, sans-serif`; ctx.fillText(label2, bx + padX, by + padY + font1 + lineH); } function drawDetections(detections) { ctx.clearRect(0, 0, els.overlay.width, els.overlay.height); lastDetections = detections; for (const det of detections) { const box = det.detection.box; const x = Math.max(0, box.x); const y = Math.max(0, box.y); const w = Math.min(box.width, els.overlay.width - x); const h = Math.min(box.height, els.overlay.height - y); const insight = makeInsight(det); drawCornerBox(x, y, w, h); drawLabel(x, y, w, insight); } } function fitCanvas() { const vw = els.video.videoWidth || 640; const vh = els.video.videoHeight || 480; if (els.overlay.width !== vw || els.overlay.height !== vh) { els.overlay.width = vw; els.overlay.height = vh; } els.stage.style.aspectRatio = `${vw} / ${vh}`; if (lastDetections.length) drawDetections(lastDetections); } function cropFaceCanvas(det, targetSize = 288, pad = 0.28, filter = "none", mirror = false) { const box = det.detection.box; const videoW = els.video.videoWidth || els.overlay.width; const videoH = els.video.videoHeight || els.overlay.height; const cx = box.x + box.width / 2; const cy = box.y + box.height / 2; const side = Math.max(box.width, box.height) * (1 + pad * 2); const sx = Math.max(0, Math.round(cx - side / 2)); const sy = Math.max(0, Math.round(cy - side / 2)); const sw = Math.min(videoW - sx, Math.round(side)); const sh = Math.min(videoH - sy, Math.round(side)); const canvas = document.createElement("canvas"); canvas.width = targetSize; canvas.height = targetSize; const c = canvas.getContext("2d", { willReadFrequently: true }); c.fillStyle = "#000"; c.fillRect(0, 0, targetSize, targetSize); c.filter = filter; if (mirror) { c.translate(targetSize, 0); c.scale(-1, 1); c.drawImage(els.video, sx, sy, sw, sh, 0, 0, targetSize, targetSize); c.setTransform(1, 0, 0, 1, 0, 0); } else { c.drawImage(els.video, sx, sy, sw, sh, 0, 0, targetSize, targetSize); } c.filter = "none"; return canvas; } function canvasToBlobUrl(canvas) { return new Promise((resolve, reject) => { canvas.toBlob(blob => { if (!blob) reject(new Error("Could not create face crop")); else resolve(URL.createObjectURL(blob)); }, "image/jpeg", 0.92); }); } async function waitForGlobal(name, timeoutMs = 14000) { const started = performance.now(); while (!window[name]) { if (performance.now() - started > timeoutMs) throw new Error(`${name} did not load`); await new Promise(resolve => setTimeout(resolve, 80)); } return window[name]; } async function loadCoreModels() { try { await waitForGlobal("faceapi"); await Promise.all([ faceapi.nets.ssdMobilenetv1.loadFromUri(FACE_API_MODEL_URL), faceapi.nets.tinyFaceDetector.loadFromUri(FACE_API_MODEL_URL), faceapi.nets.faceLandmark68TinyNet.loadFromUri(FACE_API_MODEL_URL), faceapi.nets.faceExpressionNet.loadFromUri(FACE_API_MODEL_URL), faceapi.nets.ageGenderNet.loadFromUri(FACE_API_MODEL_URL), ]); coreReady = true; setPill(els.aiDot, els.aiStatus, "AI ready", "ready"); els.startBtn.disabled = false; renderBars({}); loadPrecisionModels(); } catch (err) { console.error(err); setPill(els.aiDot, els.aiStatus, "AI model loading failed", "error"); showToast("Model loading failed. Refresh the Space and check browser network access."); } } async function loadOpenCvFerModel() { await waitForGlobal("ort", 16000); ort.env.wasm.wasmPaths = "https://cdn.jsdelivr.net/npm/onnxruntime-web@1.21.0/dist/"; ort.env.wasm.numThreads = Math.max(1, Math.min(4, navigator.hardwareConcurrency || 2)); precision.ferSession = await ort.InferenceSession.create(OPENCV_FER_MODEL_URL, { executionProviders: ["wasm"] }); updatePrecisionStatus(); } async function loadTransformerEmotionModel() { if (xenovaEnv) { xenovaEnv.allowLocalModels = false; xenovaEnv.useBrowserCache = true; xenovaEnv.backends ??= {}; xenovaEnv.backends.onnx ??= {}; xenovaEnv.backends.onnx.wasm ??= {}; xenovaEnv.backends.onnx.wasm.numThreads = Math.max(1, Math.min(4, navigator.hardwareConcurrency || 2)); } precision.emotionPipe = await xenovaPipeline("image-classification", EMOTION_TRANSFORMER_MODEL_ID, { quantized: true }); updatePrecisionStatus(); } async function loadAgeGenderModel() { if (hfEnv) { hfEnv.allowLocalModels = false; hfEnv.useBrowserCache = true; hfEnv.backends ??= {}; hfEnv.backends.onnx ??= {}; hfEnv.backends.onnx.wasm ??= {}; hfEnv.backends.onnx.wasm.numThreads = Math.max(1, Math.min(4, navigator.hardwareConcurrency || 2)); } const opts = navigator.gpu ? { device: "webgpu", dtype: "q8" } : { device: "wasm", dtype: "q8" }; try { precision.ageModel = await AutoModel.from_pretrained(AGE_GENDER_MODEL_ID, opts); } catch (err) { console.warn("Age/gender preferred backend failed; retrying wasm", err); precision.ageModel = await AutoModel.from_pretrained(AGE_GENDER_MODEL_ID, { device: "wasm", dtype: "q8" }); } precision.ageProcessor = await AutoProcessor.from_pretrained(AGE_GENDER_MODEL_ID); updatePrecisionStatus(); } async function loadPrecisionModels() { if (precision.loading) return; precision.loading = true; updatePrecisionStatus(); const tasks = [ loadOpenCvFerModel().catch(err => console.warn("OpenCV FER model unavailable", err)), loadTransformerEmotionModel().catch(err => console.warn("Transformer emotion model unavailable", err)), loadAgeGenderModel().catch(err => console.warn("Age/gender transformer unavailable", err)), ]; await Promise.allSettled(tasks); precision.loading = false; updatePrecisionStatus(); if (precision.ferSession || precision.emotionPipe || precision.ageModel) showToast("Precision models are ready."); } function averagePoint(points) { const list = (points || []).map(p => ({ x: p.x, y: p.y })).filter(p => Number.isFinite(p.x) && Number.isFinite(p.y)); if (!list.length) return null; return { x: list.reduce((a, p) => a + p.x, 0) / list.length, y: list.reduce((a, p) => a + p.y, 0) / list.length, }; } function getLandmarkPoints(det) { if (!det.landmarks) return null; const leftEye = averagePoint(det.landmarks.getLeftEye?.() || []); const rightEye = averagePoint(det.landmarks.getRightEye?.() || []); const noseList = det.landmarks.getNose?.() || []; const mouthList = det.landmarks.getMouth?.() || []; const noseTip = noseList.length ? noseList[Math.min(3, noseList.length - 1)] : null; if (!leftEye || !rightEye || !noseTip || !mouthList.length) return null; const eyes = [leftEye, rightEye].sort((a, b) => a.x - b.x); const mouths = mouthList.slice().sort((a, b) => a.x - b.x); const leftMouth = mouths[0]; const rightMouth = mouths[mouths.length - 1]; return [eyes[0], eyes[1], noseTip, leftMouth, rightMouth]; } function solveLinearSystem(A, b) { const n = b.length; const M = A.map((row, i) => row.concat([b[i]])); for (let col = 0; col < n; col += 1) { let pivot = col; for (let r = col + 1; r < n; r += 1) if (Math.abs(M[r][col]) > Math.abs(M[pivot][col])) pivot = r; if (Math.abs(M[pivot][col]) < 1e-8) return null; [M[col], M[pivot]] = [M[pivot], M[col]]; const div = M[col][col]; for (let c = col; c <= n; c += 1) M[col][c] /= div; for (let r = 0; r < n; r += 1) { if (r === col) continue; const factor = M[r][col]; for (let c = col; c <= n; c += 1) M[r][c] -= factor * M[col][c]; } } return M.map(row => row[n]); } function estimateAffine(srcPts, dstPts) { const rows = []; const vals = []; for (let i = 0; i < srcPts.length; i += 1) { const sx = srcPts[i].x; const sy = srcPts[i].y; const dx = dstPts[i].x; const dy = dstPts[i].y; rows.push([sx, sy, 1, 0, 0, 0]); vals.push(dx); rows.push([0, 0, 0, sx, sy, 1]); vals.push(dy); } const ATA = Array.from({ length: 6 }, () => Array(6).fill(0)); const ATb = Array(6).fill(0); for (let r = 0; r < rows.length; r += 1) { for (let i = 0; i < 6; i += 1) { ATb[i] += rows[r][i] * vals[r]; for (let j = 0; j < 6; j += 1) ATA[i][j] += rows[r][i] * rows[r][j]; } } return solveLinearSystem(ATA, ATb); } function makeAlignedFaceCanvas(det, targetSize = 112) { const points = getLandmarkPoints(det); if (!points) return cropFaceCanvas(det, targetSize, 0.25, "contrast(1.08)"); const dst = [ { x: 38.2946, y: 51.6963 }, { x: 73.5318, y: 51.5014 }, { x: 56.0252, y: 71.7366 }, { x: 41.5493, y: 92.3655 }, { x: 70.7299, y: 92.2041 }, ]; const sol = estimateAffine(points, dst); if (!sol) return cropFaceCanvas(det, targetSize, 0.25, "contrast(1.08)"); const canvas = document.createElement("canvas"); canvas.width = targetSize; canvas.height = targetSize; const c = canvas.getContext("2d", { willReadFrequently: true }); c.fillStyle = "#000"; c.fillRect(0, 0, targetSize, targetSize); c.setTransform(sol[0], sol[3], sol[1], sol[4], sol[2], sol[5]); c.filter = "contrast(1.08) saturate(0.95)"; c.drawImage(els.video, 0, 0); c.setTransform(1, 0, 0, 1, 0, 0); c.filter = "none"; return canvas; } function canvasToNchwTensor(canvas) { const c = canvas.getContext("2d", { willReadFrequently: true }); const { data } = c.getImageData(0, 0, canvas.width, canvas.height); const size = canvas.width * canvas.height; const out = new Float32Array(3 * size); for (let i = 0; i < size; i += 1) { const j = i * 4; out[i] = (data[j] / 255 - 0.5) / 0.5; out[size + i] = (data[j + 1] / 255 - 0.5) / 0.5; out[size * 2 + i] = (data[j + 2] / 255 - 0.5) / 0.5; } return out; } async function runOpenCvFer(det) { if (!precision.ferSession || !window.ort) return null; const canvas = makeAlignedFaceCanvas(det, 112); const input = canvasToNchwTensor(canvas); const tensor = new ort.Tensor("float32", input, [1, 3, 112, 112]); const feeds = {}; feeds[precision.ferSession.inputNames[0]] = tensor; const results = await precision.ferSession.run(feeds); const outputName = precision.ferSession.outputNames[0]; const logits = Array.from(results[outputName].data).slice(0, 7); const probs = softmax(logits); const scores = blankScores(); for (let i = 0; i < probs.length && i < ferLabels.length; i += 1) { const label = ferLabels[i]; scores[label] = Math.max(scores[label], probs[i]); } return normalizeScores(scores); } async function runTransformerEmotion(det) { if (!precision.emotionPipe) return null; const variants = [ cropFaceCanvas(det, 288, 0.14, "contrast(1.10) saturate(0.95)", false), cropFaceCanvas(det, 288, 0.30, "contrast(1.16) saturate(0.92)", false), cropFaceCanvas(det, 288, 0.08, "contrast(1.18) brightness(1.02)", false), cropFaceCanvas(det, 288, 0.20, "contrast(1.12) saturate(0.94)", true), ]; const aggregate = blankScores(); const urls = []; let count = 0; try { for (const variant of variants) { const url = await canvasToBlobUrl(variant); urls.push(url); let output; try { output = await precision.emotionPipe(url, { topK: 7 }); } catch (_) { output = await precision.emotionPipe(url); } const scores = normalizeExternalEmotion(output); for (const label of emotionLabels) aggregate[label] += scores[label] || 0; count += 1; } } finally { for (const url of urls) URL.revokeObjectURL(url); } if (!count) return null; for (const label of emotionLabels) aggregate[label] /= count; return normalizeScores(aggregate); } async function tensorValues(tensor) { if (!tensor) return []; if (tensor.tolist) { const listed = tensor.tolist(); return Array.isArray(listed) ? listed.flat(Infinity) : []; } if (tensor.data) return Array.from(tensor.data); return []; } async function runAgeGender(det) { if (!precision.ageModel || !precision.ageProcessor || !load_image) return; const canvas = cropFaceCanvas(det, 384, 0.22, "contrast(1.07) saturate(0.96)"); let url = null; try { url = await canvasToBlobUrl(canvas); const image = await load_image(url); const inputs = await precision.ageProcessor(image); const output = await precision.ageModel(inputs); const values = await tensorValues(output.logits || output[0]); if (values.length < 2) return; const rawAge = Number(values[0]); const rawGender = Number(values[1]); if (Number.isFinite(rawAge)) { const age = Math.max(0, Math.min(100, Math.round(rawAge))); precision.age = age; precision.ageAt = performance.now(); pushAgeSample(age, "precision model", 5); } if (Number.isFinite(rawGender)) { const pFemale = rawGender >= 0 && rawGender <= 1 ? rawGender : sigmoid(rawGender); precision.gender = pFemale >= 0.5 ? "Female" : "Male"; precision.genderScore = Math.max(pFemale, 1 - pFemale); precision.genderAt = performance.now(); } } finally { if (url) URL.revokeObjectURL(url); } } async function maybeRunPrecision(primary) { const now = performance.now(); if (!primary || precision.busy || (!precision.ferSession && !precision.emotionPipe && !precision.ageModel)) return; if (now - precision.lastRun < detectorConfig.precisionIntervalMs) return; precision.busy = true; precision.lastRun = now; const started = performance.now(); try { if (precision.ferSession) { const fer = await runOpenCvFer(primary); if (fer) { precision.ferScores = fer; precision.ferAt = performance.now(); } } if (precision.emotionPipe) { const transformer = await runTransformerEmotion(primary); if (transformer) { precision.transformerScores = transformer; precision.transformerAt = performance.now(); } } if (precision.ageModel && precision.ageProcessor) await runAgeGender(primary); } catch (err) { console.warn("Precision inference failed", err); } finally { precision.latency = performance.now() - started; precision.busy = false; } } function stopStream() { if (stream) for (const track of stream.getTracks()) track.stop(); stream = null; running = false; detecting = false; els.video.srcObject = null; ctx.clearRect(0, 0, els.overlay.width, els.overlay.height); els.emptyState.style.display = "grid"; els.startBtn.disabled = !coreReady; els.switchBtn.disabled = true; els.stopBtn.disabled = true; els.cameraTag.textContent = "Camera: stopped"; resetDetails(); } async function startCamera(facing = currentFacing) { if (!coreReady) { showToast("Please wait until the AI models finish loading."); return; } stopStream(); currentFacing = facing; els.cameraTag.textContent = `Camera: ${currentFacing === "user" ? "front" : "rear"} requested`; const constraints = { audio: false, video: { facingMode: { ideal: currentFacing }, width: { ideal: 1280 }, height: { ideal: 720 }, } }; try { stream = await navigator.mediaDevices.getUserMedia(constraints); } catch (firstErr) { console.warn("Preferred camera failed, falling back", firstErr); try { stream = await navigator.mediaDevices.getUserMedia({ audio: false, video: true }); } catch (err) { console.error(err); showToast("Camera permission failed. Allow camera access in your browser settings."); els.cameraTag.textContent = "Camera: permission needed"; return; } } els.video.srcObject = stream; await els.video.play(); fitCanvas(); running = true; els.emptyState.style.display = "none"; els.startBtn.disabled = true; els.switchBtn.disabled = false; els.stopBtn.disabled = false; els.cameraTag.textContent = `Camera: ${currentFacing === "user" ? "front" : "rear"}`; lastLoopTime = performance.now(); fpsSmooth = 0; detectLoop(); } async function detectFacesAccurate() { try { const options = new faceapi.SsdMobilenetv1Options({ minConfidence: 0.42 }); return await faceapi .detectAllFaces(els.video, options) .withFaceLandmarks(true) .withFaceExpressions() .withAgeAndGender(); } catch (err) { console.warn("SSD detector failed; using tiny detector", err); const tiny = new faceapi.TinyFaceDetectorOptions({ inputSize: detectorConfig.inputSize, scoreThreshold: detectorConfig.scoreThreshold }); return await faceapi .detectAllFaces(els.video, tiny) .withFaceLandmarks(true) .withFaceExpressions() .withAgeAndGender(); } } async function detectLoop() { if (!running || !coreReady || detecting) return; detecting = true; const started = performance.now(); try { fitCanvas(); const raw = await detectFacesAccurate(); const displaySize = { width: els.overlay.width, height: els.overlay.height }; const resized = faceapi.resizeResults(raw, displaySize); const elapsed = performance.now() - started; drawDetections(resized); updateDetails(resized, elapsed); setTimeout(() => { detecting = false; requestAnimationFrame(detectLoop); }, detectorConfig.intervalMs); } catch (err) { console.error(err); detecting = false; showToast("Live analysis hit an error. Restarting camera may fix it."); setTimeout(() => requestAnimationFrame(detectLoop), 250); } } els.startBtn.addEventListener("click", () => startCamera(currentFacing)); els.switchBtn.addEventListener("click", () => { currentFacing = currentFacing === "user" ? "environment" : "user"; startCamera(currentFacing); }); els.stopBtn.addEventListener("click", stopStream); els.video.addEventListener("loadedmetadata", fitCanvas); window.addEventListener("resize", fitCanvas); resetDetails(); renderBars({}); setPill(els.precisionDot, els.precisionStatus, "Precision models loading automatically", "loading"); loadCoreModels(); </script> </body> </html>
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1, viewport-fit=cover" />
<meta name="theme-color" content="#050816" />
<title>SentAI</title>
<script defer src="https://cdn.jsdelivr.net/npm/face-api.js@0.22.2/dist/face-api.min.js"></script>
<script defer src="https://cdn.jsdelivr.net/npm/onnxruntime-web@1.21.0/dist/ort.min.js"></script>
<style>
:root {
--bg-a: #050816;
--bg-b: #0d1b2f;
--card: rgba(15, 23, 42, 0.76);
--card-strong: rgba(15, 23, 42, 0.94);
--stroke: rgba(148, 163, 184, 0.24);
--text: #f8fafc;
--muted: #a7b4c8;
--soft: rgba(255,255,255,0.08);
--accent: #22d3ee;
--accent-2: #a78bfa;
--good: #2dd4bf;
--warn: #fbbf24;
--bad: #fb7185;
--shadow: 0 24px 90px rgba(0,0,0,0.42);
--radius: 28px;
font-family: Inter, ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif;
}
* { box-sizing: border-box; }
body {
margin: 0;
min-height: 100vh;
color: var(--text);
background:
radial-gradient(circle at 12% 10%, rgba(34, 211, 238, 0.26), transparent 34%),
radial-gradient(circle at 92% 2%, rgba(167, 139, 250, 0.23), transparent 34%),
linear-gradient(135deg, var(--bg-a), var(--bg-b) 52%, #112236);
overflow-x: hidden;
}
body::before {
content: "";
position: fixed;
inset: 0;
pointer-events: none;
background-image:
linear-gradient(rgba(255,255,255,0.035) 1px, transparent 1px),
linear-gradient(90deg, rgba(255,255,255,0.035) 1px, transparent 1px);
background-size: 44px 44px;
mask-image: radial-gradient(circle at 50% 20%, black 0%, transparent 75%);
}
.app-shell {
width: min(1460px, calc(100% - 32px));
margin: 0 auto;
padding: 34px 0 42px;
position: relative;
z-index: 1;
}
header.hero {
display: grid;
grid-template-columns: 1fr auto;
align-items: center;
gap: 22px;
margin-bottom: 26px;
}
.brand h1 {
margin: 0;
font-size: clamp(4.2rem, 9vw, 8rem);
line-height: 0.88;
letter-spacing: -0.08em;
font-weight: 950;
text-shadow: 0 24px 70px rgba(34, 211, 238, 0.10);
}
.brand p {
margin: 18px 0 0;
max-width: 960px;
color: var(--muted);
font-size: clamp(1rem, 1.7vw, 1.22rem);
line-height: 1.55;
}
.status-stack {
display: grid;
gap: 10px;
justify-items: end;
}
.status-pill {
display: inline-flex;
align-items: center;
gap: 10px;
border: 1px solid var(--stroke);
background: rgba(15, 23, 42, 0.64);
backdrop-filter: blur(18px);