multi-model-shot-prompts
多模型镜头提示词适配:把一份分镜表改写成适合不同视频生成模型的提示词(中/英、首帧/首尾帧/纯文生视频、运镜词表、负面约束),模型中立、不编造价格或参数。Use when the user wants the same shot list adapted for Seedance, 可灵 Kling, Veo…
Skill by xianyu110
Chalk Labs
Need help deploying multi-model-shot-prompts?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/xianyu110-multi-model-shot-prompts
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/multi-model-shot-prompts
What's inside
- Skill name
multi-model-shot-prompts- Description
- 多模型镜头提示词适配:把一份分镜表改写成适合不同视频生成模型的提示词(中/英、首帧/首尾帧/纯文生视频、运镜词表、负面约束),模型中立、不编造价格或参数。Use when the user wants the same shot list adapted for Seedance, 可灵 Kling, Veo, Runway, Hailuo/海螺, Wan/通义万相, Sora, Pika or 'any video model', asks 怎么写视频提示词, camera movement terms, or a model returns deformed products.
- License
- MIT
- Instructions
- 98 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# 多模型镜头提示词适配 · Model-Agnostic Shot Prompts
同一个镜头,不同模型「听得懂」的写法不一样。这个 skill 提供一套**通用骨架 + 适配规则**,不依赖任何一家模型的私有参数。
> Part of [ecommerce-video-skills](https://github.com/xianyu110/ecommerce-video-skills)
>
> ⚠️ 本 skill **不提供任何模型的价格、时长上限、分辨率、帧率等参数** —— 这些变化很快,请以各模型官方文档 / 控制台为准。需要时让用户确认或去官方页面查。
## 输入 Inputs
- 分镜表(可来自 `selling-point-storyboard`)
- 目标模型(可多选;不确定就用「通用」)
- 每镜的输入方式:文生视频 / 首帧图生视频 / 首尾帧
- PRODUCT LOCK(商品一致性描述,见 `image-to-video-shots`)
## 通用骨架 Universal skeleton
```text
{主体 + PRODUCT LOCK} , {主体动作} , {环境/场景} , {镜头运动} , {景别/角度} , {光线/色调} , {风格} , {节奏/时长描述}
约束:{保持商品一致} {不生成文字} {不出现其他品牌} {自然手部}
```
写作原则(对大多数模型都成立):
1. **具体动词**:「冰块落入杯中」好于「展示保温效果」。
2. **一镜一个动作,一个运镜**:同时推+摇+环绕,模型容易崩。
3. **镜头词放在固定位置**,用通用说法(见词表)。
4. **商品描述用名词+颜色+材质**,避免形容词堆砌。
5. 图生视频时,提示词**只写变化**(动作、运镜),不要重新描述整张图,否则模型会「重画」。
6. 中文模型可直接中文;多数海外模型英文更稳。两种都给最省事。
## 运镜词表 Camera vocabulary
| 中文 | English | 适合 |
|---|---|---|
| 缓慢推近 | slow push-in / dolly in | 产品亮相、细节 |
| 缓慢拉远 | slow pull-out / dolly out | 从细节到全景、收尾 |
| 横移 | truck left / right | 场景展示 |
| 环绕 | orbit 30° around the product | 白底 360 展示 |
| 俯拍下降 | crane down / top-down descending | 平铺、开箱 |
| 跟拍 | tracking shot following the hand | 使用过程 |
| 固定机位 | locked-off static camera | 带字画面、对比 |
| 微距 | macro close-up, shallow depth of field | 材质、工艺 |
| 快速推近 | quick punch-in | 钩子、重点 |
## 适配规则 Adaptation rules
| 输入模式 | 写法 |
|---|---|
| 文生视频 | 完整骨架 + 详细 PRODUCT LOCK;更容易偏离真实商品,适合氛围镜头,不适合商品特写 |
| 首帧图生视频 | 骨架里删去场景/外观描述,只保留「动作 + 运镜 + 约束」 |
| 首尾帧 | 只描述「从首帧到尾帧发生了什么」+ 运镜 |
| 支持负面提示词 | 加 `deformed product, changed logo, warped text, extra fingers, flicker, watermark` |
| 不支持负面提示词 | 把约束写成正向句:`the product keeps exactly the same shape, colour and logo` |
| 支持参考图/主体参考 | 上传白底主图作为主体参考,提示词中写「the product from the reference」 |
## 输出格式 Output
每镜一块:
```text
### Shot 3 · 卖点:24h 保冰(首帧图生视频)
中文:冰块从上方落入瓶中,溅起细小水花;镜头缓慢推近;明亮柔和棚拍光;商品形状、颜色、Logo 保持不变,不生成文字。
English: Ice cubes drop into the bottle with a small splash; slow push-in; bright soft studio light. The product keeps exactly the same shape, colour and logo. No text.
Negative (if supported): deformed product, changed logo, warped text, flicker, watermark
Fallback: motion "zoom_in" in storyboard.json
```
## 提示词模板 Prompt template(让大模型批量改写)
```text
把下面的分镜表改写为视频生成提示词。目标模型:{通用 / 你的模型名};输入模式:{首帧图生视频}。
规则:一镜一个动作一个运镜;图生视频只写变化;每镜输出中文、English、Negative(若支持)、Fallback motion;
不得写任何模型的价格、时长上限、分辨率等参数;PRODUCT LOCK:{…}
分镜表:{粘贴}
```
## 调用方式(可选)
很多视频模型可通过 OpenAI 兼容或聚合网关调用。本仓库脚本不绑定任何厂商;如需一站式调用视频/图像/大模型,可把网关地址配置为 `BASE_URL`(示例见 README「没有 Claude / API?」)。
## 检查清单 Checklist
- [ ] 每镜:一个动作、一个运镜、带 PRODUCT LOCK 或「只写变化」
- [ ] 中英双版本;约束同时有正向写法
- [ ] 没有写任何未经核实的模型参数/价格
- [ ] 每镜有本地 fallback motion
## 常见问题与修复 Failure fixes
| 问题 | 修复 |
|---|---|
| 模型「重画」了商品 | 图生视频时删掉外观描述,只写动作;加主体参考图 |
| 运镜没生效 | 把运镜词移到句首;只保留一个运镜 |
| 动作太夸张 | 加 `subtle, gentle, slow`;缩短描述 |
| 不同模型风格不统一 | 统一光线与色调描述,后期用同一 BGM/字幕统一 |From xianyu110/ecommerce-video-skills, MIT.
More from xianyu110/ecommerce-video-skills
- 0.679score
ai-voiceover-edge-tts
xianyu110
AI 口播配音:用免费的 edge-tts(微软 Edge 在线语音)为带货短视频逐句生成配音,自动去首尾静音、统计每句时长、缓存复用,可选音色/语速/音调,并给出口播稿改写规则。Use when the user needs a voice-over, 配音, 口播, 旁白, TTS…
SkillAFreshness: A★ 97Claude Code +9 - 0.679score
ffmpeg-auto-assemble
xianyu110
ffmpeg 自动拼接成片:一条命令把 storyboard.json(商品图/视频片段 + 口播稿)渲染成带配音、字幕花字、转场、BGM 自动闪避、响度标准化的 9:16 MP4,纯本地运行。Use when the user wants to assemble/edit/render a product…
SkillAFreshness: A★ 97Claude Code +9 - 0.679score
platform-spec-export
xianyu110
TikTok Shop / 抖音 / 视频号 / 小红书 / 淘宝主图视频 / Amazon 规格导出:把一条母版视频一键转成各平台推荐画幅与编码(9:16、3:4、1:1、16:9,模糊填充或智能裁切),输出规格报告,并可叠加 UI 安全区预览。Use when the user asks for export…
SkillAFreshness: A★ 97Claude Code +9 - 0.679score
subtitles-and-text-effects
xianyu110
字幕与花字:为带货短视频生成 ASS/SRT 字幕与花字(大标题弹出、关键词高亮、贴纸标签、行动号召按钮),自动避开平台底部/右侧 UI 安全区,并用 ffmpeg/libass 烧录。Use when the user wants subtitles, captions, 字幕, 花字, 大字报, keyword…
SkillAFreshness: A★ 97Claude Code +9 - 0.679score
unboxing-comparison-review
xianyu110
开箱 / 对比评测模板:生成电商开箱视频与 A/B 对比评测的结构、口播稿、拍摄清单与分镜(含公平对比规则、测试设计、缺点与适合人群),可直接转成 storyboard.json 渲染。Use when the user wants an unboxing video, 开箱, 测评, 评测, 横评, 对比…
SkillAFreshness: A★ 97Claude Code +9 - 0.679score
cover-title-ab
xianyu110
封面与标题 A/B:为带货短视频生成多套封面(自动挑选清晰帧或用干净素材图 + 不同标题与版式)和多组标题/文案,导出 3:4 / 9:16 封面文件和对比图,并给出 A/B 测试方法与记录表。Use when the user wants a video cover, 封面, 首图, thumbnail, 标题…
SkillAFreshness: A★ 97Claude Code +9 - 0.679score
image-to-video-shots
xianyu110
主图转视频:把一张或几张商品主图/白底图/场景图,写成可直接喂给图生视频模型的镜头提示词(运镜、主体动作、光线、时长、负面约束),并保证商品不变形。Use when the user has product images and wants a product video, 主图视频, 图生视频…
SkillAFreshness: A★ 97Claude Code +9 - 0.679score
live-stream-clips
xianyu110
直播切片:从电商直播录屏中找出高光片段(按时间戳 / 按字幕关键词 / 按停顿自动提议),批量导出 9:16 竖屏切片,自动模糊填充横屏画面、烧录标题与字幕、响度标准化,并给切片写标题与开头钩子。Use when the user has a livestream recording and wants 直播切片…
SkillAFreshness: A★ 97Claude Code +9 - 0.675score
multilingual-dubbing
xianyu110
多语言本地化配音:把中文带货视频本地化为英语、日语、韩语、东南亚语(印尼/泰/越/马来/菲律宾)、西语等版本——本地化改写口播(不是直译)、选对 edge-tts 音色、生成对应字幕与花字,并用同一份 storyboard 一键渲染多语言成片。Use when the user wants dubbing…
SkillAFreshness: A★ 97Claude Code +9 - 0.674score
hook-3s-script
xianyu110
3 秒钩子脚本:为电商短视频写前 3 秒开场(台词 + 画面 + 花字),按痛点/反差/结果先行/提问/数字/场景代入等钩子类型一次给出多版,并做合规自检。Use when the user asks for a hook, 开头, 黄金 3 秒, 前三秒, opening line…
SkillAFreshness: A★ 97Claude Code +9 - 0.674score
viral-remake
xianyu110
爆款拆解复刻:把抖音/TikTok 爆款带货视频(链接、转写或口述结构)拆成 8 维结构与 5-beat 节拍,再映射到你的商品,输出拆解表 + 3 个复刻角度 + 可渲染 storyboard.json。Use when the user says 爆款, 拆解, 复刻, 对标, 仿拍, viral…
SkillAFreshness: A★ 97Claude Code +9 - 0.674score
selling-point-storyboard
xianyu110
卖点分镜表:把商品卖点拆成 15–60 秒带货短视频的逐镜分镜表(时间码、画面、台词、字幕、花字、素材来源、运镜),并直接输出可渲染的 storyboard.json。Use when the user wants a storyboard, 分镜, 分镜脚本, shot list, 带货视频脚本, video…
SkillAFreshness: A★ 97Claude Code +9
Signals
- Stars
- 97
- Forks
- 17
- Open issues
- 0
- Last commit (this folder)
- Oct 4, 2026 (27 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from xianyu110
- 0.784score
sale-banner
xianyu110
Design promotional sale banners and store hero images for Taobao/Tmall (双11, 618, 年货节), Shopee/Lazada (9.9, 11.11, 12.12), TikTok Shop, Amazon Prime Day/Black…
SkillAFreshness: A★ 114Claude Code +9 - 0.784score
batch-background-swap
xianyu110
Batch-replace backgrounds or recolour SKU variants across a folder of product photos (white background, brand colour, seasonal scene, colour variants) while…
SkillAFreshness: A★ 114Claude Code +9 - 0.784score
ecommerce-listing-suite
xianyu110
Orchestrate a complete ecommerce image set for one product: confirm product facts, plan the gallery per platform (Amazon 7-image set, Taobao 5 main + detail…
SkillAFreshness: A★ 114Claude Code +9 - 0.784score
amazon-white-background
xianyu110
Create an Amazon-compliant white-background main image (pure #FFFFFF background, product fills ≥85% of the frame, no text, props or watermark) from a product…
SkillAFreshness: A★ 114Claude Code +9 - 0.784score
model-try-on
xianyu110
Dress a model in a real garment or accessory from a flat-lay, hanger or mannequin photo (on-model / virtual try-on) while preserving fabric, colour, fit…
SkillAFreshness: A★ 114Claude Code +9 - 0.784score
selling-point-infographic
xianyu110
Design selling-point / feature infographic images (one benefit per image, icons, callout lines, short headlines, comparison or spec rows) for Amazon secondary…
SkillAFreshness: A★ 114Claude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get multi-model-shot-prompts running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
multi-model-shot-prompts skill. Agent Mart. https://agentmart.dev/skills/xianyu110-multi-model-shot-prompts (last updated 2026-10-24)