[{"data":1,"prerenderedAt":505},["ShallowReactive",2],{"navigation:en":3,"\u002Fdocs\u002Fflow-videos\u002Freference\u002Faudit:en":129},[4],{"title":5,"path":6,"stem":7,"children":8,"icon":10},"Nuxfire Docs","\u002Fdocs","docs\u002F0.index",[9,11],{"title":5,"path":6,"stem":7,"icon":10},"i-ph-squares-four-duotone",{"title":12,"path":13,"stem":14,"children":15,"icon":18},"Flow Videos","\u002Fdocs\u002Fflow-videos","docs\u002Fflow-videos\u002F0.index",[16,19,36,67,93],{"title":17,"path":13,"stem":14,"icon":18},"Serelius Flow Video","i-ph-video-camera-duotone",{"title":20,"icon":21,"path":22,"stem":23,"children":24,"page":35},"Getting Started","i-ph-rocket-launch-duotone","\u002Fdocs\u002Fflow-videos\u002Fgetting-started","docs\u002Fflow-videos\u002F1.getting-started",[25,30],{"title":26,"path":27,"stem":28,"icon":29},"Overview","\u002Fdocs\u002Fflow-videos\u002Fgetting-started\u002Foverview","docs\u002Fflow-videos\u002F1.getting-started\u002F0.overview","i-ph-play-circle-duotone",{"title":31,"path":32,"stem":33,"icon":34},"Global Features","\u002Fdocs\u002Fflow-videos\u002Fgetting-started\u002Fglobal-features","docs\u002Fflow-videos\u002F1.getting-started\u002F1.global-features","i-ph-gear-six-duotone",false,{"title":37,"icon":38,"path":39,"stem":40,"children":41,"page":35},"Pipeline","i-ph-film-strip-duotone","\u002Fdocs\u002Fflow-videos\u002Fpipeline","docs\u002Fflow-videos\u002F2.pipeline",[42,47,52,57,62],{"title":43,"path":44,"stem":45,"icon":46},"Phase 1 — Script","\u002Fdocs\u002Fflow-videos\u002Fpipeline\u002Fscript","docs\u002Fflow-videos\u002F2.pipeline\u002F1.script","i-ph-scroll-duotone",{"title":48,"path":49,"stem":50,"icon":51},"Phase 2 — Style & Identity","\u002Fdocs\u002Fflow-videos\u002Fpipeline\u002Fstyle-and-identity","docs\u002Fflow-videos\u002F2.pipeline\u002F2.style-and-identity","i-ph-palette-duotone",{"title":53,"path":54,"stem":55,"icon":56},"Phase 3 — Storyboard","\u002Fdocs\u002Fflow-videos\u002Fpipeline\u002Fstoryboard","docs\u002Fflow-videos\u002F2.pipeline\u002F3.storyboard","i-ph-image-duotone",{"title":58,"path":59,"stem":60,"icon":61},"Phase 4 — Videos","\u002Fdocs\u002Fflow-videos\u002Fpipeline\u002Fvideos","docs\u002Fflow-videos\u002F2.pipeline\u002F4.videos","i-ph-video-duotone",{"title":63,"path":64,"stem":65,"icon":66},"Phase 5 — Final Assembly","\u002Fdocs\u002Fflow-videos\u002Fpipeline\u002Ffinal-assembly","docs\u002Fflow-videos\u002F2.pipeline\u002F5.final-assembly","i-ph-scissors-duotone",{"title":68,"icon":69,"path":70,"stem":71,"children":72,"page":35},"Settings & Models","i-ph-sliders-duotone","\u002Fdocs\u002Fflow-videos\u002Fsettings-and-models","docs\u002Fflow-videos\u002F3.settings-and-models",[73,78,83,88],{"title":74,"path":75,"stem":76,"icon":77},"Settings","\u002Fdocs\u002Fflow-videos\u002Fsettings-and-models\u002Fsettings","docs\u002Fflow-videos\u002F3.settings-and-models\u002F0.settings","i-ph-sliders-horizontal-duotone",{"title":79,"path":80,"stem":81,"icon":82},"AI Models & Limits","\u002Fdocs\u002Fflow-videos\u002Fsettings-and-models\u002Fmodels-and-limits","docs\u002Fflow-videos\u002F3.settings-and-models\u002F1.models-and-limits","i-ph-cpu-duotone",{"title":84,"path":85,"stem":86,"icon":87},"Option Catalogs","\u002Fdocs\u002Fflow-videos\u002Fsettings-and-models\u002Fcatalogs","docs\u002Fflow-videos\u002F3.settings-and-models\u002F2.catalogs","i-ph-list-bullets-duotone",{"title":89,"path":90,"stem":91,"icon":92},"Pipeline Calibration","\u002Fdocs\u002Fflow-videos\u002Fsettings-and-models\u002Fcalibration","docs\u002Fflow-videos\u002F3.settings-and-models\u002F3.calibration","i-ph-gauge-duotone",{"title":94,"icon":95,"path":96,"stem":97,"children":98,"page":35},"Reference","i-ph-book-open-duotone","\u002Fdocs\u002Fflow-videos\u002Freference","docs\u002Fflow-videos\u002F4.reference",[99,104,109,114,119,124],{"title":100,"path":101,"stem":102,"icon":103},"Delivered Files","\u002Fdocs\u002Fflow-videos\u002Freference\u002Fdelivered-files","docs\u002Fflow-videos\u002F4.reference\u002F0.delivered-files","i-ph-folder-duotone",{"title":105,"path":106,"stem":107,"icon":108},"Supported Video Types","\u002Fdocs\u002Fflow-videos\u002Freference\u002Fvideo-types","docs\u002Fflow-videos\u002F4.reference\u002F1.video-types","i-ph-television-duotone",{"title":110,"path":111,"stem":112,"icon":113},"What the Tool Does NOT Do","\u002Fdocs\u002Fflow-videos\u002Freference\u002Flimitations","docs\u002Fflow-videos\u002F4.reference\u002F2.limitations","i-ph-prohibit-duotone",{"title":115,"path":116,"stem":117,"icon":118},"Business Models","\u002Fdocs\u002Fflow-videos\u002Freference\u002Fbusiness-models","docs\u002Fflow-videos\u002F4.reference\u002F3.business-models","i-ph-briefcase-duotone",{"title":120,"path":121,"stem":122,"icon":123},"FAQ & Glossary","\u002Fdocs\u002Fflow-videos\u002Freference\u002Ffaq-and-glossary","docs\u002Fflow-videos\u002F4.reference\u002F4.faq-and-glossary","i-ph-question-duotone",{"title":125,"path":126,"stem":127,"icon":128},"Technical and Architectural Audit","\u002Fdocs\u002Fflow-videos\u002Freference\u002Faudit","docs\u002Fflow-videos\u002F4.reference\u002F5.audit","i-ph-shield-check-duotone",{"id":130,"title":125,"body":131,"description":499,"extension":500,"meta":501,"navigation":502,"path":126,"seo":503,"stem":127,"__hash__":504},"docs\u002Fdocs\u002Fflow-videos\u002F4.reference\u002F5.audit.md",{"type":132,"value":133,"toc":486},"minimark",[134,142,164,167,172,175,252,255,259,262,267,322,326,395,399,453,457,475,479],[135,136,137,138,141],"p",{},"This document reproduces the executive conclusions and technical analysis of the architectural audit conducted for ",[139,140,17],"strong",{},".",[143,144,145,152,158],"ul",{},[146,147,148,151],"li",{},[139,149,150],{},"Evaluator",": Senior Google Generative AI Architect specializing in Gemini 2.0, Veo 3.1, and Nano Banana Pro\u002F2.1 via Google Flow SDK.",[146,153,154,157],{},[139,155,156],{},"Audited Target",": Serelius Flow Video (v1.3.0 \u002F Schema V2).",[146,159,160,163],{},[139,161,162],{},"Audit Period",": November 2026 \u002F Q4.",[165,166],"hr",{},[168,169,171],"h2",{"id":170},"_1-executive-verdict-and-ratings","1. Executive Verdict and Ratings",[135,173,174],{},"The application was benchmarked against hundreds of custom tools built on the Google Flow platform across five core dimensions:",[176,177,178,194],"table",{},[179,180,181],"thead",{},[182,183,184,188,191],"tr",{},[185,186,187],"th",{},"Evaluated Dimension",[185,189,190],{},"Score \u002F Rating",[185,192,193],{},"Flow Community Tier",[195,196,197,209,220,231,241],"tbody",{},[182,198,199,203,206],{},[200,201,202],"td",{},"Pipeline Architecture & State Machine",[200,204,205],{},"10 \u002F 10",[200,207,208],{},"Tier 1 (Enterprise \u002F Benchmark)",[182,210,211,214,217],{},[200,212,213],{},"Token Economics & LLM Quota (Gemini 2.0 Flash)",[200,215,216],{},"9.6 \u002F 10",[200,218,219],{},"Top 0.5% across all applications",[182,221,222,225,228],{},[200,223,224],{},"Flow Credit Efficiency (Veo 3.1 & Nano Banana)",[200,226,227],{},"9.8 \u002F 10",[200,229,230],{},"Industrial Reference Standard",[182,232,233,236,238],{},[200,234,235],{},"Sandbox Security & CSP Conformance",[200,237,205],{},[200,239,240],{},"100% iframe CSP conformance",[182,242,243,246,249],{},[200,244,245],{},"Media Engineering & WebCodecs Rendering",[200,247,248],{},"9.9 \u002F 10",[200,250,251],{},"Professional Studio Grade",[135,253,254],{},"Serelius Flow Video sits at the technical pinnacle of applications built on Google Flow. While most community tools rely on naive prompt-to-video cycles without state control, Serelius adopts a client-side Non-Linear Editing (NLE) architecture with a deterministic compiler that eliminates unnecessary LLM calls and confines generative media requests to strictly validated parameters.",[168,256,258],{"id":257},"_2-token-economics-and-generation-optimization","2. Token Economics and Generation Optimization",[135,260,261],{},"The audit evaluated computational efficiency across text, image, and video generation endpoints.",[263,264,266],"h3",{"id":265},"a-gemini-20-flash-optimization","A. Gemini 2.0 Flash Optimization",[268,269,270,288,301],"ol",{},[146,271,272,280,281,284,285,141],{},[139,273,274,275,279],{},"Deterministic Client-Side Script Parsing (",[276,277,278],"code",{},"services\u002Fparser\u002FparseScript.ts",")",":\nStandard Flow tools submit the entire script to an LLM on every minor edit, consuming 10,000 to 50,000 tokens per action. Serelius executes a 100% deterministic client-side parser powered by compiled regexes and a trilingual vocabulary catalog (",[276,282,283],{},"services\u002Fparser\u002FscriptVocabulary.ts","). Continuous script editing consumes zero AI tokens, achieving an estimated ",[139,286,287],{},"~98% reduction in text token consumption",[146,289,290,296,297,300],{},[139,291,292,293,279],{},"Context Windowing for Story Conversion (",[276,294,295],{},"services\u002Fprompts\u002FstoryWindow.ts",":\nWhen converting raw articles, lessons, or stories into structured scripts, the system partitions source text into context windows with an overlap factor (",[276,298,299],{},"STORY_WINDOW_OVERLAP = 0.15","). This sustains high attention density in Gemini 2.0 Flash and eliminates hallucinated or duplicate scenes.",[146,302,303,309,310,313,314,317,318,321],{},[139,304,305,306,279],{},"Surgical Speech Pacing Repair (",[276,307,308],{},"services\u002Fpacing\u002FscriptAudit.ts",":\nPacing is audited locally via client-side word count (",[276,311,312],{},"countWords","). If deviations exceed tolerance thresholds (",[276,315,316],{},"tolerance = 2","), only the specific violating scene batch is submitted for LLM adjustment, capped at two repair rounds (",[276,319,320],{},"PACING_REPAIR_ROUNDS",").",[263,323,325],{"id":324},"b-flow-video-credit-optimization","B. Flow Video Credit Optimization",[268,327,328,352,369,386],{},[146,329,330,336,337,340,341,344,345,344,348,351],{},[139,331,332,333,279],{},"In-Flight Duplicate Prevention (",[276,334,335],{},"inFlightRef",":\nThe generation controller (",[276,338,339],{},"useGenerationControl.ts",") locks active generation keys (",[276,342,343],{},"char:\u003Ckey>",", ",[276,346,347],{},"storyboard:\u003Cid>",[276,349,350],{},"video:\u003Cid>","), preventing duplicate charges from accidental double clicks.",[146,353,354,364,365,368],{},[139,355,356,357,360,361,279],{},"Outdated Media Protection (",[276,358,359],{},"needsStoryboard"," \u002F ",[276,362,363],{},"needsVideo",":\nWith ",[276,366,367],{},"regerarDesatualizados: false"," enabled by default, script text modifications flag scenes as outdated without automatically triggering expensive video re-renders during automated batches.",[146,370,371,377,378,381,382,385],{},[139,372,373,374,279],{},"Flow SDK Constraint Enforcement (",[276,375,376],{},"services\u002Fgeneration\u002FvideoRequests.ts",":\nEnforces the Flow SDK rule that ",[276,379,380],{},"firstFrameImageMediaId"," and ",[276,383,384],{},"referenceImageMediaIds"," are mutually exclusive. It switches automatically between image-to-video interpolation and reference character modes, preventing 400 bad request errors that waste credits.",[146,387,388,394],{},[139,389,390,391,279],{},"Zero-Credit Optical Transitions (",[276,392,393],{},"0 Credits",":\nInstead of generating intermediate AI transition clips, Serelius renders Seamless Stitch, Soften Edges, Cinematic Fade, and Kinetic Push-In locally using pixel shader matrices via WebCodecs at zero credit cost.",[168,396,398],{"id":397},"_3-runtime-reliability-and-sandbox-conformance","3. Runtime Reliability and Sandbox Conformance",[268,400,401,418,431,444],{},[146,402,403,409,410,413,414,417],{},[139,404,405,406,279],{},"Sandboxed File Delivery (",[276,407,408],{},"services\u002Fmedia\u002FfileDownload.ts",":\nComplies strictly with iframe sandbox policies by routing downloads through ",[276,411,412],{},"Flow.download"," with strict file name sanitization (",[276,415,416],{},"sanitizeDownloadFilename",") rather than disallowed browser anchor tags.",[146,419,420,423,424,427,428,321],{},[139,421,422],{},"Dual-Layer Rendering Architecture",":\nEmploys MediaBunny WebCodecs (",[276,425,426],{},"renderWithMediaBunny",") for direct frame-by-frame rendering with hardware acceleration, accompanied by an automatic offscreen canvas fallback (",[276,429,430],{},"renderWithResilientMediaRecorderFallback",[146,432,433,436,437,381,440,443],{},[139,434,435],{},"High-Fidelity Audio Synchronization",":\nMixes voice-over and sound effects at 48kHz (",[276,438,439],{},"services\u002Faudio\u002FreplaceAudioTrack.ts",[276,441,442],{},"mixSceneAudio.ts","), aligning Flow Voice speech tracks with scene transitions.",[146,445,446,452],{},[139,447,448,449,279],{},"Enterprise Pre-Flight Calibration (",[276,450,451],{},"services\u002Fcalibration\u002FpipelineCalibration.ts",":\nDeterministic pre-flight engine benchmarking IndexedDB storage, scene schema coherence, prompt compilation, and WebCodecs codecs before triggering generation runs.",[168,454,456],{"id":455},"_4-roadmap-opportunities","4. Roadmap Opportunities",[268,458,459,469],{},[146,460,461,464,465,468],{},[139,462,463],{},"Native AudioData In-Memory Cache",":\nPre-decoding local base64 audio into uncompressed ",[276,466,467],{},"AudioBuffer"," in memory could further accelerate batch exports exceeding 20 scenes.",[146,470,471,474],{},[139,472,473],{},"Dynamic Prompt Compression for Contemplative Scenes",":\nTrimming default exclusion directives for dialogue-free scenes can decrease Veo 3.1 diffusion queue latency during peak demand periods.",[168,476,478],{"id":477},"_5-audit-conclusion","5. Audit Conclusion",[135,480,481,482,485],{},"Serelius Flow Video qualifies as an ",[139,483,484],{},"Enterprise Master \u002F Production Grade"," tool. It demonstrates complete mastery over Google Flow sandbox constraints, strict credit preservation, deterministic prompt compilation, and resilient client-side media engineering.",{"title":487,"searchDepth":488,"depth":488,"links":489},"",2,[490,491,496,497,498],{"id":170,"depth":488,"text":171},{"id":257,"depth":488,"text":258,"children":492},[493,495],{"id":265,"depth":494,"text":266},3,{"id":324,"depth":494,"text":325},{"id":397,"depth":488,"text":398},{"id":455,"depth":488,"text":456},{"id":477,"depth":488,"text":478},"Independent Google Senior Generative AI Architect audit report assessing token economics, credit efficiency, sandbox conformance, and rendering performance.","md",{},{"icon":128},{"title":125,"description":499},"8AiLshbaceWiejRTBZhokNXsAn_kEub2V2M6wDqL8Lk",1794055329841]