Communitygithub.com

Drama Director Staging

Turns a short-drama outline or novel scene into a shot plan: rotating camera moves, dense spoken lines, and one speaker per shot.

What is Drama Director Staging?

This is step one of a short-drama bench. Give it an outline or a scene from a novel and it writes a scene script plus plan.json. Six camera moves have to rotate so characters are not stuck in a static over-the-shoulder chat. Dialogue is normalized to spoken Mandarin, paced at about 20 to 28 characters a shot, and anyone who is not speaking is marked silent so later renders do not move every mouth at once.

Works with~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/abserari/drama-production-skills/tree/HEAD/skills/drama-director-staging

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Drama Director Staging (Scriptwriting and Shot Direction Workflow)

Stage Positioning: Stage 1 of the short drama audiovisual industrial pipeline. Transforms a text outline into a professional-grade plan.json and storyboard featuring spatial movement, actor blocking, and high-density dialogue. No Generic Fluff: Does not explain what shot sizes or lenses are. Only consolidates MiniMax H3‑constrained director scheduling criteria and a hardcore, anti‑failure SOP.


1. Operating Contract, Input, and Output

  • Input: Plot outline, scene setting, original character dialogue.
  • Output:
    1. Script.md: Standard scene dialogue text (including movement and emotion annotations);
    2. Storyboard.md and plan.json: Each shot includes camera position, 6 dynamic camera codes, speaker, silent listeners, and normalized pure Chinese dialogue.

2. Hardcore SOP: Six Cinematic Dynamic Camera Rotations

Prohibition (Anti‑example): It is strictly forbidden to use two consecutive static over-the-shoulder (OTS) shot/reverse‑shot face‑to‑face conversations! This easily causes the AI character to stand still, stare blankly, and create an awkward atmosphere.

Shot arrangement must rotate through the following 6 dynamic camera moves:

CodeMode NameEnglish PromptMandatory Use Cases
M1_TRACKINGCross‑path Slow PushSlow forward tracking shot following character walking briskly through the room, camera pushing in steadily, dynamic background movement, continuous tension.Use when the character is walking and talking, pushing through a door, or approaching an opponent – breaks static space.
M2_RACK_FOCUSProp Close‑up Rack FocusExtreme close-up starting on the prop (slammed on table/held in hands), rapid rack focus up to the speaker's intense glaring eyes, shallow depth of field, dramatic depth shift.When a key prop (blade, contract, pocket watch, chip) is slammed on a table or held in hand – shift focus from object to person.
M3_DUTCH_TILTDutch Angle Oblique Intimidation15-degree Dutch angle tilted low shot, looking up at the dominant character stepping forward intimidatingly, creating extreme psychological pressure and cinematic unease.Tense moments: villain approaches, violent intimidation, psychological imbalance, or plot twist.
M4_BACK_REVEALBack‑Turned Head SwivelCharacter begins with their back turned towards camera speaking coldly, suddenly pivots their head 90 degrees to deliver the punchline with an icy stare, dramatic rim lighting catches their jawline.Use when delivering a cold decisive line, betrayal, or an important ruling.
M5_CLOSE_WHISPEREar‑Close IntimidationTight medium close-up, speaker leans in uncomfortably close towards the listener's ear, whispered menacing intensity, listener flinches slightly while speaker holds cold direct gaze.Close‑range threats, secret deals, interrogation whispers – compress distance to the extreme.
M6_PROFILE_STANDSide‑Light Silhouette StandoffCinematic two-shot in profile, characters not looking directly into each other's eyes but glaring past each other, split lighting separating them, smoke drifting between them.Expert showdowns, misaligned gazes, subtext – never a straightforward mutual gaze.

3. Hardcore SOP: Fast‑Paced Short Drama Dialogue Density and Pure Chinese Normalization

3.1 Word Count Density Golden Rule (7.3‑second timing)

  • Standard 175 frames (7.29 seconds) dialogue slot: Strictly controlled between 20 ~ 28 pure Chinese characters (approx. 3.5 ~ 4.2 chars/second – fast‑paced commercial short drama speech rate);
  • Word count & speed criteria:
    • Characters < 16: Judged as too slow – H3 will forcibly lengthen each character's syllable, causing sluggish, dazed speech – must expand dialogue conflict or break down the action;
    • Characters > 30: Judged as too dense – H3 will suffer severe syllable swallowing, slurring, and popping – must condense to under 28 characters;
  • Pre‑/Post‑split Rule: First 4.5 seconds: high‑density utterance; last 2.8 seconds: cold stare, action reaction (gripping knife, raising eyebrow, sneering).

3.2 Pure Chinese Character Normalization (Preventing Language Drift and Foreign Talk)

Before entering the prompt compiler, the normalization processor must be invoked:

  1. 100% convert Arabic numerals to uppercase Chinese characters: 100 -> 一百, 500 -> 五百, EP01 -> 第一集;
  2. No English abbreviations or foreign letters: Boss -> 老大, OK -> 行, watch -> 电子表;
  3. Punctuation standardization: No / \ @ # $ % ^ & * _ [ ] allowed – fully standardize to Chinese full‑width punctuation , 。 ! ? :.

4. Hardcore SOP: Speaker and Silent Listener Tags

In the plan.json output, each shot must explicitly include:

{
  "seg": "seg01",
  "shot_type": "MLS",
  "camera_mode": "M1_TRACKING",
  "speaker": "Sal",
  "silent_listeners": ["Merchant", "Bystander"],
  "dialogue_text": "<d>这笔账拖了整整三年,今天要么给钱,要么留下你的右手!</d>",
  "pacing_check": {
    "char_count": 27,
    "cps": 3.7,
    "status": "OPTIMAL"
  }
}

5. Dedicated Tool Scripts

  • Compiler: python scripts/director_engine.py --validate plan.json
  • Dialogue purification & speed check: python scripts/director_engine.py --clean-dialogue "数字100转汉字"

Individual skills in this repo

This repo contains 5 individual skills — each has its own dedicated page.

comfy-h3-render-core

短剧工业化管线·第 3 工序:H3 物理视频渲染与防翻车调度。负责与本地 ComfyUI 服务交互,调度 MiniMax H3 进行 Turbo 8-step 视频生成。硬核沉淀:Windows WDDM 驱动锁防御、潜空间 32 整除铁律、单人发音锁定与全员闭嘴抑制指令装配、首帧 I2V 安全注入、分镜间显存防泄漏。

comfy-h3-video-producer

本地 ComfyUI 剧本转视频自动生产管线(MiniMax H3)。用户给剧本要求用 ComfyUI 出分镜视频,或提到"一采二采/超分/班长套件/CSGlide/工作流生成视频"时使用。覆盖:工作流握手(不写死节点号)、导演调度与段规划、H3 提示词、提交与自动监工、一采双检闸门、收片清床音与拼接、断点续产。

drama-audio-soundstage

短剧工业化管线·第 4 工序:声音工程与智能避让配乐。负责 H3 视频生成后的声音精修。硬核沉淀:谱减法音频门控(消灭风噪/电流床音)、Whisper ASR 纯普通话语种漂移拦截、FFmpeg 原生侧链动态避让混音(Sidechain Ducking,台词开口 BGM 瞬降 -14dB,闭嘴扬起)。

drama-finishing-upscaler

短剧工业化管线·第 5 工序:终期成片与 1080p 超分增强。负责全集分镜无损拼接与院线级 1080p 超分辨率放大。硬核沉淀:FFmpeg concat 零损耗直拼、RealESRGAN x2plus GPU 独占 FP16 批量超分(~4.8fps)、原生高质量避让音轨无损混流、ffprobe 工业级物理参数验收。

drama-keyframe-forge

短剧工业化管线·第 2 工序:分镜定帧与资产一致性。负责在视频渲染前,建立全局风格母卡 (Style Bible),并为每个分镜前置生成“场景+人物+道具”三合一的 16:9 高清定帧垫图。彻底根治 T2V 盲跑带来的人脸变形、角色换脸、场景变色与道具软化。

Related Skills