AIDC-AI / Pixelle-Video
AIDC-AI / Pixelle-Video
🎬 Pixelle-Video —— AI 全自动短视频引擎
🎬 Pixelle-Video —— AI Fully Automated Short Video Engine
只需输入一个主题,Pixelle-Video 就能自动完成: Simply enter a topic, and Pixelle-Video will automatically complete:
✍️ 撰写视频文案 | Writing video scripts 🎨 生成 AI 配图/视频 | Generating AI images/videos 🗣️ 合成语音解说 | Synthesizing voiceovers 🎵 添加背景音乐 | Adding background music 🎬 一键合成视频 | One-click video synthesis
零门槛,零剪辑经验,让视频创作成为一句话的事! Zero threshold, zero editing experience—making video creation as simple as a single sentence!
📋 最近更新 | Recent Updates
- 2026-01-26: 新增「动作迁移」模块,上传参考视频和图片进行动作迁移。
- Added “Motion Transfer” module; upload reference videos and images to perform motion transfer.
- 2026-01-14: 新增「数字人口播」和「图生视频」流水线,新增多语言 TTS 音色支持。
- Added “Digital Human Avatar” and “Image-to-Video” pipelines, plus multi-language TTS voice support.
- 2026-01-06: 新增 RunningHub 48G 显存机器调用支持。
- Added support for calling RunningHub 48G VRAM machines.
- 2025-12-28: 支持 RunningHub 并发限制可配置,优化 LLM 返回结构化数据的逻辑。
- Added configurable concurrency limits for RunningHub; optimized logic for LLM structured data output.
- 2025-12-17: 支持 ComfyUI API Key 配置,支持 Nano Banana 模型调用,API 接口支持模板自定义参数。
- Added ComfyUI API Key configuration, Nano Banana model support, and custom template parameters for API endpoints.
- 2025-12-10: 侧边栏内置 FAQ,锁定 edge-tts 版本修复 TTS 服务不稳定问题。
- Built-in FAQ in the sidebar; locked edge-tts version to fix TTS service instability.
- 2025-12-08: 支持固定脚本多种分割方式(段落/行/句子),优化模板选择交互逻辑支持直接预览选择。
- Added multiple segmentation methods for fixed scripts (paragraphs/lines/sentences); optimized template selection UI for direct preview.
- 2025-12-06: 修复视频生成 API 返回 URL 路径处理,支持跨平台兼容。
- Fixed video generation API URL path handling for cross-platform compatibility.
- 2025-12-05: 新增 Windows 整合包下载,优化图片与视频反推工作流。
- Added Windows integrated package download; optimized image and video reverse-inference workflows.
- 2025-12-04: 新增「自定义素材」功能,支持用户上传自己的照片和视频,AI 智能分析生成脚本。
- Added “Custom Assets” feature; users can upload their own photos/videos for AI-powered script generation.
- 2025-11-18: 优化 RunningHub 服务调用支持并行处理,新增历史记录页面,支持批量创建视频任务。
- Optimized RunningHub service calls for parallel processing; added history page and batch video task creation.
✨ 功能亮点 | Key Features
- 全自动生成: 输入主题,自动生成完整视频。
- Fully Automated: Enter a topic to generate a complete video automatically.
- AI 智能文案: 根据主题智能创作解说词,无需自己写脚本。
- AI Smart Scripting: Intelligently creates narration based on the topic; no manual scriptwriting required.
- AI 生成配图: 每句话都配上精美的 AI 插图。
- AI Image Generation: Every sentence is paired with beautiful AI-generated illustrations.
- AI 生成视频: 支持使用 AI 视频生成模型(如 WAN 2.1)创建动态视频内容。
- AI Video Generation: Supports AI video models (e.g., WAN 2.1) to create dynamic content.
- AI 生成语音: 支持 Edge-TTS、Index-TTS 等众多主流 TTS 方案。
- AI Voice Synthesis: Supports mainstream TTS solutions like Edge-TTS and Index-TTS.
- 背景音乐: 支持添加 BGM,让视频更有氛围。
- Background Music: Add BGM to enhance the video atmosphere.
- 视觉风格: 多种模板可选,打造独特视频风格。
- Visual Styles: Multiple templates available to create unique video aesthetics.
- 灵活尺寸: 支持竖屏、横屏等多种视频尺寸。
- Flexible Aspect Ratios: Supports portrait, landscape, and other video dimensions.
- 多种 AI 模型: 支持 GPT、通义千问、DeepSeek、Ollama 等。
- Multi-AI Model Support: Compatible with GPT, Qwen, DeepSeek, Ollama, etc.
- 原子能力灵活组合: 基于 ComfyUI 架构,可使用预置工作流,也可自定义任意能力。
- Flexible Modular Capabilities: Based on the ComfyUI architecture; use preset workflows or customize any component (e.g., replace image models with FLUX, swap TTS with ChatTTS).
📊 视频生成流程 | Video Generation Workflow
Pixelle-Video 采用模块化设计,整个视频生成流程清晰简洁: Pixelle-Video adopts a modular design, making the generation process clear and concise:
文案生成 → 配图规划 → 逐帧处理 → 视频合成 Script Generation → Image Planning → Frame-by-Frame Processing → Video Synthesis
每个环节都支持灵活定制,可选择不同的 AI 模型、音频引擎、视觉风格等,满足个性化创作需求。 Every step is customizable, allowing you to select different AI models, audio engines, and visual styles to meet your creative needs.
🚀 快速开始 | Quick Start
🪟 Windows 一键整合包 (推荐) | Windows One-Click Package (Recommended)
无需安装 Python、uv 或 ffmpeg,一键开箱即用! No need to install Python, uv, or ffmpeg—ready to use out of the box!
- 下载: 获取最新的 Windows 一键整合包并解压。
- Download: Get the latest Windows integrated package and extract it.
- 启动: 双击运行
start.bat启动 Web 界面。- Launch: Double-click
start.batto start the Web UI.
- Launch: Double-click
- 访问: 浏览器会自动打开
http://localhost:8501。- Access: The browser will automatically open
http://localhost:8501.
- Access: The browser will automatically open
- 配置: 在「⚙️ 系统配置」中配置 LLM API 和图像生成服务。
- Configure: Set up your LLM API and image generation service in “⚙️ System Configuration”.
💻 从源码安装 (macOS / Linux) | Install from Source
- 安装依赖: 安装
uv和ffmpeg。- Install Dependencies: Install
uvandffmpeg.
- Install Dependencies: Install
- 下载项目:
git clone https://github.com/AIDC-AI/Pixelle-Video.git- Clone Project:
git clone https://github.com/AIDC-AI/Pixelle-Video.git
- Clone Project:
- 启动:
uv run streamlit run web/app.py- Launch:
uv run streamlit run web/app.py
- Launch:
💬 社区交流 | Community
扫描下方二维码加入我们的社区,获取最新动态和技术支持: Scan the QR code below to join our community for the latest updates and technical support:
[微信群 / Discord 社区] [WeChat Group / Discord Community]
📝 许可证 | License
本项目采用 Apache 2.0 许可证。 This project is licensed under the Apache 2.0 License.