hackerFish/dsh-video-studio 预览 preview

hackerFish/dsh-video-studio

A DeepSeek Harness native plugin for AI video and motion-comic (漫剧) production: seven-stage director pipeline × multi-account quota pool with failure fallback × credential vault × four-layer prompt engineering with a score-feedback loop. Quality first, cost second.

catalog descriptioncatalog 简介 / catalog description:鲸影 DSH Video Studio:DeepSeek Harness 原生视频/漫剧生成插件——六段导演流水线、多供应商免费额度调度、四层提示词自优化,质量优先省钱第二

Project Overview项目介绍

dsh-video-studio, codenamed Whale, is a native DeepSeek Harness plugin for AI video and motion-comic production. It follows a seven-stage industry-standard pipeline that covers story writing, script breaking, storyboard planning, master asset generation, shot asset variation, video generation, and final cut. To install the plugin, users can run the command dsh plugin --profile web add github:hackerFish/dsh-video-studio --ignore-workspace-root-check directly in their DSH environment. The plugin prioritizes output quality through four-layer prompt engineering and a score-feedback loop, while controlling usage costs via a multi-account quota pool and automatic failure fallback.

The first three stages of the pipeline are completed directly by the large language model of DSH’s current session, so the plugin does not need to call an external LLM. It includes a consistency asset library for character and scene reference, supports parallel batch generation of multiple shots, and has a built-in style memory system to preserve style consistency across sessions. Five pre-built motion-comic script packs covering five popular genres are included, so creators can generate a complete work right out of the box without building a pipeline from scratch. This plugin is designed for AI content creators who want to quickly produce high-quality AI motion comics and videos.

The plugin has verified end-to-end access protocols for multiple AI service providers including Jimeng and Doubao Seedance, and supports both JianYing (CapCut) draft export for manual polishing and ffmpeg local auto-rendering for a complete final output. It follows the open-source MIT license, and all user credentials are stored only in a local encrypted vault and never transmitted to external servers. Output quality is bounded by the capability of the provider’s model, and using provider sessionIDs requires users to comply with the corresponding platform’s terms of service. The plugin is not affiliated with DeepSeek in any official capacity.

dsh-video-studio(代号“鲸”)是专为DeepSeek Harness开发的原生AI视频与漫剧制作插件,遵循七阶段行业标准制作流程:故事撰写、脚本拆分、分镜规划、主资产生成、镜头资产变体、视频生成、最终剪辑,采用四层提示工程加评分回写循环,优先保障生成质量,同时通过多账号配额池和失败回退机制控制调用成本。

前三个阶段由DSH当前会话的大模型直接完成,插件无需额外调用外部大模型。插件支持多账号轮换调度、自动失败回退,内置一致性资产库、并行批量生成、风格记忆库等功能,预设了都市、仙侠、悬疑等五种题材的漫剧脚本包,开箱即可生成完整作品,适合AI内容创作者快速制作AI漫剧与视频。

插件已验证即梦、豆包种子等多家AI服务商的接入协议,支持剪映草稿导出和ffmpeg本地自动渲染,可通过dsh plugin add命令直接安装,遵循MIT许可协议。用户凭证仅保存在本地加密 vault 中,不会外传;生成质量受服务商模型能力限制,使用服务商sessionid需遵守对应平台服务条款,插件与DeepSeek无附属关系。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 warning1 项注意
  • Only 9 stars - very few users, little community feedback星标只有 9,几乎没人在用,遇到问题缺少社区反馈
DSH walks through these 9 checksDSH 会逐条核对这 9 项

Compatibility兼容性

  • DSH, Node, OS and profile requirementsDSH 版本 / Node 版本 / 操作系统 / profile 是否满足要求
  • External dependencies and runtimes (Electron / Python / Docker, ...)外部依赖与运行时(Electron / Python / Docker 等)是否齐备
  • Conflicts with installed plugins: command names, skill / tool names, ports, duplicate MCP registration与已装插件是否冲突:命令名、skill / tool 重名、端口占用、重复 MCP 注册

Security安全性

  • Repo matches the facts registered here; archived or abandoned?仓库是否与页面登记一致,是否归档或长期停更
  • Safety of preinstall / install / postinstall and install.sh / setup.ps1preinstall / install / postinstall 与 install.sh、setup.ps1 是否安全
  • curl|bash, download-then-execute, obfuscation, unrelated domains → stop immediatelycurl|bash、下载即执行、混淆代码、无关域名 → 立刻停止
  • Typosquatting or unmaintained packages among the new dependencies新增依赖里有没有 typosquatting 或无人维护的包
  • Requested permissions vs. what the feature actually needs申请了哪些权限、是否超出功能所需(filesystem / network / shell / clipboard)
  • Any sudo / admin requirement, plus uninstall and rollback是否要求 sudo / 管理员权限,以及卸载与回滚方式

Anything uncertain must be marked unknown with a note on how to confirm it. This site's signal screen is a static snapshot, not a security audit.拿不准的必须标「未知」并说明要我怎么确认。本站的信号筛查是静态快照,不能替代安全审计。

Or use CLI install (for developers)或使用命令行安装(适合开发者)

CLI Install命令行安装

dsh plugin --profile web add github:hackerFish/dsh-video-studio --ignore-workspace-root-check

把 hackerFish/dsh-video-studio 加入你的 DSH 配置(web profile)即可启用。

READMEREADME

🐳 dsh-video-studio (Whale) — AI Video & Motion-Comic Studio for DeepSeek Harness

A DeepSeek Harness native plugin for AI video and motion-comic (漫剧) production: seven-stage director pipeline × multi-account quota pool with failure fallback × credential vault × four-layer prompt engineering with a score-feedback loop. Quality first, cost second.

中文 · Siblings: dsh-lab · awesome-dsh-skills · awesome-dsh-presets

The DSH ecosystem has 1000+ plugins — none does generative video. Whale brings the industry-validated motion-comic pipeline (waoowaoo / LumenX / MangaV / ArcReel pattern) into DSH as a hot-pluggable, per-step-controllable plugin.

Live demo — real AI-generated video (wan2.2-t2v-plus, DashScope, 1080p 5s):

whale live demo

Quantified project assessment

Pipeline — the industry-standard seven-stage workflow

story (LLM, e.g. Doubao writes the novel) → script (LLM breaks it down)
→ storyboard (LLM shot list) → master asset (MJ-style hero image)
→ shot assets (image-model variations for consistency) → video (Seedance/Jimeng/Kling)
→ final cut (JianYing draft export / local ffmpeg render)
  • The first three stages are LLM stages: inside DSH the session model itself does them — the model you're chatting with (Doubao, DeepSeek, whatever) is the brain that writes the story, script and shots; the plugin never calls an external LLM.

Mapping to the industry-standard toolchain:

Industry flow Whale
Doubao writes novel / script / shots = DSH session model doing story/script/storyboard (pick a Doubao model and it IS "Doubao writes")
MJ hero asset = master-asset stage (MJ official API adapter to come; wanx/Seedream can substitute)
image variations = shot-assets stage (wanx ✅ verified · Doubao Seedream ✅ wired)
Seedance video = video stage (Doubao Seedance ✅ / jimeng / kling / ComfyUI)
JianYing final cut = final-cut stage (JianYing draft export ✅ / ffmpeg ✅)
  • Every stage has a gate: auto / ask / manual.
  • Parallel shots, quota scheduler, style genome, distribution pack as before.

Showing the opening section of the README — the full document lives in the repository以上为 README 开头摘要,完整文档在仓库内 · View the full README on GitHub →在 GitHub 查看完整 README →

← 上一个 Prev dock 下一个 Next dsh-pua →