🌬️ Wind Comic _^v12.311

One sentence in. A finished short-form drama out — script, cast, storyboards, voiceover, timeline, mp4.
Multi-agent AI studio · reusable characters · novel→season splitting · director's control room · real-time collab · bring-your-own LLM.

一句话进,整片短剧出 —— 剧本 · 角色 · 分镜 · 配音 · 时间线 · mp4 一条龙。
多 Agent AI 创作工作室 · 可复用角色 · 长篇小说→自动分集 · 导演级控片台 · 实时协作 · 自带 LLM。

English · 简体中文 · 🔥 Pitch · 🔌 BYO LLM

▶ Watch the full 39-second promo — with voiceover & sound
_{Real cinematic footage woven with motion-graphics · 8 distinct art styles · English narration · scored by the platform's own MiniMax music engine.}

Most "AI video" tools give you a 5-second clip from a one-line prompt. Wind Comic gives you a finished short-form drama — script, character bible, multi-shot storyboards, voice-acted lines, BGM, mouth-synced characters (indicative 2D lip bar out of the box; photorealistic face-level lip-sync via BYO wav2lip/SadTalker/MuseTalk or Kling), and a final mp4 — from the same single line.

It works because it doesn't try to be one giant model. It's an honest multi-agent pipeline where each role (Writer, Director, Producer, Character Designer, Storyboard Artist, Cameo Locker, Lipsync, Editor) is a specialist that hands off with strict consistency contracts. Plus a real-time multiplayer timeline to edit with your team like Figma for film.

   "A reborn CEO confronts his cheating ex-fiancée at her wedding."
                                       │
                                       ▼
   Writer ▶ Director ▶ Style Bible ▶ Char Designer ▶ Scene Designer ▶
   ▶ Storyboard (vision-audited) ▶ Video (multi-engine race) ▶
   ▶ TTS (per-character voice) ▶ Lipsync (Kling/Sync.so/Hailuo) ▶
   ▶ Editor (j-cut/l-cut + BGM per act + CJK subtitles) ▶ final.mp4

   + Real-time collab timeline (Yjs CRDT)
   + Bring-your-own LLM (3 env vars, 0 code change)
   + Plug-in image/video providers (12+ supported)
   + Lip-sync pipeline: viseme track → align score → drift auto-correct → engine render → back into timeline   (v10)
   + Template market: save a hit project → rate / favorite / one-click remix (with its voices)               (v10)
   + Per-project cost attribution + budget guard, four-dimension publish gate                                 (v10)

Three views of the same engine. Open on GitHub to watch them animate — flowing dashes trace live data & control paths, traveling dots are data packets moving through the pipeline.

_{System architecture — five layers, top to bottom. The Director threads control across all eight agents; the LLM gateway falls back DeepSeek → MiniMax with zero code change; 12+ media engines plug in behind one router; everything lands on a dual-driver (SQLite ⇄ PostgreSQL) platform.}

_{Sequence — the lifecycle of one idea → film request, time flowing down. Two signature beats: the Vision-Audit retry loop (auto-regenerate any shot scoring < 70) and the multi-engine race (Seedance / Kling / Veo / Vidu — first good clip wins).}

_{Data flow — the artifact refinery. One line of text is refined stage-by-stage (TEXT → JSON → PNG → IMG → MP4); every artifact is persisted to the dual-driver DB + asset store and is independently reusable, so any stage can be re-run in isolation.}

🎞️ Diagrams are animated SVG, authored as code in assets/diagrams/ — crisp at any zoom, versioned with the source. (Animation renders on GitHub; the ModelScope mirror shows static PNG.)

Sketch-Lock storyboarding — every shot first renders a B/W composition sketch, then the final frame is generated locked to that composition (toggle at creation time, or per-shot in the Shot Workshop). Camera language finally survives the diffusion lottery.

Emotion-driven camera — shots missing camera movement get one derived from story beats (hook→push-in, chase→handheld, reveal→zoom-in, farewell→pull-out, emotion temperature → push/orbit/crane). Even the Ken Burns degradation follows the intended move.

Engine matrix, honestly — Veo · MiniMax (Hailuo) · Kling (official Beijing API) · HappyHorse 1.1 (Alibaba, v12.272 — joint video+audio in one pass) with a one-line env priority (VIDEO_ENGINE_ORDER=kling,minimax,veo,happyhorse), base64 first-frame channel for local assets, and an honest fallback chain: a failed shot becomes a labeled animatic, never a still image masquerading as video.

The quality loop — ffprobe-powered film health report (aspect/duration/fps/bitrate/audio/degraded shots) per project and per series; batch re-render of degraded shots (2-way concurrent); a season-export health gate so no broken episode sneaks into the compilation; an engine weather bar so you know before you hit generate.

Round-trip storyboards — export the pull sheet as CSV / Markdown / print-ready PDF (with character sheet + health appendix), edit in Excel, import it back — changed visual fields offer one-click re-render of exactly the affected shots.

Writer resilience — truncated or quote-mangled LLM scripts are salvaged (bracket-stack completion + content-quote escaping) instead of being silently replaced by a template. The real 23KB incident dump lives in the test fixtures.

v3 shipped the pipeline. v6 turned it into a production studio; v7–v9 hardened it into a platform; v10 closed the lip-sync, template-market, and cost loops. Reusable characters, a prompt IDE, novel→season auto-splitting with real voiceover, a 60-style gallery, a director's control room, team credit budgets, an industry-grade script audit (Polish Pro, v7.1), a premium design pass (v8.3), a fully-migrated Postgres backend (v9), a lip-sync delivery pipeline + template market + cost observability (v10), and a live API health board — every screen below is a real capture of the running app.

_{v10 landing — looping cinematic hero, 8 collaborating agents · 7 media engines · 3 consistency guards.}

• 🎙️ Voice & lip-sync, end-to-end — per-character voice routing (auto by name + manual pick/audition) → TTS → viseme keyframe track → measured mouth-vs-audio alignment score (Web-Audio) → drift auto-correct → pluggable engine render (wav2lip / SadTalker / MuseTalk, BYO LIPSYNC_API_URL) → written back into the timeline. One-click whole-film lip-sync with a Vision QC self-heal loop (weak shots auto re-render).
• 🧩 Template market — turn a hit project into a reusable template (style + multi-ref elements + pacing + voices), with a preview clip, ★ ratings & ♥ favorites, quality score, and one-click remix that prefills a new project.
• 💴 Cost observability + budget guard — per-project cost attribution by stage (LLM / image / video / TTS / lip-sync), saving hints, and an ok/warn/over budget guard. The publish-readiness gate is now four-dimensional: picture-vs-script · consistency · lip-sync alignability · measured mouth-vs-audio.

创作工坊 — one line → film: style presets · multi-ref element shelf · character lock · live preview · gold-neon genre gallery	模板市场 — preview clips · ★ ratings / ♥ favorites · quality score · one-click remix (carries voices)
成片质检 + 配音口型 — four-dimension publish gate · consistency trend · viseme track (animated mouth + measured alignment) · per-shot Vision scores	技术监看 · 成本归因 — per-stage cost share (video / image / LLM / TTS / lip-sync) + saving hints

Every screen above is a real capture of the running v10 app. Design language (v8.3 Taste pass — Plus Jakarta Sans + Phosphor icons, gold machined-bezel cards, spring motion, AI gold-neon genre emblems) carries across all of them.

Notes are consolidated per major version — the full minor-by-minor changelog lives in VERSIONS.md / ROADMAP.md.

| v12.49–v12.80 · Commercial Ad Factory (Stage 22) | 电商/品牌广告工厂全链 — 一句 brief(或贴商品 URL 自动出 brief)→ 竖屏成片:现代写实+photoreal 双硬锚(Director 跑偏古装/3D 的实测病根双修 + plan 确定性净化保险)· 结构化 Hook/CTA 文字卡(ffmpeg drawtext 零乱码,Hook 公式化选句:问句>感叹,批量 Hook 变体 A/B,品牌色可配)· karaoke 词级扫光字幕(对齐 TTS 真实时长,抖音/小红书安全区避让)· BGM sidechain 自动闪避 · 《广告法》合规净化(绝对化用语/医疗红线,全入口覆盖)· 质量三防线+账本(逐镜 VLM 门禁 photoreal/烤字/畸变→定向重生 · 视频瞬时错误同引擎重试 · 坏 mp4 完整性拦截 · 失败镜 Ken Burns 兜底保时长 · quality_report 健康分)· 发布预检(三平台硬指标)· rembg 商用安全抠图(产品跨镜一致)+ Kling Elements 打通 · LLM 网关 429/503 健康感知自动降级。2712 tests green。 | | v12.82–v12.120 · Ad Factory 深化 + 供给链扩容 | 视频通道扩容(qyt-vidu(Vidu Q3 via 网关 /ent/v2)入链 pri=75 · minimax 轮询超时可调 · Seedream 4.5 图像尾梯队(720x1280 原生竖屏,实测 14s/张))· 素材质量闭环(Pexels B-roll 双层兜底 + 人设感知检索 + 烤字/字幕 VLM 抽查(B-roll 与 AI 镜双向)+ 结果缓存 LRU)· 响度归一 -14 LUFS(实测 -13.62/-1.42 dBTP 命中平台标准)· 成片抽帧封面精选(VLM 打分,零 T2I 额度)+ 发布包封面链(chosen>AnyText>候选)· karaoke 长台词折行+行内缩字(libass 实渲验证)· 英文广告链(语言感知 CTA + 6 条英文合规红线)· 合规词表可扩展(env+JSON)· 导演/编剧自检修正轮入账 + 导演台 HEALTH KPI + 包装车间结构化结果面板 · LLM 健康缓存全端点对齐(拥堵实测:fable-5 断粮 403 → opus→sonnet→OpenRouter→MiniMax 五级无感降级)。2780 tests green。 |

Features that still ship. Director console · novel→season · finished-film station · team workspace · Cinema timeline are refreshed to v10 (live demo data). The style gallery, API health board, Polish Pro audit, and character turnaround are kept as earlier (v6–v8) captures because they show fuller sample output (the full style grid / live balances / a complete Pro audit / a 3-view turnaround sheet).

Every stage at a glance — what's ready, what's gone stale because you changed something upstream, and a one-click rerun that knows exactly which downstream stages it invalidates.

Paste a full novel; Wind Comic splits it into episodes by chapter markers (or by target length), picks a narration mode, and can render a real narration track + burnable subtitles for the whole season in parallel.

Lock a consistent visual identity before you generate. Search, filter by category, and apply any preset straight into the creation workshop.

Live status for every model and gateway: 正常 / 额度用尽 / 配置缺失 / 不可达, with real balance read-out and a "去充值 / 补配置" hint. Keys are never stored or returned.

Paste a draft, hit Pro: deepseek-v4-pro returns a polished script plus a full industry diagnostic — AIGC-pipeline readiness score (e.g. 85/100), style profile, first-3-second hook strength, Save-the-Cat 3-act breakdown with missing beats called out, on-the-nose dialogue lines flagged, and per-character Cameo/Seedance identity anchors so every shot stays on-model.

Every character gets a real 3-view turnaround sheet (front / three-quarter / back) with a locked structured "DNA prompt" — face geometry, skin tone, signature props, color palette, silhouette identity, full body pose — so the same actor reads identically across all 6 shots. The Cameo IP economy lets the same character travel between projects.

One project, eleven tabs of cockpit-grade control: 导演台 · 剧本 · 角色 · 场景 · 分镜 · 连贯性 · 视频 · 镜头工坊 · Cinema 时间线 · 节奏分析 · 成片质检 · 技术监看 · 参数联动 · 评论协作 · 完整播放. The finished film plays right in the workspace with a 90/100 audit badge and one-click mp4 / platform export.

Credit pool + per-member allocations, RBAC, real invite links.

Multi-track timeline; narration audio + subtitles burned in.

Director plans the story → Writer drafts dialogue under McKee structure → Style Bible Frame locks the look → Character Designer extracts an 8-dimension DNA signature of each character → Storyboard renders with Vision Audit (auto-regen on <70 score) → Video producer races multiple engines (Minimax / Veo / Kling) → Editor cuts j/l-cut on emotional beats and burns CJK subtitles.

We render one canonical "key art" frame from the Director's plan, then pass it as the first --sref of every subsequent storyboard render. Net effect: all 6 shots feel like they came from the same show, not 6 random Midjourney runs. (Most competitors only carry a 2-frame rolling chain — shot 6 doesn't know what shot 1 looked like.)

Writer prompt detects 短剧/漫剧 genres and switches to vertical canvas + injects proven hook patterns (重生回到 N 年前 · 当街掌掴 + 秘密身份 · 系统提示音突响 · etc). McKee 3-act still backs it; tropes are the surface.

The garbled-Chinese-text-in-AI-video problem solved properly: we strip dialogue text from the video prompt (so the model doesn't try to draw garbled glyphs) + add aggressive negatives (--no text --no chinese --no captions) + post-bake real subtitles with ffmpeg subtitles filter using a system CJK font (PingFang / Noto Sans CJK).

Beyond reference image hacks, we run each character's turnaround sheet through Vision LLM to extract structured features (eye shape / jaw angle / hair style / signature outfit etc.), then inject as natural-language anchor into every shot prompt. Combined with cameo-vision-retry: if a shot's character match scores <75, we auto-regen with boosted cw.

• 3 tracks: shots / BGM / subtitle
• Real BGM waveform decoded via Web Audio API (not procedural)
• Drag-to-retime + edge handles to resize duration
• Auto-snap to neighbors within 0.4s threshold + hard-clamp on overlap
• Real-time multiplayer: Yjs awareness paints other users' cursors live, presence avatars show which tab each collaborator is in, Y.Map locks prevent two people editing the same segment
• Project invites with viewer/commenter/editor role gating

Kling lip-sync API for talking heads, with Sync.so and Hailuo as auto-fallback. The pipeline strips dialogue from the prompt so the model only generates lip motion; we then sync the lips to the TTS audio in post. Real-machine limits: dialogue audio must be ≥ 2 seconds and the source video must be reachable at a public URL; ja/ko/ru currently degrade to no lip-sync (honest skip, surfaced in the engine-weather panel).

After Writer finishes, we score each shot 0-10 on a Chinese-conflict-word dictionary + detect emotional polarity reversals + cliffhanger keywords.

v12.275 turned scoring into diagnosis. An average score hides the difference between a story that builds and one that peaks in shot 1 — [1,2,4,9] and [9,4,2,1] both average 4. So v2 adds four things the average cannot see:

• Conflict-curve shape — least-squares slope + peak position + peak prominence, classified as escalating / flat / front-loaded / no-climax.
• Drag-segment localisation — names the exact shot range (e.g. "shots 3–5") where conflict flatlines, instead of reporting one global number.
• Opening density — the first third is audited separately, because completion rate is decided there.
• Duration rhythm — v1 never looked at duration; v2 flags both uniform-length monotony and long-take pile-ups.

Every finding points at the shots to change. All pure functions over existing fields — no extra LLM calls, no added cost.

Every text-LLM call (Director / Writer / Vision / Audit) goes through one OpenAI-compatible chat/completions endpoint. Want to swap to DeepSeek-r1 / GPT-4o / Claude (via OpenRouter) / Qwen-Max / local Ollama? Edit 3 lines in .env. Zero code change. See docs/llm-providers.md for the full matrix.

Every feature listed above is in main, type-checked, unit-tested, and visible at /projects/[id] if you npm install && npm run dev right now.

阵容核验 2026-08-07(Artificial Analysis 盲投竞技场,榜单数值经独立二次检索复核):两周内又出两个重量级新品。① MiniMax H3(Hailuo 3.0):7-31 发布、8-03 开放权重(HuggingFace;授权排除美/欧/英/韩),33B 全模态单次生成、2K/24fps、原生立体声,空降 T2V 带音频榜次席(Elo 1238)、I2V 第三(1190),并拿下「视频编辑(带音频)」榜首。② 字节 Seedance 2.5:7-31 公开发布,单次原生 30 秒(当前最长)、原生 4K 10bit、最多 50 个参考输入、支持区域编辑。 当前榜单 —— 带音频文生视频:Gemini Omni Flash(1244)→ MiniMax H3(1238) → Seedance 2.0 720p(1224)→ Wan 2.7(1161)→ HappyHorse-1.1(1148);带音频图生视频:Seedance 2.0 720p(1198) → Gemini Omni Flash(1191)→ MiniMax H3(1190)→ Grok Imagine 1.5(1114)→ HappyHorse-1.1(1111)。Veo 3.1 仍是画质/物理的企业安全牌(2025-10-15 公开预览,2026-01-13 更新加 4K 与原生 9:16 竖版;8s/次、可链式拼至 ~148s;$0.40/s 720p–1080p、4K $0.60/s、Fast $0.15/s);Kling 3.0 是短剧综合首选 —— 原生最高 4K / 最高 60fps / 最长 15s、最多 6 个连贯镜头,且自带原生音频(中/英/日/韩/西 多语对白 + 口型同步),API 约 $0.084–0.112/s;Wan 2.7 是最便宜的带音频 API($0.10/s)。⚠️ Sora 2 距 API 停服仅剩约 6 周(2026-09-24,消费端已于 4-26 下线),仍在其上开发的项目须立即迁移。

_{本段 8 条高风险数值经独立二次检索复核,4 条被推翻并已按订正值写入(Kling 3.0 曾被误记为「1080p/10s/无原生音频」、Veo 3.1 发布日、Runway Gen-4.5 发布日、Gemini Omni Flash 定价)。榜单 Elo 为 2026-08-07 快照,随投票持续变动。} ⭐ BYO 架构再次接住这波:榜上模型基本都开放 API,填 key 即成为本管线可调度的引擎 —— 竞品越强,本管线越强。

🔴 首个同构开源竞品出现,「开源」不再是差异点本身。 港大 ViMax(MIT,5.6k★,12 个专职 Agent:编剧/分镜师/角色提取/参考图选择 → 端到端出片)已覆盖「剧本→分镜→角色→视频」主干;另有 OpenMontage(24k★,MIT)以 Agent 技能包形式覆盖研究→脚本→资产→剪辑,但无独立 UI(依附 Claude Code/Cursor 等编程环境)。诚实结论:差异化必须从「开源」下沉到「开源 + 制作层纵深」 —— 经逐项核对,ViMax 无配音/TTS、无节奏审计、无 EDL/AAF 剪辑线导出、无团队协作;OpenMontage 无产品 UI、无原生角色一致性。截至 2026-08,全部已查竞品(含闭源 SaaS)中,没有任何一个同时具备「节奏审计 + EDL/AAF 剪辑线导出 + 开源自托管」三项 —— 这才是当下真实的护城河边界。 v12.214→244 双线推进。产品层:GPT Image / Nano Banana(Gemini)接入插件式图像 provider 链(issue #11,社区 @flobo3 提议,OPENAI_IMAGE_ENABLED / GEMINI_API_KEY 门控、原生 i2i 接角色一致性契约);多集连续生成补上「剧情记忆」(第 N 集 Writer 注入前几集前情提要 + 承接纪律,对标红果/阅文的 60~100 集连续,此前各集独立成篇)。平台/工程层:六轮独立对抗复检把安全洞从 CRITICAL 到 LOW 全清(SSRF 逐跳重验重定向 + IPv6 全隧道变体 / serve-file 签名能力 URL / WebSocket 鉴权 / 预算护栏),并把反复踩的「改了守卫却没跟到消费方」这个病固化成 CI 入库门禁(npm run gate:consumer,零容忍,上线即抓到 2 个人肉复检漏掉的真 SSRF)。 结论不变:生成层已是红海(竞品在出片/多镜/音频都第一梯队),Wind Comic 护城河收窄到「制作/平台层」——节奏审计、智能剪辑、字幕烧入、协作、自托管、开源、BYO。

Cells marked ⚠️ = the feature exists but in a limited / locked-down form (e.g. "you can only do this on a paid Pro tier through a UI panel").

本节数据全部来自联网检索(艾媒、DataEye、克劳锐、前瞻、新华网、广电总局公开文件等),含对本项目不利的事实。我们认为写清楚赛道的真实经济结构,比堆一个漂亮的 TAM 更有用。

综合降本 80–90%。 极端案例《安徽小木匠》制作成本 2900 元、收益约 50 万元(ROI ≈ 170 倍)。

① 制作只占总成本 7.5%,投流占 70–85%。 前瞻网口径:平台广告投放 82.5%、制作 7.5%、演员 5%、剧本 1.5%。这意味着「把制作成本再砍一半」对项目盈亏几乎无影响 —— 任何以「更便宜地出片」为唯一卖点的工具,价值天花板极低。

② 单位经济已经很薄。 CPM 从 2025 下半年约 60 元/千次跌到 2026 年 15–30 元;标准 ROI 仅 1.03–1.07(东南亚出海较优,1.2–1.5)。约 90% 的 AI 短剧公司处于亏损,AI 漫剧爆款率不足 0.1%,头部 5% 项目吞掉 70% 行业利润。

③ 平台正在给「纯 AI 生成」降权。 2026 年抖音/爱奇艺/腾讯视频推出分级分账新政,削减无真人出镜的全 AI 短剧保底资源,转而扶持「真人 + AI 协作」。

1. 拉高成品率与质量下限,去够那 <0.1% 的爆款率。 节奏/冲突审计、逐镜 Vision 质检与自动重生、角色一致性(DNA + 嵌入余弦)、发布预检 —— 都是在减少废片,而废片率才是这门生意的真实杀手。
2. 能交付进专业剪辑线,支撑平台正在扶持的「真人+AI 协作」。 EDL / FCPXML / 真二进制 AAF(MS-CFB,给 Avid)导出 —— AI 出的素材可以进人类剪辑师的工程,而不是只能整片直出。这是当前全部已查竞品的共同空白。
3. 自托管 + 开源 + BYO key,服务 B 端代工与品牌定制。 行业正从 C 端投流转向文旅/品牌定制 B 单(客单价数十万至百万级),这类客户对数据不出私域与成本可控有硬要求 —— 闭源 SaaS 天然做不到。

广电总局《微短剧发展管理办法》2026 年 9 月 1 日施行:

• AI 生成/辅助制作的微短剧,须在每集显著位置标注「AI创作」或「AI辅助制作」,字体不得模糊缩小;
• 视频元数据须嵌入 AI 生成溯源信息;
• 三级分类备案 —— 一/二级须播出前取得省级及以上审核许可,无备案不得发行、不得参评。

本项目当前状态(诚实披露):该办法约束的是内容播出/发行方,不是开源工具本身;本项目不运营内容,故不直接适用。工具侧已提供的手段:v12.222 的抖音直发强标 aigc_info + 出海打包结构化 AI 声明(未确认即 422)+ 成片角标(AI_WATERMARK=1 开启,文本可用 AI_WATERMARK_TEXT 覆盖为「AI创作」)。尚未提供:视频元数据溯源字段写入。若你要把成片投放到中国大陆平台,请自行开启角标并按平台要求完成备案与标识 —— 合规责任在运营者一侧。

Real puppeteer captures of the running app (node scripts/capture-v12.mjs). 核心迭代:

• 阶段十九 AI 拉片复刻(v11.1.x):拉片五栏真值表(出厂参数,非 AI 看图猜)· 外部视频 ffmpeg 拆条 + BYO Vision 打标 · 替换工作台(「全员换猫」级全局替换)+ 复刻起片(保原片镜头结构/时长)· 复刻保真度对照 · 存为私有模板。
• 项目/资产管理(v11.2.0):我的项目 / 我的资产 删除 + 下架(级联清理 + 属主守卫)。
• 模型雷达(v10.6.3):一键扫描各 API 最新模型 + 同家族自动升级(四护栏 + 回滚)。
• 钩子审计三指标(v10.6.2):开场 3 秒钩子 / 集尾悬念 / BGM 卡点对齐率。
• 阶段二十 A 智能剪辑(v12.0.x,五刀全交付):卡点剪辑(切点吸附音乐拍点)· 情绪节奏曲线(峰值镜 breathe / 动作镜快切)· 侧重强调(关键镜不压 + 沉稳转场)· 转场审美(按镜头关系选转场)· 一句指令调风格(v12.0.4,BYO)——「快节奏燃向」/「慢叙抒情」一句话调 pacing 力度 + 转场软硬,无 key 走规则、配 key LLM 解析自由文本。
• 阶段二十 B 预览音频(v12.1.x):片段预览叠播配音(静音裸片 + 同步配音轨)+ 成片音频体检自愈(ffprobe 缺流补轨)。
• 阶段二十一 角色/资产一致性升级 · 全局资产记忆库 v2(v12.2.x,五刀全交付):对标同构竞品 OiiOii「角色高维特征向量 + 跨场景一致性」——名称归一修 DNA 漏注入 · DNA/场景锚落库(rerun/重启复用、早镜不漏) · 给 global_assets.embedding 死列通电(BYO 文本嵌入 + 内存余弦检索) · 建角色入口「相似角色」推荐一键复用(防重复建/跨集漂移) · 身份漂移检测(逐镜视觉 embedding 余弦距离标 outlier 漂移镜)。全程无 key 走确定性地板(精确名+文本匹配),有 key 向量增强,诚实降级。
• 安全加固(v11.1.4):JWT 公开兜底密钥根除(进程级随机密钥,旧泄露值作废)。

我的项目 · 删除/下架管理	拉片表 + 复刻工作台

素材库 · 资产管理	API 健康 · 模型雷达

钩子审计三指标	一句指令调剪辑风格(v12.0.4)

片段预览音频:三态徽章 + 带声试听(v12.1.2)
	左:带配音(TTS 叠层)+ 带声试听开关 · 右:片段无独立音轨(成片含配乐+配音)

完整逐版本核心迭代见 VERSIONS.md;阶段计划见 docs/stage18–docs/stage20。

Below is the foundational v3 pipeline (the v6 studio screens are in the New in v6 section above). Every panel is a real puppeteer capture of the running app (run node scripts/capture-screenshots.mjs / node scripts/capture-v6.mjs to refresh).

The 创作总览 dashboard: 99 projects + 4 case studies + recent activity feed + system status (engines in use, model versions).

Cross-project reusable: 角色 / 场景 / 视频 / 音乐 / 字幕 / 模板 — 1467 assets in this demo project.

Every short film with auto-generated cinematic covers + status badges + quality donut.

The whole pipeline as a live agent flow: Writer / Character Designer / Scene Designer / Storyboard Artist / Video Producer / Editor nodes wired together with progress streaming per node, plus a chat side-rail showing every agent message in real time.

The 剧本 tab: every shot with duration, emotion tag (警觉 / 凝重 / 惊恐 / 暴风的沉着 / 镇定的专注…) and a one-line beat note (从表面到深层警觉 / 从无知到悉知威胁 / 从警戒到遭受袭击 …) so the rhythm of the cut is legible at a glance.

3-track layout (SHOTS / BGM / SUBTITLE), drag-to-retime, double-click subtitles to rewrite, drag edges to resize, real BGM waveform (Web Audio decode), live other-user cursors with name labels, segment lock indicators.

KPI: 平均冲突分 / 反转数 / 通过状态. Per-shot conflict-score bar chart with reversal arrows + emotional polarity icons. Color-coded green (≥7) / amber (4-6) / red (<4). Below: actionable warnings + suggestions.

Project-level + per-shot threaded comments with @-autocomplete and notification bell. Each shot collapses for context.

Per-shot "改 prompt 重生" (regenerate image with custom prompt + reference image upload) and "4K 重渲" (Kling Master 4K re-render, plan-gated).

Every model call is provider-pluggable (priority chain + automatic fallback). Creative and high-frequency LLM traffic are split across two model tiers, and MiniMax is always the last-resort fallback on any error / out-of-credits / timeout:

• Why two LLM tiers: the creative tier (DeepSeek -pro, a reasoning model) carries writer/director quality work; the general tier (Claude sonnet-4-6) handles high-frequency planning / validation / Vision-Audit; the -flash tier keeps draft-compare & basic polish at sub-second latency.
• MiniMax safety net: any primary LLM / video / image failure auto-routes to MiniMax (OpenAI-compatible) — surfaced live on the API Health Board (正常 / 额度用尽 / 配置缺失 / 不可达).
• XVERSE-Ent is an open-source MoE screenwriting model you can self-host (vLLM / SGLang / ModelScope) and slot in as the writer/director LLM — set XVERSE_ENABLED=true.
• Swap anything in .env.local (OPENAI_* / OPENAI_CREATIVE_* / VEO_* / IMAGE_MODEL / VE_TTS_MODEL / MINIMAX_*) — zero code change. See docs/llm-providers.md.

# 1. clone + install
git clone https://github.com/ChrisChen667788/wind-comic.git
cd wind-comic
npm install

# 2. configure (3 mandatory lines, see docs/llm-providers.md for swaps)
cp .env.example .env.local
# Edit .env.local:
#   OPENAI_API_KEY=sk-...
#   OPENAI_BASE_URL=https://api.openai.com/v1     # or any compat provider
#   OPENAI_MODEL=gpt-4o                            # or claude-opus-4 via OpenRouter, etc.

# 3. run
npm run dev                # Next.js on :3000
# Optional second terminal for real-time collab:
npm run dev:ws             # Yjs WebSocket server on :1234

# 4. open http://localhost:3000 and create your first short film

Minimum LLM: any model ≥24B parameters that responds in JSON. We've tested gpt-4o, Claude Opus 4, DeepSeek-r1, Qwen-Max, MiniMax-M2, GLM-4.5, Kimi-K2.

Optional engines (graceful fallback when missing):

• MINIMAX_API_KEY — image-01 / Hailuo-2.3 video / speech-2.8-hd TTS / music-2.6 BGM
• KELING_API_KEY — Kling Master 4K + first-last-frame fusion + lip-sync
• HAPPYHORSE_API_KEY (or reuse VECTORENGINE_API_KEY) — HappyHorse 1.1 (Alibaba). Joint video+audio in a single pass; top-5 on both Artificial Analysis boards. Enable by listing it in VIDEO_ENGINE_ORDER (it is not in the default chain, so adding the key alone changes nothing for existing users). Live-verified 2026-08-07: 1080p h264 + aac, ~170s for a 3s clip. Fixed in v12.295 — we had been sending a field that does not exist. From v12.272 we passed size: '9:16'; the upstream has no size parameter at all, and it silently ignores fields it does not recognise, so every clip came back in the default 16:9. (A deliberately bogus size still returns HTTP 200 and creates a task — that is how the silent-fallback was proven.) Per Alibaba's own HappyHorse text-to-video reference, the real fields are ratio (16:9 default, 9:16, 1:1, 4:3, 3:4, 4:5, 5:4, 9:21, 21:9), resolution (480P/720P/1080P), duration, watermark and seed. The usage block the API returns — {"SR": 1080, "ratio": "16:9"} — is simply an echo of those two, which had been in plain sight all along. Also fixed: watermark defaults to true upstream, so every HappyHorse clip carried a bottom-right "Happy Horse" watermark; it is now off by default (HAPPYHORSE_WATERMARK=1 restores it). Vertical is no longer gated off, but every finished task still cross-checks usage.ratio, and a mismatch disables that ratio for the rest of the run instead of burning shot after shot. Tune with HAPPYHORSE_RESOLUTION / HAPPYHORSE_SEED; HAPPYHORSE_SIZE is deprecated and ignored.
• VIDU_API_KEY — Vidu Q3 (long-form 16s clips)
• VEO_API_KEY — Veo 3.1-fast video fallback
• GROK_API_KEY — xAI Grok Imagine 1.5 (T2V/I2V, native audio; BYO — 2026-06 image-to-video #1; auto-preferred when set)
• JIMENG_AK / JIMENG_SK — ByteDance Seedance 2.0 (火山引擎 CV; multi-ref + native A/V; 2026-06 text-to-video #3; BYO)
• LTX_API_KEY (or FAL_KEY) — LTX-2.3 (Lightricks open-weight #2 text-to-video; self-hostable via LTX_BASE_URL; BYO)
• GEN_CONCURRENCY / GEN_CONCURRENCY_VIDEO · _STORYBOARD · _SCENE — per-stage generation concurrency (default 2, max 8). ⚠️ Higher video concurrency is faster but weakens keyframe-chain continuity (shot N pulls shot N-1's last frame) — keep low (1–2) when cross-shot 衔接 matters.
• SYNCSO_API_KEY / HAILUO_API_KEY — alternative lip-sync providers

We're open to PRs. Two things matter most:

1. Don't break the multi-agent contracts. Each agent has explicit input/output shapes — see types/agents.ts.
2. Tests gate everything. Vitest 4002/4002 must stay green. Add tests for new lib/service files.

See CONTRIBUTING.md for the repo's contribution guide.

• docs/llm-providers.md — Swap LLM provider in 3 env vars
• docs/SCREENSHOTS.md — Module-by-module screenshot manifest
• docs/MARKETING-en.md · docs/MARKETING-zh.md — Pitch deck copy
• ROADMAP.md — Full sprint-by-sprint changelog (v2.10 → v9.2.0) · VERSIONS.md — version history table
• docs/COMPETITIVE-GAP-2026-05.md — Honest analysis vs Sora/Kling/Vidu/Higgsfield

MIT. Use it, fork it, build a startup on it. We just ask: if you ship a feature on top, send a PR back.

Wind Comic's own source is MIT. But three npm dependencies ship pre-built binaries under copyleft licenses. This does not affect the MIT status of this repository, and it does not affect you if you simply run or fork the source. It does matter the moment you redistribute a bundle that embeds those binaries (a Docker image, a desktop build, a hosted appliance you hand to a customer):

CI enforces this: scripts/license-check.mjs scans the production dependency tree every run and fails the build if a copyleft dependency appears that is not registered in the table above — so a new GPL dependency can never slip in unnoticed.

Wind Comic stands on a lot of excellent open-source work:

• App & UI — Next.js (App Router + Turbopack) · React 19 · Tailwind CSS v4 · Radix UI · Phosphor Icons + Lucide · Framer Motion · React Flow for the agent-workflow DAG
• Realtime & data — Yjs + y-websocket for CRDT real-time collaboration · better-sqlite3 + node-postgres for the dual-driver persistence layer
• Media & export — FFmpeg via fluent-ffmpeg (CJK subtitle burn / audio mux) · pro NLE interchange through CMX3600 EDL · FCP7 XML · AAF (a from-scratch MS-CFB Compound File Binary writer, no third-party lib)
• Open models & engines — XVERSE-Ent, an open-source MoE screenwriting model (self-hostable via vLLM / SGLang / ModelScope) · FLUX.1 Kontext for image consistency · a provider-pluggable LLM/video chain (DeepSeek · Claude · MiniMax · Veo · Kling · Midjourney) over OpenAI-compatible endpoints
• Methods & algorithms — Robert McKee story structure + Save-the-Cat 3-act beat analysis · CRDT (conflict-free replicated data types) for collaborative editing · cref / sref + 8-dimension character DNA for cross-shot identity consistency
• Tooling — TypeScript (strict) · Vitest · Stripe — and every creator whose real-world feedback shaped the pipeline.

Features that exist because someone outside the team took the time to file a good report:

• @flobo3 — #11: proposed GPT Image (gpt-image-1) and Nano Banana (Gemini Image) as first-class image providers, including the endpoints, the env-var naming, the "reuse the already-configured OPENAI_API_KEY to lower the barrier" idea, and the observation that Nano Banana's native i2i fits the existing character-consistency contract. Shipped in v12.247–v12.247 essentially as designed.
• @MikhailNikolaev44 — #2: reported that selected characters were being ignored on shot regeneration, which uncovered three separate bugs in the reference-image path (v12.132–v12.134), plus the script-language selector request.
• @JSap0914 — #1: fixed broken README links.

Found something? Issues and PRs are genuinely read — several of the pipeline's sharpest fixes started as an outside bug report.

If Wind Comic saved you time, a star helps other creators find it.

Built with ❤️ by people who believe AI-generated drama should feel like a show, not a tech demo.
⭐ Star us if Wind Comic saved you a week.