feat: redesign creative studio entry

auto-save 2026-05-23 23:45 (~2)
auto-save 2026-05-23 23:36 (~2)
2026-05-23 23:55:05 +08:00 · 2026-05-23 23:45:06 +08:00 · 2026-05-23 23:39:41 +08:00
8 changed files with 965 additions and 1379 deletions
--- a/.gitignore
+++ b/.gitignore
@@ -17,6 +17,7 @@ __pycache__/
 deploy/.env.production
 deploy/.htpasswd
 secrets/
+.backups/

 # api
 api/.venv/
--- a/.memory/assistant-handoff.md
+++ b/.memory/assistant-handoff.md
@@ -1,6 +1,6 @@
 # 项目接力

- 生成时间：May 22, 2026 at 09:01
+- 生成时间：May 23, 2026 at 23:37
 - 项目：SKG Marketing Studio / SKG 营销内容工作台
 - 路径：/Users/kangwan/Projects/business/20260512-20260512-skg-tk-二创验证
 - 状态：active
@@ -9,7 +9,7 @@
 ## 最近助手会话概览

 - Claude：a9e0449c-d9cb-4a2a-bb16-16596dfb552a · 时间未知
- Codex：019e49d1-d114-7d52-8cf0-fbb4585f5ce6 · 时间未知
+- Codex：019e4d33-ad65-7673-934e-815226984ea6 · 时间未知
 - Cursor：未找到匹配当前项目的最近会话

 ## Claude 最近会话
@@ -92,57 +92,39 @@

 ## Codex 最近会话

- Session ID：019e49d1-d114-7d52-8cf0-fbb4585f5ce6
- Transcript：/Users/kangwan/.codex/sessions/2026/05/21/rollout-2026-05-21T17-15-47-019e49d1-d114-7d52-8cf0-fbb4585f5ce6.jsonl
+- Session ID：019e4d33-ad65-7673-934e-815226984ea6
+- Transcript：/Users/kangwan/.codex/sessions/2026/05/22/rollout-2026-05-22T09-01-32-019e4d33-ad65-7673-934e-815226984ea6.jsonl
 - 工作目录：/Users/kangwan/Projects/business/20260512-20260512-skg-tk-二创验证
 - 分支：main
 - 敏感字段：已对 token / key / password / secret 做脱敏

 ### 最近用户要求

- 分镜这个板块，我每次点击视频 都变成了选择 其实我就是想打开看下而已 你这个逻辑不太对  你看下
- 为什么要选择呢  你给我点击下载的按钮不就好了
- 而且每个生成的视频当然是自动保存呀  还需要我点击保存干啥
+- 音频解析 也需需要有中文识别  和多语言识别
+- 脚本那边是自动识别语音识别的内容然后生成脚本文案么？
+- 分镜那边

 ### Codex 最近回复

-> 生产部署正在跑第二轮，远端备份和同步已经完成，Web 镜像在构建；我等脚本验证结束。
+> 对，问题聚焦在分镜这边就更清楚了：我已经看了分镜面板当前实现，下面按“现在实际怎么跑”说。

-> 对，这个点你说得对。生成的视频本来就会写进当前 job 的 `generated_videos`，完成后自动回填 mp4，不应该让用户理解成还要点“保存视频”。
+> 分镜这边现在是“半自动”，不是完全端到端自动重写整片脚本。
 > 
-> 我已经改完并上线：
+> 当前链路是：
 > 
-> - 分镜视频候选轨现在显示 `自动保存`
-> - 视频候选普通点击只负责打开预览
-> - 右上角提供显式 `下载` 按钮
-> - “保存全部 / 保存本条规划”改成了“同步规划 / 同步本条规划”，只表达同步三字段文案规划，不再像保存视频
-> - 源码解析页也同步更新了这条交互语义
+> 1. 音频解析完成后，`job.transcript` 里有逐句时间轴：原语言文案 + 中文镜像。
+> 2. 分镜面板会自动按每个 transcript 段生成一行分镜，保留原音频时间点和原文案依据。
+> 3. 每行会先自动套默认三字段：
+>    - `文案`
+>    - `场景一句话`
+>    - `人物 + 产品 + 动作`
+> 4. 这些默认值目前主要是按段落位置和角色分类生成，比如 hook / pain / proof / solution / cta，不是每句都先让大模型深度理解原文后重写。
+> 5. 当你点单行生成视频候选或同步规划时，才会走 `quickPlanStoryboard`，把当前三字段、参考帧、产品信息扩成完整分镜 prompt。
+> 6. 你改中文字段后，系统会自动优化对应英文主字段，英文用于后续生视频。
 > 
-> 验证和发布：
+> 所以准确说：分镜会自动读取语音识别结果来切分时间轴、显示原文依据、初始化分镜行；但“根据识别内容自动生成更贴合原视频语义的新脚本文案”这层现在还不够强，偏模板化。
 > 
-> - `web/npm run build` 通过
-> - 已推送 Gitea：
->   - `4efb2ce fix: make storyboard video clicks previewable`
->   - `eb4347a fix: clarify storyboard videos auto-save`
-> - 已执行 `./scripts/deploy-prod-safe.sh` 上线，生产验证通过：`/login/ 200`、`/api/health 401`、容器内 `api:health ok`
-> 
-> 当前 `git status -sb`：
-> 
-> ```text
-> ## main...origin/main
->  M .memory/worklog.json
-> ```
-> 
-> 只剩自动日志 `.memory/worklog.json`，业务代码和文档都已提交并推送。
-> 
-> 补充：`npm run lint` 现在跑不了，因为项目里没有可执行的 `eslint`；`tsc --noEmit` 仍有一批既有的 dashboard/ui 旧类型错误，不是这次改动引入的。
-> 
-> <oai-mem-citation>
-> <citation_entries>
-> MEMORY.md:214-286|note=[used SKG worktable context and docs update expectation]
-> </citation_entries>
-> <rollout_ids>
-> 019e2fef-97a7-7523-a1ff-051a8c6…
+> 如果你想要的效果是：音频一识别完，分镜区就自动根据原音频内容生成每句对应的新 SKG 脚本文案，那需要再补一层“transcript -> 三字段候选”的自动脚本生成逻辑。这个方向是合理的。

 ## Cursor 最近会话

@@ -152,7 +134,7 @@

 - 当前分支：main
 - 未提交变更：1 项
- 最近提交：fix: clarify storyboard videos auto-save
+- 最近提交：docs: record multilingual asr deployment
 - 变更文件：
 -    M .memory/worklog.json

--- a/.memory/worklog.json
+++ b/.memory/worklog.json
@@ -1,123 +1,5 @@
 {
  "entries": [
-    {
-      "files_changed": 4,
-      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 4 项未提交变更 · 最近提交：fix: rename source video section title",
-      "ts": "2026-05-19T08:54:36Z",
-      "type": "session-heartbeat"
-    },
-    {
-      "files_changed": 3,
-      "hash": "7a7a3b5",
-      "message": "fix: narrow transcript timeline panel",
-      "ts": "2026-05-19T16:54:51+08:00",
-      "type": "commit"
-    },
-    {
-      "files_changed": 1,
-      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：fix: narrow transcript timeline panel",
-      "ts": "2026-05-19T09:04:36Z",
-      "type": "session-heartbeat"
-    },
-    {
-      "files_changed": 1,
-      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：fix: narrow transcript timeline panel",
-      "ts": "2026-05-19T09:14:36Z",
-      "type": "session-heartbeat"
-    },
-    {
-      "files_changed": 2,
-      "hash": "7c69825",
-      "message": "auto-save 2026-05-19 17:21 (~2)",
-      "ts": "2026-05-19T17:22:01+08:00",
-      "type": "commit"
-    },
-    {
-      "files_changed": 3,
-      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 3 项未提交变更 · 最近提交：auto-save 2026-05-19 17:21 (~2)",
-      "ts": "2026-05-19T09:24:36Z",
-      "type": "session-heartbeat"
-    },
-    {
-      "files_changed": 3,
-      "hash": "1f6eabc",
-      "message": "fix: add temporary filmstrip frame picker",
-      "ts": "2026-05-19T17:27:09+08:00",
-      "type": "commit"
-    },
-    {
-      "files_changed": 4,
-      "hash": "d461c65",
-      "message": "fix: lift filmstrip preview above frame",
-      "ts": "2026-05-19T17:33:37+08:00",
-      "type": "commit"
-    },
-    {
-      "files_changed": 1,
-      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：fix: lift filmstrip preview above frame",
-      "ts": "2026-05-19T09:34:36Z",
-      "type": "session-heartbeat"
-    },
-    {
-      "files_changed": 4,
-      "hash": "cb991e7",
-      "message": "fix: enlarge filmstrip frames in place",
-      "ts": "2026-05-19T17:42:02+08:00",
-      "type": "commit"
-    },
-    {
-      "files_changed": 1,
-      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：fix: enlarge filmstrip frames in place",
-      "ts": "2026-05-19T09:44:37Z",
-      "type": "session-heartbeat"
-    },
-    {
-      "files_changed": 2,
-      "hash": "8bceb90",
-      "message": "auto-save 2026-05-19 17:49 (~2)",
-      "ts": "2026-05-19T17:50:00+08:00",
-      "type": "commit"
-    },
-    {
-      "files_changed": 3,
-      "hash": "883e1d4",
-      "message": "fix: align filmstrip frames with waveform",
-      "ts": "2026-05-19T17:53:17+08:00",
-      "type": "commit"
-    },
-    {
-      "files_changed": 1,
-      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：fix: align filmstrip frames with waveform",
-      "ts": "2026-05-19T09:54:37Z",
-      "type": "session-heartbeat"
-    },
-    {
-      "files_changed": 2,
-      "hash": "97016dc",
-      "message": "fix: hide waveform filmstrip labels",
-      "ts": "2026-05-19T17:59:58+08:00",
-      "type": "commit"
-    },
-    {
-      "files_changed": 1,
-      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：fix: hide waveform filmstrip labels",
-      "ts": "2026-05-19T10:04:37Z",
-      "type": "session-heartbeat"
-    },
-    {
-      "files_changed": 2,
-      "hash": "20f730a",
-      "message": "auto-save 2026-05-19 18:06 (~2)",
-      "ts": "2026-05-19T18:06:17+08:00",
-      "type": "commit"
-    },
-    {
-      "files_changed": 3,
-      "hash": "818d785",
-      "message": "fix: enlarge filmstrip hover near waveform",
-      "ts": "2026-05-19T18:08:17+08:00",
-      "type": "commit"
-    },
    {
      "files_changed": 2,
      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 2 项未提交变更 · 最近提交：fix: enlarge filmstrip hover near waveform",
@@ -3175,37 +3057,150 @@
      "type": "assistant-session"
    },
    {
-      "ts": "2026-05-22T09:05:48+08:00",
-      "type": "commit",
-      "message": "auto-save 2026-05-22 09:01 (~2)",
+      "files_changed": 2,
      "hash": "2342892",
+      "message": "auto-save 2026-05-22 09:01 (~2)",
+      "ts": "2026-05-22T09:05:48+08:00",
+      "type": "commit"
+    },
+    {
+      "files_changed": 1,
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：auto-save 2026-05-22 09:01 (~2)",
+      "ts": "2026-05-22T01:11:31Z",
+      "type": "session-heartbeat"
+    },
+    {
+      "files_changed": 3,
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 3 项未提交变更 · 最近提交：auto-save 2026-05-22 09:01 (~2)",
+      "ts": "2026-05-22T01:21:31Z",
+      "type": "session-heartbeat"
+    },
+    {
+      "files_changed": 5,
+      "hash": "3c357b0",
+      "message": "auto-save 2026-05-22 09:22 (~5)",
+      "ts": "2026-05-22T09:22:27+08:00",
+      "type": "commit"
+    },
+    {
+      "files_changed": 7,
+      "hash": "6427935",
+      "message": "fix: support multilingual audio transcription",
+      "ts": "2026-05-22T09:26:59+08:00",
+      "type": "commit"
+    },
+    {
+      "files_changed": 2,
+      "hash": "7ae92ae",
+      "message": "auto-save 2026-05-22 09:27 (~2)",
+      "ts": "2026-05-22T09:27:54+08:00",
+      "type": "commit"
+    },
+    {
+      "files_changed": 1,
+      "hash": "0d5c326",
+      "message": "docs: record multilingual asr deployment",
+      "ts": "2026-05-22T09:30:59+08:00",
+      "type": "commit"
+    },
+    {
+      "files_changed": 1,
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：docs: record multilingual asr deployment",
+      "ts": "2026-05-22T01:31:31Z",
+      "type": "session-heartbeat"
+    },
+    {
+      "files_changed": 1,
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：docs: record multilingual asr deployment",
+      "ts": "2026-05-22T01:41:31Z",
+      "type": "session-heartbeat"
+    },
+    {
+      "files_changed": 1,
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：docs: record multilingual asr deployment",
+      "ts": "2026-05-22T01:51:31Z",
+      "type": "session-heartbeat"
+    },
+    {
+      "files_changed": 1,
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：docs: record multilingual asr deployment",
+      "ts": "2026-05-22T02:01:32Z",
+      "type": "session-heartbeat"
+    },
+    {
+      "files_changed": 1,
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：docs: record multilingual asr deployment",
+      "ts": "2026-05-22T02:11:59Z",
+      "type": "session-heartbeat"
+    },
+    {
+      "files_changed": 1,
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：docs: record multilingual asr deployment",
+      "ts": "2026-05-22T02:21:59Z",
+      "type": "session-heartbeat"
+    },
+    {
+      "files_changed": 1,
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：docs: record multilingual asr deployment",
+      "ts": "2026-05-22T02:31:59Z",
+      "type": "session-heartbeat"
+    },
+    {
+      "files_changed": 1,
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：docs: record multilingual asr deployment",
+      "ts": "2026-05-22T02:41:59Z",
+      "type": "session-heartbeat"
+    },
+    {
+      "files_changed": 1,
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：docs: record multilingual asr deployment",
+      "ts": "2026-05-22T02:51:59Z",
+      "type": "session-heartbeat"
+    },
+    {
+      "files_changed": 1,
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：docs: record multilingual asr deployment",
+      "ts": "2026-05-22T03:05:43Z",
+      "type": "session-heartbeat"
+    },
+    {
+      "files_changed": 1,
+      "message": "Codex 会话结束 · 持续 0 秒 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：docs: record multilingual asr deployment",
+      "ts": "2026-05-22T03:15:25Z",
+      "type": "session-end"
+    },
+    {
+      "files_changed": 2,
+      "message": "启动 Codex 接力会话 · 已载入 Claude / Codex 最近会话，等待下一条指令 · 分支 main · 2 项未提交变更 · 最近提交：docs: record multilingual asr deployment",
+      "ts": "2026-05-23T15:37:18Z",
+      "type": "assistant-session"
+    },
+    {
+      "ts": "2026-05-23T23:39:41+08:00",
+      "type": "commit",
+      "message": "auto-save 2026-05-23 23:36 (~2)",
+      "hash": "d551c45",
      "files_changed": 2
    },
    {
-      "ts": "2026-05-22T01:11:31Z",
+      "ts": "2026-05-23T23:45:06+08:00",
+      "type": "commit",
+      "message": "auto-save 2026-05-23 23:45 (~2)",
+      "hash": "0e55945",
+      "files_changed": 2
+    },
+    {
+      "ts": "2026-05-23T15:47:20Z",
      "type": "session-heartbeat",
-      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：auto-save 2026-05-22 09:01 (~2)",
+      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 1 项未提交变更 · 最近提交：auto-save 2026-05-23 23:45 (~2)",
      "files_changed": 1
    },
    {
-      "ts": "2026-05-22T01:21:31Z",
-      "type": "session-heartbeat",
-      "message": "Codex 会话活跃 · 最近命令：codex · 分支 main · 3 项未提交变更 · 最近提交：auto-save 2026-05-22 09:01 (~2)",
+      "ts": "2026-05-23T23:50:30+08:00",
+      "type": "commit",
+      "message": "auto-save 2026-05-23 23:50 (~3)",
+      "hash": "e13bb0b",
      "files_changed": 3
-    },
-    {
-      "ts": "2026-05-22T09:22:27+08:00",
-      "type": "commit",
-      "message": "auto-save 2026-05-22 09:22 (~5)",
-      "hash": "3c357b0",
-      "files_changed": 5
-    },
-    {
-      "ts": "2026-05-22T09:26:59+08:00",
-      "type": "commit",
-      "message": "fix: support multilingual audio transcription",
-      "hash": "6427935",
-      "files_changed": 7
    }
  ]
 }
--- a/api/main.py
+++ b/api/main.py
@@ -4275,6 +4275,32 @@ class TranslateReq(BaseModel):
    target: Literal["en", "zh"] = "en"


+class CreativeCopyReq(BaseModel):
+    goal: str
+    product: str = ""
+    audience: str = ""
+    platform: str = "TikTok / Reels"
+    tone: str = "direct"
+    seconds: int = 20
+    source_text: str = ""
+
+
+class CreativeCopyVariant(BaseModel):
+    title: str = ""
+    hook_zh: str = ""
+    script_zh: str = ""
+    script_en: str = ""
+    image_prompt_en: str = ""
+    video_prompt_en: str = ""
+    caption_zh: str = ""
+    hashtags: list[str] = Field(default_factory=list)
+
+
+class CreativeCopyResp(BaseModel):
+    model: str
+    variants: list[CreativeCopyVariant]
+
+
 class ScriptRewriteSegmentReq(BaseModel):
    index: int
    start: float = 0.0
@@ -4339,6 +4365,74 @@ def _ensure_english(text: str) -> str:
        return text


+def _creative_copy_fallback(req: CreativeCopyReq) -> CreativeCopyResp:
+    goal = req.goal.strip() or "展示 SKG 产品的核心卖点"
+    product = req.product.strip() or "SKG 健康科技产品"
+    seconds = max(6, min(60, int(req.seconds or 20)))
+    script_zh = (
+        f"开场 0-3 秒：直接展示{product}和使用场景，提出一个具体痛点。\n"
+        f"中段 3-{max(4, seconds - 5)} 秒：用三个连续镜头说明{goal}，画面保持产品清晰可见。\n"
+        f"结尾 {max(4, seconds - 5)}-{seconds} 秒：给出一句明确行动口播，收在产品近景。"
+    )
+    script_en = _ensure_english(script_zh)
+    image_prompt = _ensure_english(
+        f"{product}, premium health-tech product advertising image, clean lifestyle scene, clear product visibility, natural lighting, vertical composition"
+    )
+    video_prompt = _ensure_english(
+        f"{seconds}-second vertical short video ad for {product}. {goal}. Start with the product in use, show one clear benefit, keep camera motion smooth, realistic lifestyle lighting, no medical treatment claims."
+    )
+    return CreativeCopyResp(
+        model="fallback",
+        variants=[
+            CreativeCopyVariant(
+                title="快速成片版",
+                hook_zh=f"{product}，把一个日常痛点变成一个清楚的使用理由。",
+                script_zh=script_zh,
+                script_en=script_en,
+                image_prompt_en=image_prompt,
+                video_prompt_en=video_prompt,
+                caption_zh=f"{product}｜{goal}",
+                hashtags=["#SKG", "#健康科技", "#短视频广告"],
+            )
+        ],
+    )
+
+
+def _parse_creative_copy_response(raw: str, req: CreativeCopyReq) -> CreativeCopyResp:
+    text = (raw or "").strip()
+    text = re.sub(r"^```(?:json)?\s*", "", text, flags=re.I).strip()
+    text = re.sub(r"\s*```$", "", text).strip()
+    match = re.search(r"\{[\s\S]*\}", text)
+    json_text = match.group(0) if match else text
+    try:
+        data = json.loads(json_text)
+    except Exception:
+        return _creative_copy_fallback(req)
+    raw_items = data.get("variants") if isinstance(data, dict) else None
+    if not isinstance(raw_items, list):
+        return _creative_copy_fallback(req)
+    variants: list[CreativeCopyVariant] = []
+    for item in raw_items[:3]:
+        if not isinstance(item, dict):
+            continue
+        hashtags = item.get("hashtags") or []
+        if not isinstance(hashtags, list):
+            hashtags = []
+        variants.append(CreativeCopyVariant(
+            title=str(item.get("title") or "").strip()[:80],
+            hook_zh=str(item.get("hook_zh") or "").strip()[:180],
+            script_zh=str(item.get("script_zh") or "").strip()[:900],
+            script_en=_ensure_english(str(item.get("script_en") or item.get("script_zh") or "").strip())[:1200],
+            image_prompt_en=_ensure_english(str(item.get("image_prompt_en") or "").strip())[:1200],
+            video_prompt_en=_ensure_english(str(item.get("video_prompt_en") or "").strip())[:1400],
+            caption_zh=str(item.get("caption_zh") or "").strip()[:240],
+            hashtags=[str(tag).strip()[:40] for tag in hashtags if str(tag).strip()][:8],
+        ))
+    if not variants:
+        return _creative_copy_fallback(req)
+    return CreativeCopyResp(model=REWRITE_MODEL if LLM_API_KEY else "fallback", variants=variants)
+
+
@app.post("/translate")
 def translate_text(req: TranslateReq) -> dict:
    """单条文本翻译（给生图自定义提取元素 zh→en 用）"""
@@ -4374,6 +4468,44 @@ def translate_text(req: TranslateReq) -> dict:
        raise HTTPException(500, f"translate failed: {e}")


+@app.post("/creative/copy", response_model=CreativeCopyResp)
+def generate_creative_copy(req: CreativeCopyReq) -> CreativeCopyResp:
+    goal = req.goal.strip()
+    if not goal:
+        raise HTTPException(400, "goal required")
+    if not LLM_API_KEY:
+        return _creative_copy_fallback(req)
+    seconds = max(6, min(60, int(req.seconds or 20)))
+    prompt = (
+        "You are creating practical short-form ad material for an SKG AI creative tool. "
+        "Return strict JSON only. Create 3 distinct variants that can be pasted directly into image/video generation. "
+        "Avoid medical treatment claims; describe comfort, relaxation, daily use, visual proof, and product clarity instead. "
+        "Every variant must include title, hook_zh, script_zh, script_en, image_prompt_en, video_prompt_en, caption_zh, hashtags.\n\n"
+        f"Goal: {goal}\n"
+        f"Product: {req.product.strip() or 'SKG health-tech product'}\n"
+        f"Audience: {req.audience.strip() or 'short-form shoppers'}\n"
+        f"Platform: {req.platform.strip() or 'TikTok / Reels'}\n"
+        f"Tone: {req.tone.strip() or 'direct'}\n"
+        f"Length: {seconds}s\n"
+        f"Source/reference text:\n{req.source_text.strip()[:1500]}"
+    )
+    try:
+        resp = llm().chat.completions.create(
+            model=REWRITE_MODEL,
+            messages=[
+                {"role": "system", "content": "Return valid JSON only. No markdown. No explanation."},
+                {"role": "user", "content": prompt},
+            ],
+            response_format={"type": "json_object"},
+            temperature=0.72,
+            max_tokens=2200,
+        )
+        return _parse_creative_copy_response(resp.choices[0].message.content or "", req)
+    except Exception as e:
+        print(f"[creative copy fallback] {e}", flush=True)
+        return _creative_copy_fallback(req)
+
+
 def _fallback_script_rewrite_item(segment: ScriptRewriteSegmentReq, author_intent: str = "") -> dict:
    source = (segment.source or "").strip()
    intent = _ensure_english(author_intent or "")
@@ -4663,6 +4795,54 @@ async def create_job_from_upload(bg: BackgroundTasks, file: UploadFile = File(..
    return job


+def _write_creative_reference_frame(job_id: str, file_bytes: bytes | None = None) -> tuple[int, int]:
+    frames_dir = job_dir(job_id) / "frames"
+    frames_dir.mkdir(parents=True, exist_ok=True)
+    out = frames_dir / "000.jpg"
+    if file_bytes:
+        try:
+            with Image.open(io.BytesIO(file_bytes)) as raw:
+                im = ImageOps.exif_transpose(raw).convert("RGB")
+                im.thumbnail((1600, 1600), Image.LANCZOS)
+                width, height = im.size
+                im.save(out, "JPEG", quality=92)
+                return width, height
+        except Exception as e:
+            raise HTTPException(400, f"invalid image file: {e}")
+    im = Image.new("RGB", (1024, 1024), (246, 248, 246))
+    im.save(out, "JPEG", quality=92)
+    return im.size
+
+
+@app.post("/creative/jobs/image", response_model=Job)
+async def create_creative_image_job(file: UploadFile | None = File(default=None)) -> Job:
+    job_id = uuid.uuid4().hex[:12]
+    file_bytes: bytes | None = None
+    source_label = "blank"
+    if file and file.filename:
+        ext = Path(file.filename).suffix.lower()
+        if ext not in {".jpg", ".jpeg", ".png", ".webp"}:
+            raise HTTPException(400, f"unsupported image format: {ext}")
+        file_bytes = await file.read()
+        source_label = file.filename
+    width, height = _write_creative_reference_frame(job_id, file_bytes)
+    frame = KeyFrame(index=0, timestamp=0, url=f"/jobs/{job_id}/frames/0.jpg")
+    job = Job(
+        id=job_id,
+        url=f"creative://{source_label}",
+        status="frames_extracted",
+        progress=100,
+        message="创作任务已就绪",
+        width=width,
+        height=height,
+        duration=0,
+        frames=[frame],
+    )
+    JOBS[job_id] = job
+    save_state(job)
+    return job
+
+
@app.post("/jobs/{job_id}/analyze", response_model=Job)
 async def trigger_analyze(
    job_id: str,
--- a/docs/source-analysis.html
+++ b/docs/source-analysis.html
--- a/web/app/layout.tsx
+++ b/web/app/layout.tsx
@@ -12,8 +12,8 @@ const _playfairDisplay = Playfair_Display({
 })

 export const metadata: Metadata = {
-  title: "SKG TK 二创工作台",
-  description: "SKG AI 素材生产管线 · 节点工作流",
+  title: "SKG Creative Studio",
+  description: "SKG AI 图片、视频和文案创作台",
 }

 export default function RootLayout({
--- a/web/app/page.tsx
+++ b/web/app/page.tsx
--- a/web/lib/api.ts
+++ b/web/lib/api.ts
@@ -316,6 +316,54 @@ export async function getRuntimeHealth(): Promise<RuntimeHealth> {
  return res.json()
 }

+export interface CreativeCopyVariant {
+  title: string
+  hook_zh: string
+  script_zh: string
+  script_en: string
+  image_prompt_en: string
+  video_prompt_en: string
+  caption_zh: string
+  hashtags: string[]
+}
+
+export interface CreativeCopyResult {
+  model: string
+  variants: CreativeCopyVariant[]
+}
+
+export async function generateCreativeCopy(body: {
+  goal: string
+  product?: string
+  audience?: string
+  platform?: string
+  tone?: string
+  seconds?: number
+  source_text?: string
+}): Promise<CreativeCopyResult> {
+  const res = await fetch(`${API_BASE}/creative/copy`, {
+    method: "POST",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify(body),
+  })
+  if (!res.ok) {
+    const txt = await res.text().catch(() => "")
+    throw apiError("generateCreativeCopy", res.status, txt)
+  }
+  return res.json()
+}
+
+export async function createCreativeImageJob(file?: File | null): Promise<Job> {
+  const fd = new FormData()
+  if (file) fd.append("file", file)
+  const res = await fetch(`${API_BASE}/creative/jobs/image`, { method: "POST", body: fd })
+  if (!res.ok) {
+    const txt = await res.text().catch(() => "")
+    throw apiError("createCreativeImageJob", res.status, txt)
+  }
+  return res.json()
+}
+
 // 把 ImageRef 解析成可显示的 src URL
 export function resolveImageRefUrl(jobId: string, ref: ImageRef): string {
  if (ref.kind === "keyframe") {
Author	SHA1	Message	Date
kang	3146266383	feat: redesign creative studio entry	2026-05-23 23:55:05 +08:00
kang	0e55945352	auto-save 2026-05-23 23:45 (~2)	2026-05-23 23:45:06 +08:00
kang	d551c45006	auto-save 2026-05-23 23:36 (~2)	2026-05-23 23:39:41 +08:00