《大师级别的电影/短剧制作》Master-Level Film & Short Drama Production · عوامی Open Skill

《大师级别的电影/短剧制作》Master-Level Film & Short Drama Production

“大师级别的电影/短剧制作”凝聚一位导演十余年的创作经验，以及几十万元投入、反复生成与试错积累的 AI 制作心得。将故事判断、镜头语言、演员表演和制作经验整理为可执行的流程，适用于真人电影、短剧、动画与 AI 剧情视频。 从一个创意、一份剧本或一组参考素材出发，按以下流程推进制作： 1. 明确创作方向 确定题材、受众、情绪目标、视觉风格、画幅、时长、对白语言与声音方案，并结合预算和平台能力安排制作范围。已有定稿时，以用户确认的剧情、人物和台词为依据。 2. 梳理剧本与戏剧节奏 检查人物动机、事件因果、冲突推进、情绪转折和悬念落点。明确每场戏的看点、人物状态变化，以及观众在什么时刻产生期待、紧张、感动或喜感，让故事具备清晰的表达和制作依据。 3. 设计导演分镜 把剧本转化为具体可见的镜头，安排景别、机位、构图、运镜、光线、人物站位与视线。根据动作、对白和反应安排时长，确定镜头的起始状态、主要行动、结束状态及剪辑接点。 4. 建立角色、场景与道具资产 整理角色身份、服装、饰品、关键道具和场景参考，区分人物身份图、剧情状态图、背景图与构图图。复用已经合格的资产，为换装、干湿变化、持物变化和特殊姿态补充必要参考，减少角色跑偏与重复制作。 5. 设计表演与动作 将情绪转化为可见的眼神、眉眼、嘴形、呼吸、肩颈及身体动作。接触与运动明确谁发力、接触哪里、向哪里移动、怎样失衡或恢复，以及对方如何反应，让夸张表演仍有可信的动作因果。 6. 制作并检查首尾帧 核对人物身份、服装、姿势、道具归属、背景、构图与光源。固定机位镜头优先以合格首帧为基础局部修改尾帧；复杂动作另外检查必要的接触、释放、承重或最低点状态，避免只看首尾就认定全过程正确。 7. 编写适配模型的视频提示词 根据所选平台的实际能力，明确参考素材用途、起始状态、运动路径、结束状态、摄影机行为、对白及声音。区分图像外观描述与视频运动指令，清除过期版本和相互冲突的要求，形成可直接使用的提示词。 8. 按制作单元生成与测试 在确认的范围和预算内分段推进，优先验证复杂动作、身份稳定和关键接戏。记录每次输入、修改内容与实际结果，使用针对性小样定位问题，减少无方向的重复抽卡。 9. 检查成片与跨镜头连续性 检查角色、服装、道具、站位、视线、运动方向、光线、对白、口型、音效与配乐。检查关键中间动作和镜头接头，区分计划状态与实际生成状态，验收合格后再衔接下一段。 10. 定位问题并局部返修 对闪烁、角色变形、道具漂移、动作缺失、背景变化或对白异常，定位具体时间段与可能原因。优先修正参考资产、起始状态或冲突指令，保留已通过的部分，在预算内完成有明确目标的返修。 11. 剪辑、声音与画面整理 按照故事和情绪节奏组织镜头，处理动作接点、对白停顿、反应与转场；整理配音、环境声、拟音及音乐，统一画面色彩，并根据任务完成字幕、封面和导出方案。 12. 验收与交付 核对最终画幅、分辨率、时长、音画同步和内容完整性，整理成片、素材、提示词及版本信息。根据实际完成的工作提供交付清单，明确尚未验证或需要继续处理的部分。English Detailed Description “Master-Level Film & Short Drama Production” brings together over a decade of a director’s creative experience and practical lessons from hundreds of thousands of yuan invested in AI generation, testing, and refinement. It turns storytelling, cinematic language, performance direction, and production experience into an actionable workflow for live-action films, short dramas, animation, and AI narrative videos. Starting with an idea, a script, or reference materials, the workflow covers: 1. Define the creative direction Establish the genre, audience, emotional goals, visual style, aspect ratio, duration, dialogue language, and sound approach. Plan the scope around the budget and available tools while preserving approved stories, characters, and dialogue. 2. Review the script and dramatic rhythm Examine character motivation, cause and effect, conflict, emotional turns, and suspense. Identify each scene’s appeal, character changes, and intended audience response to create a clear foundation for production. 3. Plan the shots Translate the script into visible actions and deliberate framing. Define shot sizes, camera positions, composition, movement, lighting, blocking, and eyelines. Allocate time for action, dialogue, and reactions, with clear starting states, ending states, and edit points. 4. Build the visual assets Organize character identities, costumes, accessories, props, and locations. Separate identity references from scene-state, background, and composition references. Reuse accepted assets and add only the variations needed for changes in clothing, wetness, props, or poses. 5. Direct performance and action Express emotion through visible gaze, facial movement, breathing, posture, and gestures. For physical interactions, define contact, force, direction, balance, support, and reactions so that exaggerated performances retain clear action logic. 6. Prepare and review start/end frames Check identity, costume, pose, prop ownership, background, framing, and lighting. For fixed-camera shots, prefer editing an accepted start image to create the end image. Review essential intermediate contact, release, and support states in complex actions. 7. Write model-specific video prompts Adapt prompts to the selected platform’s actual capabilities. Specify reference purposes, starting states, motion paths, ending states, camera behavior, dialogue, and sound. Separate visual appearance from motion instructions and remove outdated or conflicting requirements. 8. Generate and test production units Work within the agreed scope and budget. Test difficult actions, identity consistency, and critical transitions early. Record inputs, revisions, and observed results, using focused tests to reduce repeated generation without a clear objective. 9. Review footage and continuity Check characters, costumes, props, positions, eyelines, movement, lighting, dialogue, lip sync, sound effects, and music. Inspect important intermediate actions and transitions, and approve usable results before continuing to dependent shots. 10. Diagnose and repair specific problems Locate flicker, deformation, drifting props, missing actions, background changes, or dialogue errors. Prioritize corrections to references, starting states, or conflicting instructions while preserving accepted material and respecting the revision budget. 11. Edit and refine picture and sound Arrange shots around narrative and emotional rhythm. Refine action cuts, dialogue pauses, reactions, and transitions; organize voices, ambience, effects, and music; and prepare color, captions, covers, and export settings as required. 12. Validate and deliver Check aspect ratio, resolution, duration, audiovisual sync, and completeness. Organize finished videos, assets, prompts, and version information, clearly identifying completed work and any unresolved or unverified items.

@2254459940

ShortDrama

3 1 1

https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4224794/2026-10-07/rqn3pe8j.png

---
name: master-film-short-drama
description: Develop ideas into films and short dramas through scriptwriting, shot planning, visual assets, start/end frames, video prompts, production review, repairs, and editing. 按用户要求完成电影与短剧创作制作，保留已确认剧情与人物。
---

# 大师级别的电影/短剧制作
# Master-Level Film & Short Drama Production

## 任务与输入 / Purpose and inputs
将创意、剧本或参考素材转化为可执行、可接戏的制作方案。读取当前用户要求、最新稿和实际素材，确定题材、受众、画幅、时长、画风、语言、声音和预算。用户已完成的阶段复用，不从头强制重做。用户最新要求优先；已确认剧情、角色与台词不得擅改。
Turn an idea, script, or references into an actionable production plan. Read current requirements, approved material and actual assets. Establish genre, audience, aspect ratio, duration, style, language, sound and budget. Continue from the requested stage; preserve approved story, characters and dialogue.

## 1 创意与三幕结构 / Idea and structure
先写一句话故事：主角为了具体目标，面对阻碍，采用策略，最终作出有代价的选择。三幕为建立与进入、行动与升级、选择与兑现。短视频按时长合并节拍，不硬套长剧固定集数。写清观众看到什么，先产生什么情绪，又在哪个动作或结果处获得回报。方向确认后展开剧本。
Create a logline, protagonist goal, obstacles, strategy and consequential choice. Use setup, escalation and payoff; compress beats for short formats. Identify concrete emotional payoffs and confirm the direction before expanding the script.

## 2 剧本与人工修订 / Script and revision
每场明确时间地点、人物目标、可见行动、对方反应、转折和结果。对白自然、有信息或关系作用，不重复画面。检查人物动机、知情渠道、时间、空间、道具数量和去向。输出问题—依据—修订建议，按授权修改；人工确认后登记当前版本。
Specify location, time, goals, visible action, reactions, turns and outcomes. Review dialogue, motivation, knowledge, timing, space and props. Deliver evidence-based revisions within scope and record the approved version.

## 3 导演分镜 / Shot planning
先查已有资产，再说明有无剧本改动。场级按光线、动作、表演、分镜输出。逐镜列镜号、预计时长、主体、景别、机位、构图、摄影机行为、起态、主要动作、终态、原句对白及语气、声音、资产。景别必须显示所写动作；运动和摄影机分开。时长根据试读、行动和反应安排，没有统一3秒下限。一个镜头聚焦主要行动或情绪，完成够用就切。
Inspect existing assets, state proposed script changes, then plan lighting, action, performance and shots. Include framing, camera behavior, duration, start/action/end states, exact dialogue, sound and references. Allocate time from actual performance; no universal minimum shot length.

## 4 空间与接戏 / Space and continuity
先确定门窗、树、出口等固定地标和人物真实位置，再选机位。正反打绑定对应背景；换机位不搬人物或改变光源。记录上镜终态到下镜起态：位置、朝向、视线、姿势、支撑、衣服干湿、饰品、左右手、道具归属、情绪、未完动作。画外人物不冻结、不瞬移。推撞追赶等明确路线、接触、力、位移和恢复。
Establish landmarks and real positions before camera placement. Match reverse-angle backgrounds and track costume, props, hands, support, eyelines, emotion and unfinished actions. Keep offscreen progress plausible; specify contact and force for physical action.

## 5 剧本资产 / Visual assets
区分角色身份、剧情状态、场景机位、道具和构图参考。每个资产用稳定ID，记录实际文件、用途、版本和验收状态。相同状态复用，只有真实服装、戴摘、干湿、污渍、损坏、持物或姿态变化才补图。逐张实际看图，缺重要参考先报告或在授权内补齐。身份图不替代本镜背景与姿态。
Separate identity, state, scene, prop and composition references. Reuse accepted assets and create only needed variations. Inspect actual images, record IDs and status, and resolve missing essential references before dependent generation.

## 6 首尾帧 / Start and end frames
核对首帧身份、服装、构图、位置、姿态、支撑、持物与光线。固定机位优先由合格首帧局部编辑尾帧，允许动作造成的遮挡、衣料与阴影变化。机位移动按明确设计执行。复杂接触、释放、承重、下沉回升须检查中间状态。首尾正确不证明过程正确；普通多参考上传不是首尾绑定。
Review starting identity, composition, support and props. Prefer local edits of an accepted start image for fixed-camera endings. Inspect intermediate contact and release states. Correct endpoints do not prove correct motion; verify the platform's actual input mode.

## 7 六段视频提示词 / Six-section video prompts
输出可复制提示词，按当前真实素材标签填写：
subject_definitions: 逐张绑定身份、状态、背景或首尾用途。
summary: 画幅、时长、画风、镜头结构和主要行动。
retention_analysis: 保留内容及允许动作变化。
detailed_description: 起始站位、触发、运动路径、接触受力、表演、摄影机、准确对白、结束状态。
overall_soundscape: 已确认对白、环境声与动作拟音。
non_diegetic_music: 项目指定配乐或明确全段无配乐。
图片写画面状态，视频写运动过程。正向为主，限制少而具体、不与行动冲突。真实核对参考、时长、音频和控制能力；未支持或未设置的参数不宣称启用。无配乐文字不是硬开关，实际音轨另验。
Write copy-ready prompts with those six headings. Describe image states separately from video motion. Verify real input, duration and audio capabilities, use targeted constraints, and do not invent control settings.

## 8 生成前与制作 / Preflight and generation
生成前完成空间、接戏、版本检查。两遍核对：空间动作；台词表演与图文对应。关键矛盾未解只能交待核实草稿。按用户确认版本和已授权预算生成，写提示词不等于授权付费。复杂段先测试短单元，记录输入、改动、预期与实际结果。默认每失