[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-deepfakes--faceswap":3,"tool-deepfakes--faceswap":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":70,"readme_en":71,"readme_zh":72,"quickstart_zh":73,"use_case_zh":74,"hero_image_url":75,"owner_login":76,"owner_name":77,"owner_avatar_url":78,"owner_bio":77,"owner_company":77,"owner_location":77,"owner_email":77,"owner_twitter":77,"owner_website":77,"owner_url":79,"languages":80,"stars":92,"forks":93,"last_commit_at":94,"license":95,"difficulty_score":10,"env_os":96,"env_gpu":97,"env_ram":98,"env_deps":99,"category_tags":103,"github_topics":104,"view_count":117,"oss_zip_url":77,"oss_zip_packed_at":77,"status":16,"created_at":118,"updated_at":119,"faqs":120,"releases":149},1064,"deepfakes\u002Ffaceswap","faceswap","Deepfakes Software For All","FaceSwap 是一款基于深度学习的开源工具，专注于实现图片和视频中的人脸自动识别与替换。通过训练神经网络模型，它能够精准捕捉面部特征并完成跨图像\u002F视频的自然融合，例如将演员的面部表情迁移到另一张面孔上，或在视频中批量替换特定人脸。\n\n这项技术降低了深度伪造技术的使用门槛，解决了传统AI换脸需要复杂代码能力和算力投入的难题。用户无需机器学习背景，即可通过图形化界面（GUI）完成从素材提取、模型训练到效果生成的全流程。其支持多种预训练模型（如Phaze-A、Villain），在消费级显卡上即可实现高质量输出。\n\n适合数字内容创作者、影视特效爱好者、AI研究者及普通用户使用。设计师可将其用于创意视频制作，研究人员能基于开源框架探索生成对抗网络（GAN）的优化方案，而普通用户通过社区教程也能快速上手。项目提供详细的中文安装指南和活跃的Discord支持社区，兼顾不同技术层级的需求。\n\n技术上采用卷积神经网络（CNN）实现面部关键点检测，并通过自编码器架构保持换脸后的自然光影过渡。其模型训练模块支持自定义数据集导入，为专业用户提供了算法调优空间。需要注意的是，开发者在宣言中强调技术的伦理使用","FaceSwap 是一款基于深度学习的开源工具，专注于实现图片和视频中的人脸自动识别与替换。通过训练神经网络模型，它能够精准捕捉面部特征并完成跨图像\u002F视频的自然融合，例如将演员的面部表情迁移到另一张面孔上，或在视频中批量替换特定人脸。\n\n这项技术降低了深度伪造技术的使用门槛，解决了传统AI换脸需要复杂代码能力和算力投入的难题。用户无需机器学习背景，即可通过图形化界面（GUI）完成从素材提取、模型训练到效果生成的全流程。其支持多种预训练模型（如Phaze-A、Villain），在消费级显卡上即可实现高质量输出。\n\n适合数字内容创作者、影视特效爱好者、AI研究者及普通用户使用。设计师可将其用于创意视频制作，研究人员能基于开源框架探索生成对抗网络（GAN）的优化方案，而普通用户通过社区教程也能快速上手。项目提供详细的中文安装指南和活跃的Discord支持社区，兼顾不同技术层级的需求。\n\n技术上采用卷积神经网络（CNN）实现面部关键点检测，并通过自编码器架构保持换脸后的自然光影过渡。其模型训练模块支持自定义数据集导入，为专业用户提供了算法调优空间。需要注意的是，开发者在宣言中强调技术的伦理使用，倡导将工具应用于创意表达、影视修复等正当场景。","# deepfakes_faceswap\n\n### Important information for **Patreon** and **PayPal** supporters. Please see this forum post: https:\u002F\u002Fforum.faceswap.dev\u002Fviewtopic.php?f=14&t=3120\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Ffaceswap.dev\">\u003Cimg src=\"https:\u002F\u002Fi.imgur.com\u002FzHvjHnb.png\">\u003C\u002Fimg>\u003C\u002Fa>\n\u003Cbr \u002F>FaceSwap is a tool that utilizes deep learning to recognize and swap faces in pictures and videos.\n\u003C\u002Fp>\n\u003Cp align=\"center\">\n\u003Cimg src = \"https:\u002F\u002Fi.imgur.com\u002FnWHFLDf.jpg\">\u003C\u002Fimg>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n\u003Ca href=\"https:\u002F\u002Fwww.patreon.com\u002FbePatron?u=23238350\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_acde232a22d6.png\">\u003C\u002Fimg>\u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002FFC54sYg\">\u003Cimg src=\"https:\u002F\u002Fi.imgur.com\u002FgIpztkv.png\">\u003C\u002Fimg>\u003C\u002Fa>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fwww.dailymotion.com\u002Fvideo\u002Fx810mot\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_3e5a8b151398.png\">\u003C\u002Fimg>\u003C\u002Fa>\n\u003Cbr \u002F>Emma Stone\u002FScarlett Johansson FaceSwap using the Phaze-A model\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=r1jng79a5xc\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_ab44b8f10460.jpg\">\u003C\u002Fimg>\u003C\u002Fa>\n\u003Cbr \u002F>Jennifer Lawrence\u002FSteve Buscemi FaceSwap using the Villain model\n\u003C\u002Fp>\n\n\n![Build Status](https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Factions\u002Fworkflows\u002Fpytest.yml\u002Fbadge.svg) [![Documentation Status](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_13d664e1afd7.png)](https:\u002F\u002Ffaceswap.readthedocs.io\u002Fen\u002Flatest\u002F?badge=latest)\n\nMake sure you check out [INSTALL.md](INSTALL.md) before getting started.\n\n- [deepfakes\\_faceswap](#deepfakes_faceswap)\n    - [Important information for **Patreon** and **PayPal** supporters. Please see this forum post: https:\u002F\u002Fforum.faceswap.dev\u002Fviewtopic.php?f=14\\&t=3120](#important-information-for-patreon-and-paypal-supporters-please-see-this-forum-post-httpsforumfaceswapdevviewtopicphpf14t3120)\n- [Manifesto](#manifesto)\n  - [FaceSwap has ethical uses.](#faceswap-has-ethical-uses)\n- [How To setup and run the project](#how-to-setup-and-run-the-project)\n- [Overview](#overview)\n  - [Extract](#extract)\n  - [Train](#train)\n  - [Convert](#convert)\n  - [GUI](#gui)\n- [General notes:](#general-notes)\n- [Help I need support!](#help-i-need-support)\n  - [Discord Server](#discord-server)\n  - [FaceSwap Forum](#faceswap-forum)\n- [Donate](#donate)\n  - [Patreon](#patreon)\n  - [One time Donations](#one-time-donations)\n    - [@torzdf](#torzdf)\n    - [@andenixa](#andenixa)\n- [How to contribute](#how-to-contribute)\n  - [For people interested in the generative models](#for-people-interested-in-the-generative-models)\n  - [For devs](#for-devs)\n  - [For non-dev advanced users](#for-non-dev-advanced-users)\n  - [For end-users](#for-end-users)\n- [About machine learning](#about-machine-learning)\n  - [How does a computer know how to recognize\u002Fshape faces? How does machine learning work? What is a neural network?](#how-does-a-computer-know-how-to-recognizeshape-faces-how-does-machine-learning-work-what-is-a-neural-network)\n\n# Manifesto\n\n## FaceSwap has ethical uses.\n\nWhen faceswapping was first developed and published, the technology was groundbreaking, it was a huge step in AI development. It was also completely ignored outside of academia because the code was confusing and fragmentary. It required a thorough understanding of complicated AI techniques and took a lot of effort to figure it out. Until one individual brought it together into a single, cohesive collection. It ran, it worked, and as is so often the way with new technology emerging on the internet, it was immediately used to create inappropriate content. Despite the inappropriate uses the software was given originally, it was the first AI code that anyone could download, run and learn by experimentation without having a Ph.D. in math, computer theory, psychology, and more. Before \"deepfakes\" these techniques were like black magic, only practiced by those who could understand all of the inner workings as described in esoteric and endlessly complicated books and papers.\n\n\"Deepfakes\" changed all that and anyone could participate in AI development. To us, developers, the release of this code opened up a fantastic learning opportunity. It allowed us to build on ideas developed by others, collaborate with a variety of skilled coders, experiment with AI whilst learning new skills and ultimately contribute towards an emerging technology which will only see more mainstream use as it progresses.\n\nAre there some out there doing horrible things with similar software? Yes. And because of this, the developers have been following strict ethical standards. Many of us don't even use it to create videos, we just tinker with the code to see what it does. Sadly, the media concentrates only on the unethical uses of this software. That is, unfortunately, the nature of how it was first exposed to the public, but it is not representative of why it was created, how we use it now, or what we see in its future. Like any technology, it can be used for good or it can be abused. It is our intention to develop FaceSwap in a way that its potential for abuse is minimized whilst maximizing its potential as a tool for learning, experimenting and, yes, for legitimate faceswapping.\n\nWe are not trying to denigrate celebrities or to demean anyone. We are programmers, we are engineers, we are Hollywood VFX artists, we are activists, we are hobbyists, we are human beings. To this end, we feel that it's time to come out with a standard statement of what this software is and isn't as far as us developers are concerned.\n\n- FaceSwap is not for creating inappropriate content.\n- FaceSwap is not for changing faces without consent or with the intent of hiding its use.\n- FaceSwap is not for any illicit, unethical, or questionable purposes.\n- FaceSwap exists to experiment and discover AI techniques, for social or political commentary, for movies, and for any number of ethical and reasonable uses.\n\nWe are very troubled by the fact that FaceSwap can be used for unethical and disreputable things. However, we support the development of tools and techniques that can be used ethically as well as provide education and experience in AI for anyone who wants to learn it hands-on. We will take a zero tolerance approach to anyone using this software for any unethical purposes and will actively discourage any such uses.\n\n# How To setup and run the project\nFaceSwap is a Python program that will run on multiple Operating Systems including Windows, Linux, and MacOS.\n\nSee [INSTALL.md](INSTALL.md) for full installation instructions. You will need a modern GPU with CUDA support for best performance. Many AMD GPUs are supported through ROCm (Linux).\n\n# Overview\nThe project has multiple entry points. You will have to:\n - Gather photos and\u002For videos\n - **Extract** faces from your raw photos\n - **Train** a model on the faces extracted from the photos\u002Fvideos\n - **Convert** your sources with the model\n\nCheck out [USAGE.md](USAGE.md) for more detailed instructions.\n\n## Extract\nFrom your setup folder, run `python faceswap.py extract`. This will take photos from `src` folder and extract faces into `extract` folder.\n\n## Train\nFrom your setup folder, run `python faceswap.py train`. This will take photos from two folders containing pictures of both faces and train a model that will be saved inside the `models` folder.\n\n## Convert\nFrom your setup folder, run `python faceswap.py convert`. This will take photos from `original` folder and apply new faces into `modified` folder.\n\n## GUI\nAlternatively, you can run the GUI by running `python faceswap.py gui`\n\n# General notes:\n- All of the scripts mentioned have `-h`\u002F`--help` options with arguments that they will accept. You're smart, you can figure out how this works, right?!\n\nNB: there is a conversion tool for video. This can be accessed by running `python tools.py effmpeg -h`. Alternatively, you can use [ffmpeg](https:\u002F\u002Fwww.ffmpeg.org) to convert video into photos, process images, and convert images back to the video.\n\n\n**Some tips:**\n\nReusing existing models will train much faster than starting from nothing.\nIf there is not enough training data, start with someone who looks similar, then switch the data.\n\n# Help I need support!\n## Discord Server\nYour best bet is to join the [FaceSwap Discord server](https:\u002F\u002Fdiscord.gg\u002FFC54sYg) where there are plenty of users willing to help. Please note that, like this repo, this is a SFW Server!\n\n## FaceSwap Forum\nAlternatively, you can post questions in the [FaceSwap Forum](https:\u002F\u002Ffaceswap.dev\u002Fforum). Please do not post general support questions in this repo as they are liable to be deleted without response.\n\n# Donate\nThe developers work tirelessly to improve and develop FaceSwap. Many hours have been put in to provide the software as it is today, but this is an extremely time-consuming process with no financial reward. If you enjoy using the software, please consider donating to the devs, so they can spend more time implementing improvements.\n\n## Patreon\nThe best way to support us is through our Patreon page:\n\n[![become-a-patron](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_acde232a22d6.png)](https:\u002F\u002Fwww.patreon.com\u002FbePatron?u=23238350)\n\n## One time Donations\nAlternatively you can give a one off donation to any of our Devs:\n### @torzdf\n There is very little FaceSwap code that hasn't been touched by torzdf. He is responsible for implementing the GUI, FAN aligner, MTCNN detector and porting the Villain, DFL-H128 and DFaker models to FaceSwap, as well as significantly improving many areas of the code.\n\n**Bitcoin:** bc1qpm22suz59ylzk0j7qk5e4c7cnkjmve2rmtrnc6\n\n**Ethereum:** 0xd3e954dC241B87C4E8E1A801ada485DC1d530F01\n\n**Monero:** 45dLrtQZ2pkHizBpt3P3yyJKkhcFHnhfNYPMSnz3yVEbdWm3Hj6Kr5TgmGAn3Far8LVaQf1th2n3DJVTRkfeB5ZkHxWozSX\n\n**Paypal:** [![torzdf](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_74f7d23b80fa.gif)](https:\u002F\u002Fwww.paypal.com\u002Fcgi-bin\u002Fwebscr?cmd=_s-xclick&hosted_button_id=JZ8PP3YE9J62L)\n\n### @andenixa\nCreator of the Unbalanced and OHR models, as well as expanding various capabilities within the training process. Andenixa is currently working on new models and will take requests for donations.\n\n**Paypal:** [![andenixa](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_74f7d23b80fa.gif)](https:\u002F\u002Fwww.paypal.com\u002Fcgi-bin\u002Fwebscr?cmd=_s-xclick&hosted_button_id=NRVLQYGS6NWTU)\n\n# How to contribute\n\n## For people interested in the generative models\n - Go to the 'faceswap-model' to discuss\u002Fsuggest\u002Fcommit alternatives to the current algorithm.\n\n## For devs\n - Read this README entirely\n - Fork the repo\n - Play with it\n - Check issues with the 'dev' tag\n - For devs more interested in computer vision and openCV, look at issues with the 'opencv' tag. Also feel free to add your own alternatives\u002Fimprovements\n\n## For non-dev advanced users\n - Read this README entirely\n - Clone the repo\n - Play with it\n - Check issues with the 'advuser' tag\n - Also go to the '[faceswap Forum](https:\u002F\u002Ffaceswap.dev\u002Fforum)' and help others.\n\n## For end-users\n - Get the code here and play with it if you can\n - You can also go to the [faceswap Forum](https:\u002F\u002Ffaceswap.dev\u002Fforum) and help or get help from others.\n - Be patient. This is a relatively new technology for developers as well. Much effort is already being put into making this program easy to use for the average user. It just takes time!\n - **Notice** Any issue related to running the code has to be opened in the [faceswap Forum](https:\u002F\u002Ffaceswap.dev\u002Fforum)!\n\n# About machine learning\n\n## How does a computer know how to recognize\u002Fshape faces? How does machine learning work? What is a neural network?\nIt's complicated. Here's a good video that makes the process understandable:\n[![How Machines Learn](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_dcd36ddb5b05.jpg)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=R9OHn5ZF4Uo)\n\nHere's a slightly more in depth video that tries to explain the basic functioning of a neural network:\n[![How Machines Learn](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_e802ba861962.jpg)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=aircAruvnKk)\n\ntl;dr: training data + trial and error\n","# deepfakes_faceswap\n\n### 致 **Patreon** 和 **PayPal** 支持者的重要信息。请参阅此论坛帖子：https:\u002F\u002Fforum.faceswap.dev\u002Fviewtopic.php?f=14&t=3120\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Ffaceswap.dev\">\u003Cimg src=\"https:\u002F\u002Fi.imgur.com\u002FzHvjHnb.png\">\u003C\u002Fimg>\u003C\u002Fa>\n\u003Cbr \u002F>FaceSwap 是一个利用深度学习 (Deep Learning) 来识别和交换图片及视频中人脸的工具。\n\u003C\u002Fp>\n\u003Cp align=\"center\">\n\u003Cimg src = \"https:\u002F\u002Fi.imgur.com\u002FnWHFLDf.jpg\">\u003C\u002Fimg>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n\u003Ca href=\"https:\u002F\u002Fwww.patreon.com\u002FbePatron?u=23238350\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_acde232a22d6.png\">\u003C\u002Fimg>\u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002FFC54sYg\">\u003Cimg src=\"https:\u002F\u002Fi.imgur.com\u002FgIpztkv.png\">\u003C\u002Fimg>\u003C\u002Fa>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fwww.dailymotion.com\u002Fvideo\u002Fx810mot\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_3e5a8b151398.png\">\u003C\u002Fimg>\u003C\u002Fa>\n\u003Cbr \u002F>使用 Phaze-A 模型进行的 Emma Stone\u002FScarlett Johansson 人脸交换\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=r1jng79a5xc\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_ab44b8f10460.jpg\">\u003C\u002Fimg>\u003C\u002Fa>\n\u003Cbr \u002F>使用 Villain 模型进行的 Jennifer Lawrence\u002FSteve Buscemi 人脸交换\n\u003C\u002Fp>\n\n\n![Build Status](https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Factions\u002Fworkflows\u002Fpytest.yml\u002Fbadge.svg) [![Documentation Status](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_13d664e1afd7.png)](https:\u002F\u002Ffaceswap.readthedocs.io\u002Fen\u002Flatest\u002F?badge=latest)\n\n在开始之前，请务必查看 [INSTALL.md](INSTALL.md)。\n\n- [deepfakes\\_faceswap](#deepfakes_faceswap)\n    - [致 **Patreon** 和 **PayPal** 支持者的重要信息。请参阅此论坛帖子：https:\u002F\u002Fforum.faceswap.dev\u002Fviewtopic.php?f=14\\&t=3120](#important-information-for-patreon-and-paypal-supporters-please-see-this-forum-post-httpsforumfaceswapdevviewtopicphpf14t3120)\n- [宣言](#manifesto)\n  - [FaceSwap 具有符合伦理的用途。](#faceswap-has-ethical-uses)\n- [如何设置和运行项目](#how-to-setup-and-run-the-project)\n- [概述](#overview)\n  - [提取](#extract)\n  - [训练](#train)\n  - [转换](#convert)\n  - [图形用户界面](#gui)\n- [一般说明：](#general-notes)\n- [帮助，我需要支持！](#help-i-need-support)\n  - [Discord 服务器](#discord-server)\n  - [FaceSwap 论坛](#faceswap-forum)\n- [捐赠](#donate)\n  - [Patreon](#patreon)\n  - [一次性捐赠](#one-time-donations)\n    - [@torzdf](#torzdf)\n    - [@andenixa](#andenixa)\n- [如何贡献](#how-to-contribute)\n  - [对于对生成模型 (Generative Models) 感兴趣的人](#for-people-interested-in-the-generative-models)\n  - [对于开发者](#for-devs)\n  - [对于非开发者的进阶用户](#for-non-dev-advanced-users)\n  - [对于最终用户](#for-end-users)\n- [关于机器学习](#about-machine-learning)\n  - [计算机如何知道如何识别\u002F塑造人脸？机器学习如何工作？什么是神经网络 (Neural Network)？](#how-does-a-computer-know-how-to-recognizeshape-faces-how-does-machine-learning-work-what-is-a-neural-network)\n\n# 宣言\n\n## FaceSwap 具有符合伦理的用途。\n\n当人脸交换技术最初被开发和发布时，这项技术是突破性的，它是人工智能 (AI) 发展的一大步。最初它在学术界之外被完全忽视，因为代码令人困惑且支离破碎。它需要对复杂的人工智能 (AI) 技术有透彻的理解，并且需要花费大量精力才能弄清楚。直到一个人将其整合成一个单一、连贯的集合。它可以运行，它可以工作，正如互联网上出现的新技术经常发生的那样，它立即被用于创建不当内容。尽管该软件最初被用于不当用途，但它是任何人都可以下载、运行并通过实验学习的第一段人工智能 (AI) 代码，而无需拥有数学、计算机理论、心理学等方面的博士学位 (Ph.D.)。在\"deepfakes\"之前，这些技术就像黑魔法，只有那些能够理解所有内部运作的人才能实践，正如那些深奥且无限复杂的书籍和论文中所描述的那样。\n\n\"Deepfakes\"改变了这一切，任何人都可以参与人工智能 (AI) 开发。对我们开发者来说，这段代码的发布开启了一个绝佳的学习机会。它允许我们建立在他人开发的想法之上，与各种熟练的编码者合作，在学习新技能的同时实验人工智能 (AI)，并最终为一项新兴技术做出贡献，随着技术的发展，这项技术只会看到更多的主流应用。\n\n是否有人在使用类似软件做可怕的事情？是的。正因为如此，开发人员一直在遵循严格的道德标准。我们中的许多人甚至不使用它来创建视频，我们只是摆弄代码看看它能做什么。遗憾的是，媒体只关注该软件的不道德用途。不幸的是，这是它最初向公众曝光的性质，但这并不代表它被创建的原因、我们现在如何使用它，或者我们在其未来中看到什么。像任何技术一样，它可以用于善，也可以被滥用。我们的意图是以一种最小化其滥用潜力，同时最大化其作为学习、实验工具以及是的，用于合法人脸交换的潜力的方式来开发 FaceSwap。\n\n我们并不是要贬低名人或侮辱任何人。我们是程序员，我们是工程师，我们是好莱坞视觉特效 (VFX) 艺术家，我们是活动家，我们是爱好者，我们是人类。为此，我们认为现在是时候发表一份标准声明，说明就我们开发者而言，这个软件是什么以及不是什么。\n\n- FaceSwap 不是用于创建不当内容。\n- FaceSwap 不是用于在未经同意的情况下或意图隐藏其使用的情况下更换人脸。\n- FaceSwap 不是用于任何非法、不道德或可疑的目的。\n- FaceSwap 的存在是为了实验和发现人工智能 (AI) 技术，用于社会或政治评论，用于电影，以及用于任何数量的道德和合理用途。\n\n我们对 FaceSwap 可用于不道德和声名狼藉的事情感到非常不安。然而，我们支持开发可以合乎道德地使用的工具和技术，并为任何想要动手学习的人提供人工智能 (AI) 教育和经验。我们将对任何将此软件用于任何不道德目的的人采取零容忍态度，并将积极劝阻任何此类用途。\n\n# 如何设置和运行项目\nFaceSwap 是一个 Python 程序，可运行于多种操作系统 (Operating Systems)，包括 Windows、Linux 和 MacOS。\n\n请参阅 [INSTALL.md](INSTALL.md) 获取完整的安装说明。为了获得最佳性能，您需要一个支持 CUDA 的现代图形处理器 (GPU)。许多 AMD 图形处理器 (GPU) 通过 ROCm (Linux) 得到支持。\n\n# 概述\n该项目有多个入口点。你需要：\n - 收集照片和\u002F或视频\n - 从原始照片中**提取 (Extract)** 人脸\n - 在从照片\u002F视频中提取的人脸上**训练 (Train)** 模型 (model)\n - 使用模型 (model)**转换 (Convert)** 你的源文件\n\n查看 [USAGE.md](USAGE.md) 获取更详细的说明。\n\n## 提取 (Extract)\n在你的设置文件夹中，运行 `python faceswap.py extract`。这将从 `src` 文件夹获取照片并将人脸提取到 `extract` 文件夹中。\n\n## 训练 (Train)\n在你的设置文件夹中，运行 `python faceswap.py train`。这将从两个包含两张人脸照片的文件夹中获取图片，并训练一个模型 (model)，该模型将保存在 `models` 文件夹内。\n\n## 转换 (Convert)\n在你的设置文件夹中，运行 `python faceswap.py convert`。这将从 `original` 文件夹获取照片并将新人脸应用到 `modified` 文件夹中。\n\n## 图形用户界面 (GUI)\n或者，你可以通过运行 `python faceswap.py gui` 来启动图形用户界面 (GUI)。\n\n# 一般注意事项：\n- 所有提到的脚本都有 `-h`\u002F`--help` 选项，显示它们接受的参数。你很聪明，你能弄明白这是怎么工作的，对吧？！\n\n注意 (NB)：有一个用于视频的工具。可以通过运行 `python tools.py effmpeg -h` 访问。或者，你可以使用 [ffmpeg](https:\u002F\u002Fwww.ffmpeg.org) 将视频转换为照片，处理图像，然后再将图像转换回视频。\n\n\n**一些提示：**\n\n重用现有模型 (model) 的训练速度比从头开始快得多。\n如果没有足够的训练数据，先从看起来相似的人开始，然后切换数据。\n\n# 帮助，我需要支持！\n## Discord 服务器\n最好的选择是加入 [FaceSwap Discord 服务器](https:\u002F\u002Fdiscord.gg\u002FFC54sYg)，那里有很多愿意帮助的用户。请注意，像这个仓库 (repo) 一样，这是一个适合工作场所 (SFW) 的服务器！\n\n## FaceSwap 论坛\n或者，你可以在 [FaceSwap 论坛](https:\u002F\u002Ffaceswap.dev\u002Fforum) 发帖提问。请不要在这个仓库 (repo) 中发布一般支持问题，因为它们可能会被删除且不予回复。\n\n# 捐赠\n开发人员 (Dev) 不知疲倦地改进和开发 FaceSwap。为了提供如今的软件，他们投入了许多小时，但这是一个极其耗时且没有经济回报的过程。如果你喜欢使用该软件，请考虑向开发人员 (Dev) 捐赠，以便他们能花更多时间实施改进。\n\n## Patreon\n支持我们的最佳方式是通过我们的 Patreon 页面：\n\n[![become-a-patron](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_acde232a22d6.png)](https:\u002F\u002Fwww.patreon.com\u002FbePatron?u=23238350)\n\n## 一次性捐赠\n或者，你可以向我们的任何开发人员 (Dev) 进行一次性的捐赠：\n### @torzdf\n 很少有 FaceSwap 代码未经过 torzdf 的处理。他负责实现图形用户界面 (GUI)、FAN 对齐器 (FAN aligner)、MTCNN 检测器 (MTCNN detector)，并将 Villain、DFL-H128 和 DFaker 模型 (model) 移植到 FaceSwap，以及显著改进了代码的许多领域。\n\n**Bitcoin:** bc1qpm22suz59ylzk0j7qk5e4c7cnkjmve2rmtrnc6\n\n**Ethereum:** 0xd3e954dC241B87C4E8E1A801ada485DC1d530F01\n\n**Monero:** 45dLrtQZ2pkHizBpt3P3yyJKkhcFHnhfNYPMSnz3yVEbdWm3Hj6Kr5TgmGAn3Far8LVaQf1th2n3DJVTRkfeB5ZkHxWozSX\n\n**Paypal:** [![torzdf](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_74f7d23b80fa.gif)](https:\u002F\u002Fwww.paypal.com\u002Fcgi-bin\u002Fwebscr?cmd=_s-xclick&hosted_button_id=JZ8PP3YE9J62L)\n\n### @andenixa\nUnbalanced 和 OHR 模型 (model) 的创建者，并扩展了训练过程中的各种功能。Andenixa 目前正在研究新模型 (model)，并将接受捐赠请求。\n\n**Paypal:** [![andenixa](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_74f7d23b80fa.gif)](https:\u002F\u002Fwww.paypal.com\u002Fcgi-bin\u002Fwebscr?cmd=_s-xclick&hosted_button_id=NRVLQYGS6NWTU)\n\n# 如何贡献\n\n## 对于对生成模型 (generative models) 感兴趣的人\n - 前往 'faceswap-model' 讨论\u002F建议\u002F提交当前算法 (algorithm) 的替代方案。\n\n## 对于开发人员 (Dev)\n - 完整阅读本 README\n - Fork (分叉) 这个仓库 (repo)\n - 尝试使用它\n - 查看带有 'dev' 标签的问题\n - 对于对计算机视觉 (computer vision) 和 openCV 更感兴趣的开发人员 (Dev)，查看带有 'opencv' 标签的问题。也欢迎添加你自己的替代方案\u002F改进\n\n## 对于非开发人员 (Dev) 的进阶用户\n - 完整阅读本 README\n - Clone (克隆) 这个仓库 (repo)\n - 尝试使用它\n - 查看带有 'advuser' 标签的问题\n - 同时前往 '[faceswap 论坛](https:\u002F\u002Ffaceswap.dev\u002Fforum)' 帮助他人。\n\n## 对于最终用户\n - 在这里获取代码并尝试使用它（如果可以的话）\n - 你也可以前往 [faceswap 论坛](https:\u002F\u002Ffaceswap.dev\u002Fforum) 帮助他人或获取帮助。\n - 要有耐心。这对开发人员 (Dev) 来说也是一项相对较新的技术。已经投入了大量精力使该程序易于普通用户使用。这只是需要时间！\n - **注意** 任何与运行代码相关的问题必须在 [faceswap 论坛](https:\u002F\u002Ffaceswap.dev\u002Fforum) 中提出！\n\n# 关于机器学习 (machine learning)\n\n## 计算机如何知道如何识别\u002F塑造人脸？机器学习 (machine learning) 如何工作？什么是神经网络 (neural network)？\n这很复杂。这里有一个很好的视频可以让过程易于理解：\n[![How Machines Learn](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_dcd36ddb5b05.jpg)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=R9OHn5ZF4Uo)\n\n这里有一个稍微更深入的视频，试图解释神经网络 (neural network) 的基本功能：\n[![How Machines Learn](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_readme_e802ba861962.jpg)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=aircAruvnKk)\n\n太长不看 (tl;dr)：训练数据 + 试错","# FaceSwap 快速上手指南\n\n## 简介\nFaceSwap 是一款利用深度学习技术识别并交换图片或视频中人脸的工具。本项目支持 Windows、Linux 和 MacOS 系统，旨在提供易于使用的 AI 换脸解决方案。\n\n## 环境准备\n\n在开始之前，请确保您的系统满足以下要求：\n\n- **操作系统**：Windows、Linux 或 MacOS。\n- **编程语言**：Python（项目为 Python 程序）。\n- **硬件要求**：\n  - **推荐**：支持 CUDA 的现代 GPU 以获得最佳性能。\n  - **支持**：许多 AMD GPU 可通过 ROCm（Linux）支持。\n- **依赖项**：完整的前置依赖列表请参阅项目根目录下的 `INSTALL.md` 文件。\n\n## 安装步骤\n\n由于安装过程涉及具体的依赖配置，请严格按照以下指引操作：\n\n1. 克隆或下载本项目到本地。\n2. 查阅项目根目录下的 `INSTALL.md` 文件获取完整的安装指令。\n3. 确保配置好 Python 环境及所需的 GPU 驱动（CUDA 或 ROCm）。\n\n> **注意**：在开始使用前，务必阅读 `INSTALL.md` 以确保环境配置正确。\n\n## 基本使用\n\n项目主要包含四个核心步骤：提取（Extract）、训练（Train）、转换（Convert）以及图形界面（GUI）。\n\n### 1. 准备工作\n收集包含目标人脸的照片或视频素材。\n\n### 2. 提取人脸 (Extract)\n从原始照片中提取人脸数据。\n```bash\npython faceswap.py extract\n```\n*说明：将从 `src` 文件夹读取照片，并将提取的人脸保存至 `extract` 文件夹。*\n\n### 3. 训练模型 (Train)\n使用提取的人脸数据训练模型。\n```bash\npython faceswap.py train\n```\n*说明：读取包含两张人脸照片的两个文件夹，训练模型并保存至 `models` 文件夹。*\n\n### 4. 转换人脸 (Convert)\n使用训练好的模型应用换脸效果。\n```bash\npython faceswap.py convert\n```\n*说明：从 `original` 文件夹读取照片，将换脸后的结果保存至 `modified` 文件夹。*\n\n### 5. 图形界面 (GUI)\n如果您更倾向于可视化操作，可以启动 GUI：\n```bash\npython faceswap.py gui\n```\n\n### 视频处理提示\n项目包含视频转换工具，可通过以下命令查看帮助：\n```bash\npython tools.py effmpeg -h\n```\n或者使用 [ffmpeg](https:\u002F\u002Fwww.ffmpeg.org) 将视频转换为图片，处理后再转回视频。\n\n## 注意事项\n\n- **伦理声明**：本软件仅用于实验、学习、社会评论、电影制作等合法合规用途。**严禁**用于制作不当内容、未经同意更换人脸或任何非法、不道德的目的。\n- **技术支持**：\n  - 遇到问题请优先访问 [FaceSwap 论坛](https:\u002F\u002Ffaceswap.dev\u002Fforum)。\n  - 也可加入 [Discord 服务器](https:\u002F\u002Fdiscord.gg\u002FFC54sYg) 寻求帮助（请注意这是一个 SFW 服务器）。\n  - **请勿**在本代码仓库中发布一般性支持问题，否则可能会被删除。\n- **训练技巧**：\n  - 复用现有模型比从头训练速度更快。\n  - 如果训练数据不足，可先使用相似人脸进行训练，再切换数据。","某独立电影制作团队在后期审查时，发现因主角档期冲突，部分关键镜头使用了替身演员，导致面部特征不连贯，严重影响观影体验。\n\n### 没有 faceswap 时\n- 传统 VFX 流程需要美术师逐帧手工绘制修饰，耗时数周甚至数月，严重拖慢上映进度\n- 外包给专业特效公司费用高昂，单次修改就可能严重超出独立制作的有限预算\n- 合成效果往往生硬，肤色过渡与光影变化难以自然融合，观众容易察觉违和感\n- 团队成员需精通 After Effects 等专业软件，技术门槛高且前期学习成本巨大\n\n### 使用 faceswap 后\n- faceswap 利用深度学习自动提取特征并训练模型，将数周的手工工作量压缩至几天内完成\n- faceswap 支持本地开源部署且无授权费用，显著降低制作成本，节省资金可用于其他环节\n- faceswap 生成结果逼真度高，算法自动匹配光源方向和表情细节，视觉过渡自然流畅\n- faceswap 提供友好的 GUI 界面和详细文档，普通后期人员经过简单培训即可上手操作\n\nfaceswap 通过自动化深度学习技术，让小型创作团队也能以低成本实现电影级的面部替换效果。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdeepfakes_faceswap_d038eec3.png","deepfakes",null,"https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fdeepfakes_32876187.jpg","https:\u002F\u002Fgithub.com\u002Fdeepfakes",[81,85,89],{"name":82,"color":83,"percentage":84},"Python","#3572A5",98.7,{"name":86,"color":87,"percentage":88},"Shell","#89e051",0.8,{"name":90,"color":77,"percentage":91},"NSIS",0.4,55119,13401,"2026-04-05T11:19:05","GPL-3.0","Windows, Linux, macOS","需要支持 CUDA 的现代 GPU 以获得最佳性能，Linux 下支持部分 AMD GPU (ROCm)，具体显存和 CUDA 版本未说明","未说明",{"notes":100,"python":98,"dependencies":101},"完整安装步骤请参阅 INSTALL.md 文件。项目强调道德使用，禁止用于非法或不道德目的。技术支持可通过 Discord 服务器或官方论坛获取。",[102],"ffmpeg (可选，用于视频处理)",[13,14],[67,105,106,107,108,76,109,110,111,112,113,114,115,116],"face-swap","deep-learning","deeplearning","deep-neural-networks","deepface","deep-face-swap","fakeapp","neural-networks","neural-nets","openfaceswap","myfakeapp","machine-learning",140,"2026-03-27T02:49:30.150509","2026-04-06T06:52:14.057133",[121,126,130,135,140,144],{"id":122,"question_zh":123,"answer_zh":124,"source_url":125},4752,"转换速度慢且 GPU 使用率为 0% 是正常的吗？","不正常。通常是因为启用了 \"seamless\" 模式，该模式会导致 GPU 利用率低且转换极慢。禁用 \"seamless\" 后速度可显著提升（例如从 1.8 it\u002Fs 提升至 5.5 it\u002Fs）。此外，确保 dlib 编译时启用了 CUDA 支持。","https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Fissues\u002F184",{"id":127,"question_zh":128,"answer_zh":129,"source_url":125},4753,"如何正确编译支持 CUDA 的 dlib？","需要自行编译 dlib 并启用 CUDA 支持。建议使用编译参数：`--yes USE_AVX_INSTRUCTIONS --yes DLIB_USE_CUDA`。注意某些 setup.py 参数（如 -G）可能不被识别，需根据环境调整。编译成功后 face-recognition 的 CNN 功能才能正常工作。",{"id":131,"question_zh":132,"answer_zh":133,"source_url":134},4754,"使用 CNN 检测器时人脸 landmarks 不匹配或对齐错误怎么办？","这通常是因为 landmarks 检测错误导致的缩放问题。建议改用 `face_alignment pytorch cnn` 代替 `face_recognition dlib cnn`。如果是代码层面问题，检查 `flip_input` 是否必须设为 False。","https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Fissues\u002F187",{"id":136,"question_zh":137,"answer_zh":138,"source_url":139},4755,"应该选择哪种模型分辨率（128\u002F256\u002F512）？","Original 128 适合大多数消费级 GPU。Orig256 和 Orig512 可能无法适应消费者 GPU 显存。如果需要高清换脸，建议使用已实现的 OriginalHiRes 模型。高分辨率模型可能需要牺牲部分颜色数据以保持内存和速度限制。","https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Fissues\u002F385",{"id":141,"question_zh":142,"answer_zh":143,"source_url":139},4756,"如何改善换脸后眼睛和瞳孔的自然度？","建议增加训练迭代次数，维护者建议运行至 300 epochs 以上可能会有帮助。此外，目前批处理大小（batch size）设为 16 已足够，未来可能需要添加眼部掩膜（mask for eyes）来专门处理眼睛区域。",{"id":145,"question_zh":146,"answer_zh":147,"source_url":148},4757,"如何高效筛选和清理提取的人脸图片？","可以使用内置的图像排序工具。按相似度排序命令：`python.exe faceswap\\sorttool.py -i %WORKSPACE%\\data_src\\aligned -by similarity`。按模糊度排序命令：`python.exe faceswap\\sorttool.py -i %WORKSPACE%\\data_src\\aligned -by blur`。排序后可轻松删除不需要的脸或模糊图片。","https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Fissues\u002F252",[150,155,160,165,170],{"id":151,"version":152,"summary_zh":153,"released_at":154},104291,"v3.0.0","**Installer for Faceswap.py**\r\n\r\nThis installer will install missing prerequisites (Git, MiniConda, PyTorch), set up the environment and create a Desktop shortcut for launching straight into the FaceSwap GUI. If running this installer, you do not need to manually install any programs yourself.\r\n\r\nFor **Nvidia** users it will install a local version if CUDA and cuDNN into a Conda environment. A CUDA 11.8+ capable graphics card is required.\r\n\r\nFor **AMD** Users The ROCm version of Torch will be installed, but you will need to ensure that you have ROCm 6.0-6.4 installed to enable use of AMD Graphics cards.\r\n\r\n# Windows Users\r\n**Note: Only Nvidia + CPU are supported under Windows. To use AMD cards, you will need to use Windows Subsystem for Linux (WSL2) and run the ROCm version of Faceswap.**\r\n\r\nYou only need to download the file [faceswap_setup_x64.exe](https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Freleases\u002Flatest\u002Fdownload\u002Ffaceswap_setup_x64.exe) from this page. You should **not** run the installer as an Administrator.\r\n\r\n**NB:** It may, at times, look like the installer has hung. It hasn't. The installer has to download and install a lot of prerequisites, and it is impossible to retrieve a live update of the output into the Details window. Please be patient.\r\n\r\n**IMPORTANT**: Windows SmartScreen may flag this program as unsafe. This is because it needs to download third party applications from the internet. It is entirely safe, and the source code can be reviewed here: https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Ftree\u002Fmaster\u002F.install\u002Fwindows If you are still not convinced, you can review the source code and compile the installer yourself.\r\n\r\n# Linux Users\r\nYou only need to download the file [faceswap_setup_x64.sh](https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Freleases\u002Flatest\u002Fdownload\u002Ffaceswap_setup_x64.sh) from this page.\r\n\r\nJust download and run as a normal user:\r\n```bash .\u002Ffaceswap_setup_x64.sh```\r\nThen follow the prompts.\r\n\r\n# Mac Users\r\nJust run the following command in `Terminal` to download and run the install script:\r\n```\r\nbash \u003C(curl -s https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Freleases\u002Flatest\u002Fdownload\u002Ffaceswap_setup_macos.sh)\r\n```\r\nThen follow the prompts\r\n\r\nIf the above doesn't work, then download the file  [faceswap_setup_macos.sh](https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Freleases\u002Flatest\u002Fdownload\u002Ffaceswap_setup_macos.sh) from this page then execute as a normal user:\r\n\r\n```bash .\u002Ffaceswap_setup_macos.sh```\r\n\r\n\r\n\r\n","2025-12-21T02:44:08",{"id":156,"version":157,"summary_zh":158,"released_at":159},104292,"v2.3.0","**Installer for Faceswap.py 2.3**\r\n\r\n**Note: This is the installer for final version of Faceswap using Tensorflow backend and with native support for AMD GPUs under Windows. This branch is no longer updated, but the installer is retained for legacy purposes. This version supports Cuda 11.2 and ROCm versions 5.2 and 5.4**\r\n\r\nThis installer will install missing prerequisites (Git, MiniConda, Tensorflow), set up the environment and create a Desktop shortcut for launching straight into the FaceSwap GUI. If running this installer, you do not need to manually install any programs yourself.\r\n\r\nFor **Nvidia** users it will install CUDA and cuDNN into a Conda environment. It is recommended to remove any system-wide installs of CUDA you have from your PC. A CUDA 9.0+ capable graphics card is required.\r\n\r\nFor **AMD** Users it will install PlaidML to enable use of AMD Graphics cards.\r\n\r\n# Windows Users\r\nYou only need to download the file [faceswap_setup_23_x64.exe](https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Freleases\u002Fdownload\u002Fv2.3.0\u002Ffaceswap_setup_23_x64.exe) from this page. You should **not** run the installer as an Administrator.\r\n\r\n**NB:** It may, at times, look like the installer has hung. It hasn't. The installer has to download and install a lot of prerequisites, and it is impossible to retrieve a live update of the output into the Details window. Please be patient.\r\n\r\n**IMPORTANT**: Windows SmartScreen may flag this program as unsafe. This is because it needs to download third party applications from the internet. It is entirely safe, and the source code can be reviewed here: https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Fblob\u002Fr2.3\u002F.install\u002Fwindows If you are still not convinced, you can review the source code and compile the installer yourself.\r\n\r\n# Linux Users\r\nYou only need to download the file [faceswap_setup_23_x64.sh](https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Freleases\u002Fdownload\u002Fv2.3.0\u002Ffaceswap_setup_23_x64.sh) from this page.\r\n\r\nJust download and run as a normal user:\r\n```bash .\u002Ffaceswap_setup_23_x64.sh```\r\nThen follow the prompts.\r\n\r\n# Mac Users\r\nJust run the following command in `Terminal` to download and run the install script:\r\n```\r\nbash \u003C(curl -s https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Freleases\u002Fdownload\u002Fv2.3.0\u002Ffaceswap_setup_23_macos.sh)\r\n```\r\nThen follow the prompts\r\n\r\nIf the above doesn't work, then download the file  [faceswap_setup_macos.sh](https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Freleases\u002Fdownload\u002Fv2.3.0\u002Ffaceswap_setup_23_macos.sh) from this page then execute as a normal user:\r\n\r\n```bash .\u002Ffaceswap_setup_23_macos.sh```\r\n\r\n\r\n\r\n","2025-12-19T23:13:33",{"id":161,"version":162,"summary_zh":163,"released_at":164},104293,"v2.2.0","**NB: This is the installer for final version of Faceswap with PlaidML support for AMD GPUs. This branch is no longer updated, but the installer is retained for legacy purposes.**\r\n\r\n**Installer for Faceswap.py**\r\n\r\nThis installer will install missing prerequisites (Git, MiniConda, Tensorflow), set up the environment and create a Desktop shortcut for launching straight into the FaceSwap GUI. If running this installer, you do not need to manually install any programs yourself.\r\n\r\nFor **Nvidia** users it will install CUDA and cuDNN into a Conda environment. It is recommended to remove any system-wide installs of CUDA you have from your PC. A CUDA 9.0+ capable graphics card is required.\r\n\r\nFor **AMD** Users it will install PlaidML to enable use of AMD Graphics cards.\r\n\r\n# Windows Users\r\nYou only need to download the file [faceswap_setup_x64.exe](https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Freleases\u002Fdownload\u002Fv2.2.0\u002Ffaceswap_setup_x64.exe) from this page. You should **not** run the installer as an Administrator.\r\n\r\n**NB:** It may, at times, look like the installer has hung. It hasn't. The installer has to download and install a lot of prerequisites, and it is impossible to retrieve a live update of the output into the Details window. Please be patient.\r\n\r\n**IMPORTANT**: Windows SmartScreen may flag this program as unsafe. This is because it needs to download third party applications from the internet. It is entirely safe, and the source code can be reviewed here: https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Ftree\u002Fmaster\u002F.install\u002Fwindows If you are still not convinced, you can review the source code and compile the installer yourself.\r\n\r\n# Linux Users\r\nYou only need to download the file [faceswap_setup_x64.sh](https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Freleases\u002Fdownload\u002Fv2.2.0\u002Ffaceswap_setup_x64.sh) from this page.\r\n\r\nJust download and run as a normal user:\r\n```bash .\u002Ffaceswap_setup_x64.sh```\r\nThen follow the prompts.\r\n","2023-06-18T18:08:19",{"id":166,"version":167,"summary_zh":168,"released_at":169},104294,"v1.0.0","**NB: This is the install for the Tensorflow 1.x release of Faceswap. This branch is no longer updated, but the installer is retained for legacy purposes.**\r\n\r\n**Installer for Faceswap.py**\r\n\r\nThis installer will install missing prerequisites (Git, MiniConda, Tensorflow), set up the environment and create a Desktop shortcut for launching straight into the FaceSwap GUI. If running this installer, you do not need to manually install any programs yourself.\r\n\r\nFor **Nvidia** users it will install CUDA and cuDNN into a Conda environment. It is recommended to remove any system-wide installs of CUDA you have from your PC. A CUDA 9.0+ capable graphics card is required.\r\n\r\nFor **AMD** Users it will install PlaidML to enable use of AMD Graphics cards.\r\n\r\n# Windows Users\r\nYou only need to download the file [faceswap_v1.0_setup_x64.exe](https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Freleases\u002Fdownload\u002Fv1.0.0\u002Ffaceswap_v1.0_setup_x64.exe) from this page.\r\n\r\n**NB:** It may, at times, look like the installer has hung. It hasn't. The installer has to download and install a lot of prerequisites, and it is impossible to retrieve a live update of the output into the Details window. Please be patient.\r\n\r\n**IMPORTANT**: Windows SmartScreen may flag this program as unsafe. This is because it needs to download third party applications from the internet. It is entirely safe, and the source code can be reviewed here: https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Fblob\u002Fr1.0\u002F.install\u002Fwindows If you are still not convinced, you can review the source code and compile the installer yourself.\r\n\r\n# Linux Users\r\nYou only need to download the file [faceswap_v1.0_setup_x64.sh](https:\u002F\u002Fgithub.com\u002Fdeepfakes\u002Ffaceswap\u002Freleases\u002Fdownload\u002Fv1.0.0\u002Ffaceswap_v1.0_setup_x64.sh) from this page.\r\n\r\nJust download and run as a normal user:\r\n```bash .\u002Ffaceswap_v1.0_setup_x64.sh```\r\nThen follow the prompts.\r\n\r\n","2020-08-14T17:25:06",{"id":171,"version":172,"summary_zh":173,"released_at":174},104295,"v0.90","This is a snapshot of the repo prior to the refactor of training, to support old models.","2019-02-23T03:16:21"]