[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8,"_9":10},"actionData","errors","root",{"_153":154,"_155":156},"routes/_app.x.$handle.$name",{"_11":12,"_13":14,"_15":16,"_18":19,"_17":-7,"_116":-5,"_20":117,"_22":118,"_119":120,"_121":122,"_25":123,"_24":-5,"_124":-5,"_125":-5,"_126":53,"_127":128,"_129":44,"_130":-5,"_131":-5,"_132":133},"routes/_app.x.$handle.$name.branch.$branch.code.$",{"_11":12,"_13":14,"_15":16,"_17":-7,"_18":19,"_20":21,"_22":23,"_24":-7,"_25":26,"_27":-5,"_28":-5,"_29":-5,"_30":-5,"_31":-5},"handle","mariamshoaib","name","esrever-vid","branchName","main","version","path","","project",{"_33":61,"_13":14,"_45":62,"_63":-5,"_64":53,"_65":66,"_67":66,"_68":19,"_69":-5,"_70":71,"_72":-5,"_73":53,"_74":-5,"_75":-5,"_76":77,"_78":79,"_80":-5,"_81":-5,"_82":-5,"_83":-5,"_84":85,"_86":85,"_87":88,"_89":88,"_90":88,"_91":85,"_92":85,"_93":94,"_95":96,"_97":98,"_99":53},"tree",[56],"file","contents",[32],"upstream","pull","ahead","behind","envVars",{"_33":34,"_13":35,"_36":37,"_38":-5,"_39":40,"_41":42,"_43":44,"_45":46,"_17":47,"_48":49,"_50":51,"_52":53,"_54":-5,"_18":55},"id","9a45edf2-c0b3-469a-91b0-27856a4f3a1a","main.ts","fileId","f47a348c-7a3a-11f1-bea7-1607ee4eb77e","parentId","branchId","fde9bf96-e2f6-4887-8bb9-9b7e71662fcc","branchVersion",4,"latest",true,"createdAt",["D",1783453154281],3,"fileType","http","content","import { GoogleGenerativeAI } from \"npm:@google/generative-ai\";\n\nconst GEMINI_API_KEY = Deno.env.get(\"GEMINI_API_KEY\");\n\ninterface VideoEsreverResult {\n verdict: \"synthetic\" | \"authentic\" | \"tampered\" | \"inconclusive\";\n confidence: number;\n signals: Signal[];\n container_metadata: Record;\n tamper_flags: TamperFlag[];\n temporal_analysis: TemporalAnalysis;\n model_signature: ModelSignature;\n reconstructed_prompt: string;\n processing_time_ms: number;\n}\n\ninterface Signal {\n type: string;\n weight: \"high\" | \"medium\" | \"low\";\n description: string;\n}\n\ninterface TamperFlag {\n field: string;\n issue: string;\n severity: \"critical\" | \"moderate\" | \"minor\";\n}\n\ninterface TemporalAnalysis {\n temporal_consistency: boolean;\n flicker_detected: boolean;\n physics_anomalies: string[];\n texture_drift: boolean;\n motion_blur_natural: boolean;\n confidence: number;\n notes: string;\n}\n\ninterface ModelSignature {\n detected_model: string | null;\n confidence: number;\n fingerprints: string[];\n}\n\nfunction parseMP4Metadata(buffer: Uint8Array): Record {\n const metadata: Record = {};\n\n const readUint32 = (offset: number): number =>\n (buffer[offset] << 24) | (buffer[offset + 1] << 16) |\n (buffer[offset + 2] << 8) | buffer[offset + 3];\n\n const readString = (offset: number, length: number): string => {\n let str = \"\";\n for (let i = 0; i < length && offset + i < buffer.length; i++) {\n const c = buffer[offset + i];\n if (c >= 32 && c < 127) str += String.fromCharCode(c);\n }\n return str.trim();\n };\n\n // MP4 epoch starts at Jan 1, 1904\n const mp4EpochOffset = 2082844800;\n const toDate = (secs: number): string | null => {\n if (secs === 0) return null;\n return new Date((secs - mp4EpochOffset) * 1000).toISOString();\n };\n\n let offset = 0;\n const boxes: string[] = [];\n\n while (offset < buffer.length - 8) {\n const size = readUint32(offset);\n if (size < 8 || offset + size > buffer.length) break;\n\n const type = readString(offset + 4, 4);\n boxes.push(type);\n\n if (type === \"ftyp\") {\n metadata.major_brand = readString(offset + 8, 4);\n metadata.minor_version = readUint32(offset + 12);\n const compatBrands: string[] = [];\n for (let i = offset + 16; i < offset + size; i += 4) {\n const brand = readString(i, 4);\n if (brand) compatBrands.push(brand);\n }\n metadata.compatible_brands = compatBrands;\n }\n\n if (type === \"mvhd\") {\n const version = buffer[offset + 8];\n if (version === 0) {\n metadata.creation_time = toDate(readUint32(offset + 12));\n metadata.modification_time = toDate(readUint32(offset + 16));\n const timescale = readUint32(offset + 20);\n const duration = readUint32(offset + 24);\n metadata.duration_seconds = timescale > 0 ? duration / timescale : null;\n }\n }\n\n if (type === \"udta\") {\n // Try to extract encoder/tool string from udta\n const udtaContent = readString(offset + 8, Math.min(size - 8, 512));\n if (udtaContent) metadata.user_data = udtaContent;\n }\n\n offset += size;\n }\n\n metadata.box_sequence = boxes;\n\n // Check for suspicious encoder strings\n const knownAIEncoders = [\n \"runway\",\n \"sora\",\n \"kling\",\n \"pika\",\n \"luma\",\n \"hailuo\",\n \"veo\",\n \"dream machine\",\n \"minimax\",\n \"wan\",\n \"cogvideo\",\n ];\n const userDataStr = ((metadata.user_data as string) || \"\").toLowerCase();\n const majorBrand = ((metadata.major_brand as string) || \"\").toLowerCase();\n\n for (const encoder of knownAIEncoders) {\n if (userDataStr.includes(encoder)) {\n metadata.detected_encoder = encoder;\n break;\n }\n }\n\n // Flag missing mvhd (always present in real camera footage)\n if (!metadata.creation_time) {\n metadata.missing_creation_time = true;\n }\n\n return metadata;\n}\n\nfunction detectContainerTampering(\n metadata: Record,\n): TamperFlag[] {\n const flags: TamperFlag[] = [];\n\n if (metadata.detected_encoder) {\n flags.push({\n field: \"udta/encoder\",\n issue:\n `AI video tool detected in container metadata: \"${metadata.detected_encoder}\"`,\n severity: \"critical\",\n });\n }\n\n if (metadata.missing_creation_time) {\n flags.push({\n field: \"mvhd/creation_time\",\n issue:\n \"No creation timestamp in container — absent in AI-generated video, present in all real camera footage\",\n severity: \"critical\",\n });\n }\n\n const creation = metadata.creation_time as string | null;\n const modification = metadata.modification_time as string | null;\n if (creation && modification) {\n if (new Date(modification) < new Date(creation)) {\n flags.push({\n field: \"mvhd/timestamps\",\n issue:\n \"Modification time precedes creation time — impossible for unaltered file\",\n severity: \"critical\",\n });\n }\n }\n\n const brand = (metadata.major_brand as string) || \"\";\n const aiTypicalBrands = [\"iso5\", \"isom\"];\n if (aiTypicalBrands.includes(brand.toLowerCase())) {\n // isom/iso5 alone isn't conclusive but combined with missing timestamps is\n if (metadata.missing_creation_time) {\n flags.push({\n field: \"ftyp/major_brand\",\n issue:\n `Generic brand \"${brand}\" with no camera metadata — consistent with AI generation pipeline output`,\n severity: \"moderate\",\n });\n }\n }\n\n const boxes = (metadata.box_sequence as string[]) || [];\n if (!boxes.includes(\"moov\")) {\n flags.push({\n field: \"container/structure\",\n issue: \"Missing moov box — malformed or heavily re-encoded file\",\n severity: \"critical\",\n });\n }\n\n return flags;\n}\n\nasync function runGeminiVideoAnalysis(\n videoBuffer: Uint8Array,\n mimeType: string,\n): Promise<\n { temporal: TemporalAnalysis; modelSig: ModelSignature; prompt: string }\n> {\n const genAI = new GoogleGenerativeAI(GEMINI_API_KEY!);\n const model = genAI.getGenerativeModel({ model: \"gemini-2.5-flash\" });\n\n const base64Video = btoa(\n Array.from(videoBuffer).map((b) => String.fromCharCode(b)).join(\"\"),\n );\n\n const forensicsPrompt =\n `You are a forensic video analyst specializing in detecting AI-generated synthetic video.\n\nAnalyze this video across these dimensions:\n\nTEMPORAL CONSISTENCY:\n- Does lighting, shadow direction, and color temperature remain consistent across all frames?\n- Are there flickering artifacts or sudden texture changes between frames?\n- Does the background remain stable or show warping/drift?\n\nPHYSICS:\n- Is motion blur natural and direction-consistent with movement?\n- Do secondary elements (hair, clothing, water, smoke, leaves) move with natural physics?\n- Are object interactions (hands touching surfaces, feet on ground) physically coherent?\n\nAI MODEL FINGERPRINTS — check specifically for:\n- Runway Gen-4: characteristic motion smearing on edges, over-smooth skin, temporal texture shimmer on hair\n- Kling: slightly unnatural human gait, face morphing between cuts, overly cinematic color grade\n- Sora: almost-right physics that breaks on secondary motion, dreamlike background drift\n- Pika: lower resolution artifacts, choppy motion on fast movement, obvious looping\n- Hailuo/MiniMax: face consistency issues across angles, unnatural eye movement\n- Veo 2: very high quality but subtle lighting inconsistency at subject boundaries\n- Luma Dream Machine: strong on objects/products, weaker on human faces, characteristic soft focus\n\nAUDIO-VISUAL SYNC (if audio present):\n- Does lip movement match audio precisely?\n- Is ambient audio consistent with environment shown?\n\nReturn ONLY this JSON, no markdown:\n{\n \"temporal_analysis\": {\n \"temporal_consistency\": boolean,\n \"flicker_detected\": boolean,\n \"physics_anomalies\": [\"list of specific anomalies\"],\n \"texture_drift\": boolean,\n \"motion_blur_natural\": boolean,\n \"confidence\": number between 0 and 1,\n \"notes\": \"precise technical summary\"\n },\n \"model_signature\": {\n \"detected_model\": \"model name or null\",\n \"confidence\": number between 0 and 1,\n \"fingerprints\": [\"specific visual fingerprints observed\"]\n },\n \"reconstructed_prompt\": \"the most likely text prompt used to generate this video, written as if prompting an AI video model, or null if video appears authentic\"\n}`;\n\n const result = await model.generateContent([\n {\n inlineData: {\n mimeType: mimeType,\n data: base64Video,\n },\n },\n forensicsPrompt,\n ]);\n\n const raw = result.response.text();\n const clean = raw.replace(/```json|```/g, \"\").trim();\n\n try {\n const parsed = JSON.parse(clean);\n return {\n temporal: parsed.temporal_analysis,\n modelSig: parsed.model_signature,\n prompt: parsed.reconstructed_prompt ||\n \"Prompt reconstruction unavailable\",\n };\n } catch {\n return {\n temporal: {\n temporal_consistency: true,\n flicker_detected: false,\n physics_anomalies: [],\n texture_drift: false,\n motion_blur_natural: true,\n confidence: 0,\n notes: \"Gemini analysis failed to parse — treat as inconclusive\",\n },\n modelSig: {\n detected_model: null,\n confidence: 0,\n fingerprints: [],\n },\n prompt: \"Analysis failed\",\n };\n }\n}\n\nfunction assembleVerdict(\n tamperFlags: TamperFlag[],\n temporal: TemporalAnalysis,\n modelSig: ModelSignature,\n): {\n verdict: VideoEsreverResult[\"verdict\"];\n confidence: number;\n signals: Signal[];\n} {\n const signals: Signal[] = [];\n let syntheticScore = 0;\n let tamperScore = 0;\n\n for (const flag of tamperFlags) {\n if (flag.severity === \"critical\") {\n tamperScore += 0.35;\n signals.push({\n type: \"container\",\n weight: \"high\",\n description: flag.issue,\n });\n } else if (flag.severity === \"moderate\") {\n tamperScore += 0.20;\n signals.push({\n type: \"container\",\n weight: \"medium\",\n description: flag.issue,\n });\n } else {\n tamperScore += 0.08;\n signals.push({\n type: \"container\",\n weight: \"low\",\n description: flag.issue,\n });\n }\n }\n\n if (temporal.confidence > 0) {\n if (temporal.flicker_detected) {\n syntheticScore += 0.25;\n signals.push({\n type: \"temporal\",\n weight: \"high\",\n description: \"Inter-frame flickering detected\",\n });\n }\n if (temporal.texture_drift) {\n syntheticScore += 0.20;\n signals.push({\n type: \"temporal\",\n weight: \"high\",\n description: \"Texture drift between frames\",\n });\n }\n if (!temporal.motion_blur_natural) {\n syntheticScore += 0.15;\n signals.push({\n type: \"temporal\",\n weight: \"medium\",\n description: \"Unnatural motion blur pattern\",\n });\n }\n if (temporal.physics_anomalies.length > 0) {\n syntheticScore += 0.15 * Math.min(temporal.physics_anomalies.length, 3) /\n 3;\n signals.push({\n type: \"physics\",\n weight: \"medium\",\n description: `Physics anomalies: ${\n temporal.physics_anomalies.slice(0, 3).join(\", \")\n }`,\n });\n }\n if (!temporal.temporal_consistency) {\n syntheticScore += 0.20;\n signals.push({\n type: \"temporal\",\n weight: \"high\",\n description: \"Temporal inconsistency across frames\",\n });\n }\n }\n\n if (modelSig.detected_model && modelSig.confidence > 0.5) {\n syntheticScore += modelSig.confidence * 0.30;\n signals.push({\n type: \"model_signature\",\n weight: \"high\",\n description: `Probable generation model: ${modelSig.detected_model} (${\n Math.round(modelSig.confidence * 100)\n }% confidence)`,\n });\n if (modelSig.fingerprints.length > 0) {\n signals.push({\n type: \"model_fingerprint\",\n weight: \"medium\",\n description: `Fingerprints: ${\n modelSig.fingerprints.slice(0, 3).join(\", \")\n }`,\n });\n }\n }\n\n syntheticScore = Math.min(syntheticScore, 0.98);\n tamperScore = Math.min(tamperScore, 0.98);\n\n let verdict: VideoEsreverResult[\"verdict\"] = \"inconclusive\";\n let confidence = 0;\n\n if (syntheticScore >= 0.65) {\n verdict = \"synthetic\";\n confidence = syntheticScore;\n } else if (tamperScore >= 0.5 && syntheticScore < 0.5) {\n verdict = \"tampered\";\n confidence = tamperScore;\n } else if (\n syntheticScore < 0.25 && tamperScore < 0.25 && signals.length === 0\n ) {\n verdict = \"authentic\";\n confidence = 1 - Math.max(syntheticScore, tamperScore);\n } else {\n verdict = \"inconclusive\";\n confidence = 0.5;\n }\n\n return { verdict, confidence: parseFloat(confidence.toFixed(2)), signals };\n}\n\nexport default async function handler(req: Request): Promise {\n const headers = {\n \"Content-Type\": \"application/json\",\n \"Access-Control-Allow-Origin\": \"*\",\n \"Access-Control-Allow-Methods\": \"POST, OPTIONS\",\n \"Access-Control-Allow-Headers\": \"Content-Type\",\n };\n\n if (req.method === \"OPTIONS\") {\n return new Response(null, { status: 204, headers });\n }\n if (req.method !== \"POST\") {\n return new Response(JSON.stringify({ error: \"POST only\" }), {\n status: 405,\n headers,\n });\n }\n\n const startTime = Date.now();\n\n try {\n const body = await req.json();\n let videoBuffer: Uint8Array;\n let mimeType = \"video/mp4\";\n\n if (body.url) {\n const res = await fetch(body.url);\n if (!res.ok) throw new Error(\"Failed to fetch video from URL\");\n mimeType = res.headers.get(\"content-type\")?.split(\";\")[0] || \"video/mp4\";\n videoBuffer = new Uint8Array(await res.arrayBuffer());\n } else if (body.base64) {\n const binary = atob(body.base64.replace(/^data:video\\/\\w+;base64,/, \"\"));\n videoBuffer = new Uint8Array(binary.length);\n for (let i = 0; i < binary.length; i++) {\n videoBuffer[i] = binary.charCodeAt(i);\n }\n if (body.mimeType) mimeType = body.mimeType;\n } else {\n return new Response(\n JSON.stringify({ error: \"Provide 'url' or 'base64' in request body\" }),\n { status: 400, headers },\n );\n }\n\n const containerMetadata = parseMP4Metadata(videoBuffer);\n const tamperFlags = detectContainerTampering(containerMetadata);\n const { temporal, modelSig, prompt } = await runGeminiVideoAnalysis(\n videoBuffer,\n mimeType,\n );\n const { verdict, confidence, signals } = assembleVerdict(\n tamperFlags,\n temporal,\n modelSig,\n );\n\n const result: VideoEsreverResult = {\n verdict,\n confidence,\n signals,\n container_metadata: containerMetadata,\n tamper_flags: tamperFlags,\n temporal_analysis: temporal,\n model_signature: modelSig,\n reconstructed_prompt: prompt,\n processing_time_ms: Date.now() - startTime,\n };\n\n return new Response(JSON.stringify(result, null, 2), {\n status: 200,\n headers,\n });\n } catch (err) {\n return new Response(\n JSON.stringify({\n error: \"Analysis failed\",\n detail: (err as Error).message,\n }),\n { status: 500, headers },\n );\n }\n}","deleted",false,"label",[35],{"_33":34,"_13":35,"_36":37,"_38":-5,"_39":40,"_41":42,"_43":44,"_45":57,"_17":47,"_48":49,"_50":51,"_52":53,"_54":-5,"_58":59,"_18":60},["D",1783453154281],"children",[],[35],"4c2c7ca7-4fdb-46ac-928f-2b6fd4c97540",["D",1783452905333],"deletedAt","legacyEnvVars","privacy","public","httpPrivacy","description","imageUrl","userId","9586015e-2e16-427e-a2a9-932991d3b871","pinnedFileId","pinPreview","pinMigrated","screenshotVersion","likeCount",0,"orgId","019ecd1b-ddbf-74fb-b5e7-e3b7d3397e81","tursoName","tursoHostname","tursoMeasuredAt","pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_11":12,"_115":111},"org",{"_11":12,"_108":109,"_110":111,"_70":71,"_112":113,"_114":53},"branch",{"_17":42,"_13":16,"_33":40,"_100":61,"_101":-5,"_102":-5,"_103":-5,"_104":-5,"_105":-5,"_106":14,"_107":12},"isOwned","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","displayName","Mariam Shoaib","avatar","https://lh3.googleusercontent.com/a/ACg8ocIonqHMwFmivqQc9nfb3Pt_LjrC6Wsgw3cUq1NkQenzRYXVA7zM=s96-c","tier","free","canRestrictVals","profileImageUrl","adoptOrgId",{"_33":61,"_13":14,"_45":149,"_63":-5,"_64":53,"_65":66,"_67":66,"_68":19,"_69":-5,"_70":71,"_72":-5,"_73":53,"_74":-5,"_75":-5,"_76":77,"_78":79,"_80":-5,"_81":-5,"_82":-5,"_83":-5,"_84":85,"_86":85,"_87":88,"_89":88,"_90":88,"_91":85,"_92":85,"_93":150,"_95":151,"_97":152,"_99":53},[146],"files",[146],"counts",{"_136":77,"_137":138,"_139":77,"_31":140,"_141":142},[134],"upstreamProject","upstreamBranch","emptyEnvVars","flags",[],"hasHttpFiles","grants","orgs","customDomains",{},{"_33":34,"_13":35,"_36":37,"_38":-5,"_39":40,"_41":42,"_43":44,"_45":46,"_17":47,"_48":49,"_50":51,"_52":53,"_54":-5,"_18":135},[35],"pulls","branches",1,"forks",{"_143":138,"_145":77},"connections",{"_143":77,"_144":77},"count","unfilledCount","emptyCount",{"_33":34,"_13":35,"_36":37,"_38":-5,"_39":40,"_41":42,"_43":44,"_45":57,"_17":47,"_48":49,"_50":51,"_52":53,"_54":-5,"_58":147,"_18":148},[],[35],["D",1783452905333],{"_11":12,"_115":111},{"_11":12,"_108":109,"_110":111,"_70":71,"_112":113,"_114":53},{"_17":42,"_13":16,"_33":40,"_100":61,"_101":-5,"_102":-5,"_103":-5,"_104":-5,"_105":-5,"_106":14,"_107":12},"ctx",{"_93":-5,"_131":186,"_187":188,"_189":190,"_191":53,"_192":53,"_193":53,"_194":44,"_195":-5,"_196":-7,"_197":-7,"_198":77,"_199":53,"_200":53,"_201":-5,"_202":53,"_203":53,"_204":53,"_205":-5},"ENV",{"_157":158,"_159":160,"_161":16,"_162":163,"_164":165,"_166":167,"_168":169,"_170":171,"_172":173,"_174":175,"_176":177,"_178":179,"_180":181,"_182":183,"_184":185},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","fdb67791413f4386b4dbf18f354029a617f6160f","RENDER_GIT_BRANCH","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","CLOUDFLARE_TURNSTILE_SITE_KEY","0x4AAAAAACtaIZJH7s9rBMZ6","GOOGLE_CLIENT_ID","875989696899-jtsqkhlbdv5jrlsk8631p7ipjg3c6o62.apps.googleusercontent.com","STRIPE_PUBLISHABLE_KEY","pk_live_51N2zvWEpZTTLipIkVEwXrUG9AT0yeOOEYxszarZT1jpsZV58Ly6wlg3SltZpHN26CUalOBmUSY7adQRQiSkCZz2500qSXNkf8Y","API_URL","https://api.val.town",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"]