[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-TachibanaYoshino--AnimeGANv3":3,"tool-TachibanaYoshino--AnimeGANv3":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",140436,2,"2026-04-05T23:32:43",[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":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":79,"owner_email":80,"owner_twitter":79,"owner_website":79,"owner_url":81,"languages":82,"stars":91,"forks":92,"last_commit_at":93,"license":79,"difficulty_score":23,"env_os":94,"env_gpu":95,"env_ram":96,"env_deps":97,"category_tags":105,"github_topics":106,"view_count":23,"oss_zip_url":79,"oss_zip_packed_at":79,"status":16,"created_at":116,"updated_at":117,"faqs":118,"releases":152},2940,"TachibanaYoshino\u002FAnimeGANv3","AnimeGANv3","Use AnimeGANv3 to make your own animation works, including turning photos or videos into anime.","AnimeGANv3 是一款强大的开源人工智能工具，旨在帮助用户轻松将普通照片或视频转化为风格独特的动画作品。它有效解决了传统图像风格化处理中常见的速度慢、画面闪烁或细节丢失等痛点，让用户能够快速生成高质量、连贯性强的动漫风格影像。\n\n无论是希望尝试创意玩法的普通用户，还是需要高效原型验证的设计师、开发者及研究人员，都能从中受益。普通用户可通过预置模型一键转换人像为吉卜力、皮克斯、迪士尼等多种流行画风；专业用户则可利用其开放的代码、训练数据及 ONNX、TensorRT、RKNN 等多平台推理支持，进行深度定制与性能优化。\n\n其核心技术基于一种新颖的“双尾生成对抗网络”架构，在保持快速推理的同时显著提升了画面稳定性与艺术表现力。项目持续更新，不仅新增了油画、素描、8 位像素等多种风格模型，还提供了专为肖像优化的高效推理仓库，兼顾易用性与扩展性。如果你渴望探索图像到动画的无限可能，AnimeGANv3 值得尝试。","# AnimeGANv3   \n\nPaper Title: A Novel Double-Tail Generative Adversarial Network for Fast Photo Animation.\n## Let's use AnimeGANv3 to produce our own animation.\n\n\u003Cdiv align=\"center\">    \n           \n[![manuscript](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fmanuscript-PDF-gold?logo=googledocs&logoColor=gold)](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3\u002Fblob\u002Fmaster\u002Fdoc\u002FAnimeGANv3_manuscript.pdf)\n[![Paper](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fcs.CV-Paper-violet?logo=docusign&logoColor=violet)](https:\u002F\u002Fwww.jstage.jst.go.jp\u002Farticle\u002Ftransinf\u002FE107.D\u002F1\u002FE107.D_2023EDP7061\u002F_pdf\u002F-char\u002Fen)\n[![Project Page](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FProject-Website-green?logo=googlechrome&logoColor=green)](https:\u002F\u002Ftachibanayoshino.github.io\u002FAnimeGANv3\u002F)\n[![HuggingFace](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Leaderboard-40D1F5)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002FTachibanaYoshino\u002FAnimeGANv3)\n[![Video](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FYouTube-Video-b31b1b?logo=youtube&logoColor=red)](https:\u002F\u002Fyoutu.be\u002FEosubeJmAnE)\n[![twitter](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ftwitter-Asher-1D9BF0?logo=x&logoColor=#1D9BF0)](https:\u002F\u002Ftwitter.com\u002Fasher_9527)\n[![LICENSE](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-AnimeGANv3-AB82FF?logo=leagueoflegends&logoColor=AB82FF)](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3?tab=readme-ov-file#scroll-license)\n[![Github](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FTachibanaYoshino\u002FAnimeGANv3?logo=githubsponsors&logoColor=#EA4AAA)](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3)\n[![Open In Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fdrive.google.com\u002Ffile\u002Fd\u002F1Er23bL36pkr67Q9f1P28BuMP6yZKf-yz\u002Fview?usp=sharing)\n![Badge](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_518264477dbd.png)\n\n\u003C\u002Fdiv>       \n\n## 📢 Updates    \n* `2025-08-22` Added a new AnimeGANv3 model for Face to Ghibli-c1 style. Its [model files](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3_Portrait_Inference\u002Freleases\u002Fdownload\u002F1.0\u002FAnimeGANv3_large_Ghibli_c1_e299.onnx) and [training data](https:\u002F\u002Fwww.patreon.com\u002Fposts\u002Freleased-ghibli-137135263?utm_medium=clipboard_copy&utm_source=copyLink&utm_campaign=postshare_creator&utm_content=join_link) have been open source. The training data of **Ghibli-o1** style can be accessed [here](https:\u002F\u002Fwww.patreon.com\u002Fposts\u002Freleased-ghibli-137019265?utm_medium=clipboard_copy&utm_source=copyLink&utm_campaign=postshare_creator&utm_content=join_link).     \n* `2025-02-16` Released the **TensorRT**-based model conversion and inference [repo](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3_Portrait_Inference\u002Ftree\u002FtensorRT).\n* `2024-12-15` Released the **rknn**-based model conversion and inference [repo](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3_Portrait_Inference\u002Ftree\u002Frknn). \n* `2024-11-27` Disney and Trump styles updated to version 2.0.    \n* `2024-10-31` Added a new styles of AnimeGANv3: Portrait to Pixar. :jack_o_lantern:    \n* `2024-08-28` A [repo](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3_Portrait_Inference) more suitable for portrait style inference based on the AnimeGANv3 models has been released. Highly recommended.🔥  \n* `2023-12-10` Added a new AnimeGANv3 model for Portrait to Oil-painting style. Its onnx is available [here](https:\u002F\u002Fwww.patreon.com\u002Fposts\u002Fanimeganv3-s-oil-94445425?utm_medium=clipboard_copy&utm_source=copyLink&utm_campaign=postshare_creator&utm_content=join_link).     \n* `2023-11-23` The code and the [manuscript](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3\u002Fblob\u002Fmaster\u002Fdoc\u002FAnimeGANv3_manuscript.pdf) are released.  🦃   \n* `2023-10-31` Added three new styles of AnimeGANv3: Portrait to Cute, 8bit and Sketch-0 style. :ghost:   \n* `2023-09-18` Added a new AnimeGANv3 model for Face to Kpop style.     \n* `2023-01-16` Added a new AnimeGANv3-photo.exe for the inference of AnimeGANv3's onnx model.     \n* `2023-01-13` Added a new AnimeGANv3 model for Face to comic style.     \n* `2022-12-25` Added the tiny model (2.4MB) of [~~Nordic myth style~~]() and USA style 2.0. It can go upto 50 FPS on iphone14 with 512*512 input. :santa:       \n* `2022-11-24` ~~Added a new AnimeGANv3 model for Face to Nordic myth style.~~  🦃      \n* `2022-11-06` Added a new AnimeGANv3 model for Face to Disney style **V1.0**.         \n* `2022-10-31` Added a new AnimeGANv3 model for Face to USA cartoon and Disney style **V1.0**.  :jack_o_lantern:    \n* `2022-10-07` The USA cartoon Style of AnimeGANv3 is integrated to [**ProfileProfile**](https:\u002F\u002Fapps.apple.com\u002Fin\u002Fapp\u002Fprofileprofile\u002Fid1636884362\n) with [Core ML](https:\u002F\u002Fdeveloper.apple.com\u002Fdocumentation\u002Fcoreml). Install it by the Apple Store and have a try.        \n* `2022-09-26` [Official online demo](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002FTachibanaYoshino\u002FAnimeGANv3) is integrated to [Huggingface Spaces](https:\u002F\u002Fhuggingface.co\u002Fspaces) with [Gradio](https:\u002F\u002Fgithub.com\u002Fgradio-app\u002Fgradio). [![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Spaces-blue)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002FTachibanaYoshino\u002FAnimeGANv3)     \n* `2022-09-24` Added a new great AnimeGANv3 model for Face to USA cartoon Style.    \n* `2022-09-18` Update a new AnimeGANv3 model for Photo to Hayao Style.    \n* `2022-08-01` Added a new AnimeGANv3 onnx model [**(Colab)**](https:\u002F\u002Fwww.patreon.com\u002Fposts\u002Fnew-animeganv3-69895469?utm_medium=clipboard_copy&utm_source=copyLink&utm_campaign=postshare_creator) for Face to [Arcane](https:\u002F\u002Fwww.netflix.com\u002Fsg\u002Ftitle\u002F81435684) style.    \n* `2022-07-13` Added a new AnimeGANv3 onnx model [**(Colab)**](https:\u002F\u002Fcolab.research.google.com\u002Fdrive\u002F1XYNWwM8Xq-U7KaTOqNap6A-Yq1f-V-FB?usp=sharing) for Face to portrait sketch.\n* `2021-12-25` The paper of AnimeGANv3 will be released in 2022.  :christmas_tree:  \n---------  \n\n## 🎮 Usage\n       \n* Official online demo is released in [![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Spaces-blue)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002FTachibanaYoshino\u002FAnimeGANv3).      \n\n* Download this repository and use AnimeGANv3's [UI tool](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3_gui.exe) and pre-trained *.onnx to turn your photos into anime. :blush:    \n\n* 🛠️ Installation\n  1. Clone repo  \n      ```bash  \n      git clone https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3.git\n      cd AnimeGANv3   \n      ```\n  \n  1. Install dependent packages\n      ```bash\n      pip install -r requirements.txt  \n      ```\n  1. Inference with *.onnx\n      ```bash\n      python deploy\u002Ftest_by_onnx.py -i inputs\u002Fimgs\u002F -o output\u002Fresults -m deploy\u002FAnimeGANv3_Hayao_36.onnx  \n      ```\n  1. video to anime with *.onnx\n      ```bash\n      python tools\u002Fvideo2anime.py -i inputs\u002Fvid\u002F1.mp4 -o output\u002Fresults -m deploy\u002FAnimeGANv3_Hayao_36.onnx  \n      ```\n\u003Cbr\u002F>    \n\n## 🚀 Landscape Demos     \n### :fire: Video to anime (Hayao Style)   \n\u003Cp>\n\u003Ca href=\"https:\u002F\u002Fyoutu.be\u002FEosubeJmAnE\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FYouTube-video1-red?logo=youtube&logoColor=red\"\u002F>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fyoutu.be\u002F5qLUflWb45E\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FYouTube-video2-green?logo=youtube&logoColor=red\"\u002F>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=iFjiaPlhVm4\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FYouTube-video3-pink?logo=youtube&logoColor=red\"\u002F>\u003C\u002Fa>\n\u003C\u002Fp>   \n\n____     \n\n### :art: Photo to Hayao Style    \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_f5ca646d508d.jpg)      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_750840353289.jpg)   \n\n\u003Cdetails>\n\u003Csummary>\u003Cstrong>   more surprise\u003C\u002Fstrong> 👈\u003C\u002Fsummary>    \n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_b3012f2e8549.jpg)   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_9b659256f39f.jpg)   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_2919a497a0ac.jpg)   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_3f06cb57918b.jpg)   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_4360ea70d6bd.jpg)   \n\u003C\u002Fdetails>    \n\n___   \n\u003Cbr\u002F>   \n\n### :art: Photo to Shinkai Style \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_41fa24a60cc2.jpg)  \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_9631550885dc.jpg)  \n     \n\u003Cdetails>\n\u003Csummary>\u003Cstrong>   more surprise\u003C\u002Fstrong> 👈 \u003C\u002Fsummary>    \n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_1dc40922d364.jpg)   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_69af5b649cf9.jpg)   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_13a4f2b5dbb4.jpg)  \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_db859de3315e.jpg)  \n\u003C\u002Fdetails>   \n\n___\n\u003Cbr\u002F>   \n\n## 🚀 Portrait Style Demos     \n**The paper has been completed in 2022. The study of portrait stylization is an extension of the paper.**     \n\n\u003Cdetails>\n\u003Csummary>\u003Cstrong>   Some exhibits \u003C\u002Fstrong> 👈\u003C\u002Fsummary>   \n       \n### :art: Face to USA cartoon style     \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_fde39e25d290.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_41f3808c1f8e.jpg)    \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F9644b1f5-78a4-4dcd-9da0-0186fbf5ab94\n\n___    \n### :art: Face to Disney cartoon style     \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_bb0234fbb65d.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_dfcc3ca8b47c.jpg)  \n\n| v1.9 | v2.0 |\n|:-:|:-:| \n|\u003Cvideo  src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F9cc111e9-8a1d-4c22-b0d0-0c430aca98d5\" type=\"video\u002Fmp4\"> \u003C\u002Fvideo>|\u003Cvideo  src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F53f927cd-bf2b-4699-a02a-8e635ff0403c\" type=\"video\u002Fmp4\"> \u003C\u002Fvideo>| \n___    \n\n### :art: Face to Ghibli-c1 style   \n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_efbdee7dd8eb.png) \n___   \n\n### :art: Face to Trump style    \n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_2d36d1d4671f.jpg)     \n\n| v1.9 | v2.0 |\n|:-:|:-:| \n|\u003Cvideo  src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fc20b4d99-125a-45c8-a1fe-c0c46a714b55\" type=\"video\u002Fmp4\"> \u003C\u002Fvideo>|\u003Cvideo  src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F6231a36d-1418-4175-adb5-c415adb8784e\" type=\"video\u002Fmp4\"> \u003C\u002Fvideo>| \n         \n\u003Cdetails>\n\u003Csummary>\u003Cstrong>   more surprise\u003C\u002Fstrong> 👈\u003C\u002Fsummary>\n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_15b246ddef33.gif)   \n \n\u003C\u002Fdetails>   \n\n___    \n\n### :art: Face to Arcane style   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_20d253ec2ec5.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_62c6ad7bf51d.jpg)   \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fab082d32-cc77-4c89-92c1-6a50cfa6a77b\n\n___    \n### :art: Portrait to comic style   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_9a0c9ae0e242.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_c3e700cdd2ce.jpg)    \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F3e999a8e-a331-46f6-863c-c01fd50591c8   \n\n___    \n### :art: Face to Kpop style   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_87cd475350c9.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_145c79d0a43b.jpg)  \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F3a59537c-fff2-4c86-8462-d53b07ff596b\n\n___    \n### :art: Portrait to Oil-painting style   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_f2fce433e003.gif)     \n\n\u003Cdetails>\n\u003Csummary>\u003Cstrong>   more surprise\u003C\u002Fstrong> 👈 \u003C\u002Fsummary>\n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_85dd1c67faf2.jpg)     \n\u003C\u002Fdetails>  \n\n___    \n### :art: Portrait to Cute style   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_61a3b8ea4437.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_cfe4feafd52e.jpg)    \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F0b105ee7-8116-4456-931c-ec196200e288  \n___  \n### :art: Portrait to Pixar style   \n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_c46d8c1e00d7.png) \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fd9c4e931-3b3c-4b03-9531-63d9e391b4df\n\n___    \n  \n### :art: Portrait to Sketch-0 style    \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_b53d5e902a82.jpg)  \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fed3f3511-4583-41d8-aad9-e47fdd2f5c32\n\n___  \n\n### :art: Portrait to 8bit style   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_f534f4e7a3f2.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_8625ea475f3c.jpg)    \n\n___\n### :art: Face to portrait sketch   \n[![Open In Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fdrive\u002F1XYNWwM8Xq-U7KaTOqNap6A-Yq1f-V-FB?usp=sharing)     \n      \n| input | Face | panoramic image|\n| :-: |:-:| :-:|\n|\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_fef2f617c61a.jpg\" height=\"60%\" width=\"60%\">|\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_1a9669c1d350.png\" height=\"60%\" width=\"60%\">|\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_87058dc8802c.png\" height=\"60%\" width=\"60%\">|\n|\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_261cf3768e03.jpg\" height=\"60%\" width=\"60%\">|\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_4d5f2adafaff.png\" height=\"60%\" width=\"60%\" >|\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_66fb0453d6d8.png\" height=\"60%\" width=\"60%\">|     \n    \n\u003Cdetails>\n\u003Csummary>\u003Cstrong>   more surprise\u003C\u002Fstrong> 👈\u003C\u002Fsummary>     \n       \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_6ba54a14d365.jpg)    \n       \n\u003C\u002Fdetails>    \n\n\u003C\u002Fdetails>  \n\n\u003Cbr\u002F>   \n\n## 🔨 Train\n\n#### 1. Download dataset and pretrained vgg19   \n1. [vgg19](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGAN\u002Freleases\u002Fdownload\u002Fvgg16%2F19.npy\u002Fvgg19_no_fc.npy)   \n2. [Hayao dataset](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv2\u002Freleases\u002Fdownload\u002F1.0\u002FHayao.tar.gz)   \n3. [Shinkai dataset](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv2\u002Freleases\u002Fdownload\u002F1.0\u002FShinkai.tar.gz)   \n4. [photo dataset](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGAN\u002Freleases\u002Fdownload\u002Fdataset-1\u002Fdataset.zip)   \n\n#### 2. Do edge_smooth  \n```bash\n    cd tools && python edge_smooth.py --dataset Hayao --img_size 256\n  ```\n\n#### 3. Do superPixel\n```bash\n    cd tools && python visual_superPixel_seg_image.py\n  ```  \n\n#### 4. Train  \n```bash\n    python train.py --style_dataset Hayao --init_G_epoch 5 --epoch 100\n  ```  \n\n\u003Cbr\u002F>   \n\n## ✒️ Citation   \nConsider citing as below if you find this repository helpful to your project:   \n```bibtex\n@article{Liu2024dtgan,\n  title={A Novel Double-Tail Generative Adversarial Network for Fast Photo Animation},\n  author={Gang LIU and Xin CHEN and Zhixiang GAO},\n  journal={IEICE Transactions on Information and Systems},\n  volume={E107.D},\n  number={1},\n  pages={72-82},\n  year={2024},\n  doi={10.1587\u002Ftransinf.2023EDP7061}\n}\n```\n\n## :scroll: License  \nThis repo is made freely available to academic and non-academic entities for non-commercial purposes such as academic research, teaching, scientific publications. Permission is granted to use the AnimeGANv3 given that you agree to my license terms. Regarding the request for commercial use, please contact us via email to help you obtain the authorization letter.    \n\n## :e-mail: Author  \nAsher Chan `asher_chan@foxmail.com`\n    \n","# AnimeGANv3   \n\n论文标题：一种用于快速照片动画的新型双尾生成对抗网络。\n## 让我们使用AnimeGANv3来制作属于自己的动画。\n\n\u003Cdiv align=\"center\">    \n           \n[![manuscript](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fmanuscript-PDF-gold?logo=googledocs&logoColor=gold)](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3\u002Fblob\u002Fmaster\u002Fdoc\u002FAnimeGANv3_manuscript.pdf)\n[![Paper](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fcs.CV-Paper-violet?logo=docusign&logoColor=violet)](https:\u002F\u002Fwww.jstage.jst.go.jp\u002Farticle\u002Ftransinf\u002FE107.D\u002F1\u002FE107.D_2023EDP7061\u002F_pdf\u002F-char\u002Fen)\n[![Project Page](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FProject-Website-green?logo=googlechrome&logoColor=green)](https:\u002F\u002Ftachibanayoshino.github.io\u002FAnimeGANv3\u002F)\n[![HuggingFace](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20 Face-Leaderboard-40D1F5)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002FTachibanaYoshino\u002FAnimeGANv3)\n[![Video](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FYouTube-Video-b31b1b?logo=youtube&logoColor=red)](https:\u002F\u002Fyoutu.be\u002FEosubeJmAnE)\n[![twitter](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ftwitter-Asher-1D9BF0?logo=x&logoColor=#1D9BF0)](https:\u002F\u002Ftwitter.com\u002Fasher_9527)\n[![LICENSE](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-AnimeGANv3-AB82FF?logo=leagueoflegends&logoColor=AB82FF)](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3?tab=readme-ov-file#scroll-license)\n[![Github](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FTachibanaYoshino\u002FAnimeGANv3?logo=githubsponsors&logoColor=#EA4AAA)](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3)\n[![Open In Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fdrive.google.com\u002Ffile\u002Fd\u002F1Er23bL36pkr67Q9f1P28BuMP6yZKf-yz\u002Fview?usp=sharing)\n![Badge](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_518264477dbd.png)\n\n\u003C\u002Fdiv>       \n\n## 📢 更新    \n* `2025-08-22` 新增了面向吉卜力风格c1的人像转换模型。其[模型文件](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3_Portrait_Inference\u002Freleases\u002Fdownload\u002F1.0\u002FAnimeGANv3_large_Ghibli_c1_e299.onnx)和[训练数据](https:\u002F\u002Fwww.patreon.com\u002Fposts\u002Freleased-ghibli-137135263?utm_medium=clipboard_copy&utm_source=copyLink&utm_campaign=postshare_creator&utm_content=join_link)已开源。**Ghibli-o1**风格的训练数据可在此获取[这里](https:\u002F\u002Fwww.patreon.com\u002Fposts\u002Freleased-ghibli-137019265?utm_medium=clipboard_copy&utm_source=copyLink&utm_campaign=postshare_creator&utm_content=join_link)。     \n* `2025-02-16` 发布了基于**TensorRT**的模型转换与推理[仓库](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3_Portrait_Inference\u002Ftree\u002FtensorRT)。\n* `2024-12-15` 发布了基于**rknn**的模型转换与推理[仓库](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3_Portrait_Inference\u002Ftree\u002Frknn)。 \n* `2024-11-27` 迪士尼和特朗普风格更新至2.0版本。    \n* `2024-10-31` 新增了AnimeGANv3的一种新风格：人像转皮克斯风格。 :jack_o_lantern:    \n* `2024-08-28` 发布了一个更适合基于AnimeGANv3模型进行人像风格推理的[仓库](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3_Portrait_Inference)。强烈推荐。🔥  \n* `2023-12-10` 新增了一种人像转油画风格的AnimeGANv3模型。其onnx文件可在[这里](https:\u002F\u002Fwww.patreon.com\u002Fposts\u002Fanimeganv3-s-oil-94445425?utm_medium=clipboard_copy&utm_source=copyLink&utm_campaign=postshare_creator&utm_content=join_link)找到。     \n* `2023-11-23` 代码和[论文](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3\u002Fblob\u002Fmaster\u002Fdoc\u002FAnimeGANv3_manuscript.pdf)正式发布。 🦃   \n* `2023-10-31` 新增了三种AnimeGANv3的新风格：人像转可爱风、8bit风格和素描0风格。 :ghost:   \n* `2023-09-18` 新增了一种面部转Kpop风格的AnimeGANv3模型。     \n* `2023-01-16` 新增了一个用于推理AnimeGANv3 onnx模型的AnimeGANv3-photo.exe工具。     \n* `2023-01-13` 新增了一种面部转漫画风格的AnimeGANv3模型。     \n* `2022-12-25` 新增了[~~北欧神话风格~~]()的小型模型（2.4MB）以及美国风格2.0版。在iPhone 14上，使用512*512分辨率的输入时，最高可达50 FPS。 :santa:       \n* `2022-11-24` ~~新增了一种面部转北欧神话风格的AnimeGANv3模型。~~  🦃      \n* `2022-11-06` 新增了一种面部转迪士尼风格的AnimeGANv3模型**V1.0**。         \n* `2022-10-31` 新增了一种面部转美国卡通和迪士尼风格的AnimeGANv3模型**V1.0**。 :jack_o_lantern:    \n* `2022-10-07` AnimeGANv3的美国卡通风格已集成到[**ProfileProfile**](https:\u002F\u002Fapps.apple.com\u002Fin\u002Fapp\u002Fprofileprofile\u002Fid1636884362\n)中，并使用了[Core ML](https:\u002F\u002Fdeveloper.apple.com\u002Fdocumentation\u002Fcoreml)技术。请通过Apple Store下载并试用。        \n* `2022-09-26` [官方在线演示](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002FTachibanaYoshino\u002FAnimeGANv3)已通过[Gradio](https:\u002F\u002Fgithub.com\u002Fgradio-app\u002Fgradio)集成到[Huggingface Spaces](https:\u002F\u002Fhuggingface.co\u002Fspaces)平台中。[![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20 Face-Spaces-blue)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002FTachibanaYoshino\u002FAnimeGANv3)     \n* `2022-09-24` 新增了一种优秀的AnimeGANv3模型，用于将面部转换为美国卡通风格。    \n* `2022-09-18` 更新了一种新的AnimeGANv3模型，用于将照片转换为宫崎骏风格。    \n* `2022-08-01` 新增了一个AnimeGANv3 onnx模型[**(Colab)**](https:\u002F\u002Fwww.patreon.com\u002Fposts\u002Fnew-animeganv3-69895469?utm_medium=clipboard_copy&utm_source=copyLink&utm_campaign=postshare_creator)用于面部转[《英雄联盟：激斗峡谷》](https:\u002F\u002Fwww.netflix.com\u002Fsg\u002Ftitle\u002F81435684)风格。    \n* `2022-07-13` 新增了一个AnimeGANv3 onnx模型[**(Colab)**](https:\u002F\u002Fcolab.research.google.com\u002Fdrive\u002F1XYNWwM8Xq-U7KaTOqNap6A-Yq1f-V-FB?usp=sharing)用于面部转人像素描。    \n* `2021-12-25` AnimeGANv3的论文将于2022年发表。 :christmas_tree:  \n---------  \n\n## 🎮 使用方法\n       \n* 官方在线演示已在[![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20 Face-Spaces-blue)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002FTachibanaYoshino\u002FAnimeGANv3)发布。      \n\n* 下载此仓库，并使用AnimeGANv3的[UI工具](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3_gui.exe)和预训练的*.onnx文件，将你的照片变成动漫风格。 :blush:    \n\n* 🛠️ 安装步骤\n  1. 克隆仓库  \n      ```bash  \n      git clone https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3.git\n      cd AnimeGANv3   \n      ```\n  \n  1. 安装依赖包\n      ```bash\n      pip install -r requirements.txt  \n      ```\n  1. 使用*.onnx进行推理\n      ```bash\n      python deploy\u002Ftest_by_onnx.py -i inputs\u002Fimgs\u002F -o output\u002Fresults -m deploy\u002FAnimeGANv3_Hayao_36.onnx  \n      ```\n  1. 使用*.onnx将视频转换为动漫\n      ```bash\n      python tools\u002Fvideo2anime.py -i inputs\u002Fvid\u002F1.mp4 -o output\u002Fresults -m deploy\u002FAnimeGANv3_Hayao_36.onnx  \n      ```\n\u003Cbr\u002F>    \n\n## 🚀 风景图演示\n\n### :fire: 视频转宫崎骏风格动画   \n\u003Cp>\n\u003Ca href=\"https:\u002F\u002Fyoutu.be\u002FEosubeJmAnE\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FYouTube-video1-red?logo=youtube&logoColor=red\"\u002F>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fyoutu.be\u002F5qLUflWb45E\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FYouTube-video2-green?logo=youtube&logoColor=red\"\u002F>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=iFjiaPlhVm4\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FYouTube-video3-pink?logo=youtube&logoColor=red\"\u002F>\u003C\u002Fa>\n\u003C\u002Fp>   \n\n____     \n\n### :art: 照片转宫崎骏风格  \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_f5ca646d508d.jpg)      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_750840353289.jpg)   \n\n\u003Cdetails>\n\u003Csummary>\u003Cstrong>   更多惊喜\u003C\u002Fstrong> 👈\u003C\u002Fsummary>    \n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_b3012f2e8549.jpg)   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_9b659256f39f.jpg)   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_2919a497a0ac.jpg)   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_3f06cb57918b.jpg)   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_4360ea70d6bd.jpg)   \n\u003C\u002Fdetails>    \n\n___   \n\u003Cbr\u002F>   \n\n### :art: 照片转新海诚风格 \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_41fa24a60cc2.jpg)  \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_9631550885dc.jpg)  \n     \n\u003Cdetails>\n\u003Csummary>\u003Cstrong>   更多惊喜\u003C\u002Fstrong> 👈 \u003C\u002Fsummary>    \n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_1dc40922d364.jpg)   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_69af5b649cf9.jpg)   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_13a4f2b5dbb4.jpg)  \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_db859de3315e.jpg)  \n\u003C\u002Fdetails>   \n\n___\n\u003Cbr\u002F>   \n\n## 🚀 人像风格演示     \n**该论文已于2022年完成。人像风格化研究是该论文的延伸。**     \n\n\u003Cdetails>\n\u003Csummary>\u003Cstrong>   一些展示\u003C\u002Fstrong> 👈\u003C\u002Fsummary>   \n       \n### :art: 脸部转美国卡通风格     \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_fde39e25d290.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_41f3808c1f8e.jpg)    \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F9644b1f5-78a4-4dcd-9da0-0186fbf5ab94\n\n___    \n### :art: 脸部转迪士尼卡通风格     \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_bb0234fbb65d.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_dfcc3ca8b47c.jpg)  \n\n| v1.9 | v2.0 |\n|:-:|:-:| \n|\u003Cvideo  src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F9cc111e9-8a1d-4c22-b0d0-0c430aca98d5\" type=\"video\u002Fmp4\"> \u003C\u002Fvideo>|\u003Cvideo  src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F53f927cd-bf2b-4699-a02a-8e635ff0403c\" type=\"video\u002Fmp4\"> \u003C\u002Fvideo>| \n___    \n\n### :art: 脸部转吉卜力-c1风格   \n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_efbdee7dd8eb.png) \n___   \n\n### :art: 脸部转特朗普风格    \n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_2d36d1d4671f.jpg)     \n\n| v1.9 | v2.0 |\n|:-:|:-:| \n|\u003Cvideo  src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fc20b4d99-125a-45c8-a1fe-c0c46a714b55\" type=\"video\u002Fmp4\"> \u003C\u002Fvideo>|\u003Cvideo  src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F6231a36d-1418-4175-adb5-c415adb8784e\" type=\"video\u002Fmp4\"> \u003C\u002Fvideo>| \n         \n\u003Cdetails>\n\u003Csummary>\u003Cstrong>   更多惊喜\u003C\u002Fstrong> 👈\u003C\u002Fsummary>\n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_15b246ddef33.gif)   \n \n\u003C\u002Fdetails>   \n\n___    \n\n### :art: 脸部转《英雄联盟：双城之战》风格   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_20d253ec2ec5.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_62c6ad7bf51d.jpg)   \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fab082d32-cc77-4c89-92c1-6a50cfa6a77b\n\n___    \n### :art: 人像转漫画风格   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_9a0c9ae0e242.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_c3e700cdd2ce.jpg)    \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F3e999a8e-a331-46f6-863c-c01fd50591c8   \n\n___    \n### :art: 脸部转K-pop风格   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_87cd475350c9.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_145c79d0a43b.jpg)  \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F3a59537c-fff2-4c86-8462-d53b07ff596b\n\n___    \n### :art: 人像转油画风格   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_f2fce433e003.gif)     \n\n\u003Cdetails>\n\u003Csummary>\u003Cstrong>   更多惊喜\u003C\u002Fstrong> 👈 \u003C\u002Fsummary>\n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_85dd1c67faf2.jpg)     \n\u003C\u002Fdetails>  \n\n___    \n### :art: 人像转可爱风格   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_61a3b8ea4437.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_cfe4feafd52e.jpg)    \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F0b105ee7-8116-4456-931c-ec196200e288  \n___  \n### :art: 人像转皮克斯风格   \n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_c46d8c1e00d7.png) \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fd9c4e931-3b3c-4b03-9531-63d9e391b4df\n\n___    \n  \n### :art: 人像转素描-0风格    \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_b53d5e902a82.jpg)  \n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fed3f3511-4583-41d8-aad9-e47fdd2f5c32\n\n___  \n\n### :art: 人像转8bit风格   \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_f534f4e7a3f2.gif)     \n      \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_8625ea475f3c.jpg)    \n\n___\n\n### :art: 人脸转肖像素描   \n[![在 Colab 中打开](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fdrive\u002F1XYNWwM8Xq-U7KaTOqNap6A-Yq1f-V-FB?usp=sharing)     \n      \n| 输入 | 人脸 | 全景图|\n| :-: |:-:| :-:|\n|\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_fef2f617c61a.jpg\" height=\"60%\" width=\"60%\">|\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_1a9669c1d350.png\" height=\"60%\" width=\"60%\">|\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_87058dc8802c.png\" height=\"60%\" width=\"60%\">|\n|\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_261cf3768e03.jpg\" height=\"60%\" width=\"60%\">|\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_4d5f2adafaff.png\" height=\"60%\" width=\"60%\" >|\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_66fb0453d6d8.png\" height=\"60%\" width=\"60%\">|     \n    \n\u003Cdetails>\n\u003Csummary>\u003Cstrong>   更多惊喜\u003C\u002Fstrong> 👈\u003C\u002Fsummary>     \n       \n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_readme_6ba54a14d365.jpg)    \n       \n\u003C\u002Fdetails>    \n\n\u003C\u002Fdetails>  \n\n\u003Cbr\u002F>   \n\n## 🔨 训练\n\n#### 1. 下载数据集和预训练的 VGG19   \n1. [vgg19](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGAN\u002Freleases\u002Fdownload\u002Fvgg16%2F19.npy\u002Fvgg19_no_fc.npy)   \n2. [宫崎骏数据集](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv2\u002Freleases\u002Fdownload\u002F1.0\u002FHayao.tar.gz)   \n3. [新海诚数据集](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv2\u002Freleases\u002Fdownload\u002F1.0\u002FShinkai.tar.gz)   \n4. [照片数据集](https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGAN\u002Freleases\u002Fdownload\u002Fdataset-1\u002Fdataset.zip)   \n\n#### 2. 进行边缘平滑  \n```bash\n    cd tools && python edge_smooth.py --dataset Hayao --img_size 256\n  ```\n\n#### 3. 进行超像素分割  \n```bash\n    cd tools && python visual_superPixel_seg_image.py\n  ```  \n\n#### 4. 训练  \n```bash\n    python train.py --style_dataset Hayao --init_G_epoch 5 --epoch 100\n  ```  \n\n\u003Cbr\u002F>   \n\n## ✒️ 引用   \n如果您觉得本仓库对您的项目有所帮助，请参考以下引用格式：   \n```bibtex\n@article{Liu2024dtgan,\n  title={一种用于快速照片动画的新颖双尾生成对抗网络},\n  author={Gang LIU 和 Xin CHEN 和 Zhixiang GAO},\n  journal={IEICE 信息与系统汇刊},\n  volume={E107.D},\n  number={1},\n  pages={72-82},\n  year={2024},\n  doi={10.1587\u002Ftransinf.2023EDP7061}\n}\n```\n\n## :scroll: 许可协议  \n本仓库免费向学术及非学术机构开放，仅限于非商业用途，如学术研究、教学和科学出版。只要您同意本许可条款，即可使用 AnimeGANv3。如需商业用途，请通过电子邮件联系我们，以获取授权书。    \n\n## :e-mail: 作者  \nAsher Chan `asher_chan@foxmail.com`","# AnimeGANv3 快速上手指南\n\nAnimeGANv3 是一个基于双尾生成对抗网络（Double-Tail GAN）的开源项目，能够将真实照片或视频快速转换为多种动漫风格（如宫崎骏、新海诚、迪士尼、皮克斯等）。\n\n## 1. 环境准备\n\n在开始之前，请确保您的开发环境满足以下要求：\n\n*   **操作系统**：Linux, macOS 或 Windows\n*   **Python 版本**：推荐 Python 3.8 及以上版本\n*   **硬件建议**：虽然支持 CPU 推理，但若有 NVIDIA GPU 并安装了 CUDA，处理速度将显著提升。\n*   **前置依赖**：\n    *   Git（用于克隆代码库）\n    *   pip（Python 包管理工具）\n\n> **💡 国内加速建议**：\n> 建议使用国内镜像源安装依赖，以提升下载速度。\n> *   清华源：`https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple`\n> *   阿里源：`https:\u002F\u002Fmirrors.aliyun.com\u002Fpypi\u002Fsimple\u002F`\n\n## 2. 安装步骤\n\n### 第一步：克隆项目代码\n打开终端或命令行工具，执行以下命令获取源代码：\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3.git\ncd AnimeGANv3\n```\n\n### 第二步：安装依赖包\n使用 pip 安装项目所需的 Python 库。**推荐使用国内镜像源**：\n\n```bash\npip install -r requirements.txt -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\n```\n\n### 第三步：准备模型文件\n项目运行需要预训练的 `.onnx` 模型文件。\n*   您可以从项目的 Release 页面、HuggingFace 空间或作者提供的链接下载模型。\n*   将下载的模型文件（例如 `AnimeGANv3_Hayao_36.onnx`）放置在项目目录下的 `deploy\u002F` 文件夹中（或根据实际路径调整后续命令）。\n\n## 3. 基本使用\n\n安装完成后，您可以直接使用提供的脚本对图片或视频进行风格转换。以下是最常用的两个示例：\n\n### 场景一：图片转动漫风格\n将输入文件夹中的图片转换为指定风格（以宫崎骏风格为例）：\n\n```bash\npython deploy\u002Ftest_by_onnx.py -i inputs\u002Fimgs\u002F -o output\u002Fresults -m deploy\u002FAnimeGANv3_Hayao_36.onnx\n```\n\n*   `-i`: 输入图片所在的文件夹路径。\n*   `-o`: 输出结果保存的文件夹路径。\n*   `-m`: 使用的 `.onnx` 模型文件路径。\n\n### 场景二：视频转动漫风格\n将视频文件转换为动漫风格视频：\n\n```bash\npython tools\u002Fvideo2anime.py -i inputs\u002Fvid\u002F1.mp4 -o output\u002Fresults -m deploy\u002FAnimeGANv3_Hayao_36.onnx\n```\n\n*   `-i`: 输入视频文件路径。\n*   `-o`: 输出视频保存的文件夹路径。\n*   `-m`: 使用的 `.onnx` 模型文件路径。\n\n> **提示**：项目支持多种风格模型（如 Shinkai, Disney, Arcane, Trump 等），只需更换 `-m` 参数对应的模型文件即可体验不同效果。更多模型下载地址请参考项目 README 中的 \"Updates\" 章节或官方 HuggingFace 空间。","一位独立游戏开发者正在为一款复古风格的视觉小说制作角色立绘，需要将大量真人参考照片快速转化为统一的动画风格素材。\n\n### 没有 AnimeGANv3 时\n- **美术成本高昂**：聘请画师手绘每张立绘费用昂贵，且对于只有单人开发的项目来说预算难以承受。\n- **风格难以统一**：不同批次绘制或不同画师产出的线条与上色习惯存在差异，导致角色在场景中显得突兀。\n- **迭代效率低下**：当需要调整角色表情或发型时，必须重新委托画师修改，沟通成本高且等待周期长。\n- **视频素材缺失**：若想制作动态开场动画，将真人实拍视频转为动画帧几乎不可能手动完成，工作量巨大。\n\n### 使用 AnimeGANv3 后\n- **一键批量转换**：利用其支持的“宫崎骏”或“皮克斯”等多样式模型，开发者可将数百张真人照片瞬间转化为高质量动画立绘。\n- **风格高度一致**：基于同一训练模型生成的图像，在线条勾勒和色彩渲染上保持完美一致，确保游戏视觉体验流畅。\n- **实时快速迭代**：开发者可随时更换输入照片或切换风格模型（如从“可爱风”切至\"8 位像素风”），几分钟内即可预览新效果。\n- **动态内容生成**：借助其对视频的处理能力，轻松将真人短剧片段转换为动画过场，极大丰富了游戏的叙事表现力。\n\nAnimeGANv3 通过高效的风格迁移技术，让个人开发者也能以极低的成本和专业级的质量，独立完成从静态立绘到动态影像的动画化创作。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FTachibanaYoshino_AnimeGANv3_bb46d458.png","TachibanaYoshino","Asher Chan","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002FTachibanaYoshino_c1505889.jpg","AI Algorithm Engineer",null,"asher_chan@foxmail.com","https:\u002F\u002Fgithub.com\u002FTachibanaYoshino",[83,87],{"name":84,"color":85,"percentage":86},"Python","#3572A5",99.9,{"name":88,"color":89,"percentage":90},"Shell","#89e051",0.1,2001,262,"2026-03-18T16:00:37","Windows, Linux, macOS","非必需。支持 CPU 推理（ONNX 模型）；可选 NVIDIA GPU 加速或特定硬件加速（如 TensorRT, RKNN, Core ML for iOS）。具体显存和 CUDA 版本未在提供的片段中明确说明，但提到 Tiny 模型可在 iPhone 14 上以 50 FPS 运行。","未说明",{"notes":98,"python":99,"dependencies":100},"该工具主要提供预训练的 ONNX 模型用于推理，无需复杂的训练环境。官方推荐使用独立的 UI 工具 (AnimeGANv3_gui.exe) 或直接运行 Python 脚本加载 .onnx 模型。支持多种风格转换（如宫崎骏、新海诚、迪士尼等）。针对移动端和边缘设备提供了 TensorRT、RKNN 和 Core ML 的转换仓库。视频转动漫功能包含在工具集中。","未说明 (通常建议 Python 3.8+ 以兼容主流深度学习库)",[101,102,103,104],"onnxruntime","opencv-python","numpy","torch (若需训练或特定后端)",[13,14],[107,108,109,110,111,112,113,114,115],"animegan","animeganv2","animeganv3","tensorflow","onnx","colab","coreml","huggingface","tflite","2026-03-27T02:49:30.150509","2026-04-06T08:46:47.510073",[119,124,129,134,139,144,148],{"id":120,"question_zh":121,"answer_zh":122,"source_url":123},13580,"如何根据不同的 CUDA 版本正确安装 AnimeGANv3 的依赖环境？","请根据您的 CUDA 版本选择对应的安装命令：\n1. **CUDA 10**: 创建 Python 3.7 环境，安装 cudatoolkit=10.0.130, cudnn, tensorflow-gpu==1.15.0, tf2onnx==1.10.1 等。\n2. **CUDA 11**: 创建 Python 3.8 环境，安装 nvidia-pyindex, nvidia-tensorflow[horovod], tf2onnx==1.15.1, coremltools==7.0 等。\n3. **CUDA 12**: 创建 Python 3.8 环境，具体包版本需参考最新文档（通常涉及较新的 tensorflow 和 onnxruntime 版本）。\n建议使用 conda 创建独立虚拟环境以避免冲突。","https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3\u002Fissues\u002F52",{"id":125,"question_zh":126,"answer_zh":127,"source_url":128},13581,"为什么在 Ubuntu 上训练时报错 'cv2.error: (-215:Assertion failed) !_src.empty()'？","该错误通常表示 OpenCV 未能成功读取图像文件。主要原因包括：\n1. **图片路径错误**：请仔细检查代码中指定的数据集图片路径是否正确。\n2. **文件格式问题**：确保图片文件未损坏且格式受支持。\n3. **目录结构不符**：数据集的目录树结构需严格参照项目说明，特别是 AnimeGANv3_hayao.py 和 AnimeGANv3_shinkai.py 第 54 行的配置差异。若需生成 smooth_noise 数据，还需先运行 tools\u002Fget_saltNoise.py 脚本处理。","https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3\u002Fissues\u002F46",{"id":130,"question_zh":131,"answer_zh":132,"source_url":133},13582,"能否将 AnimeGANv3 模型从 FP32 量化为 INT8 以在 RK3588 等设备上运行？","目前 **rknn_toolkit2 不支持**直接对从 TensorFlow 导出的 ONNX 模型进行 INT8 量化。原因是 TensorFlow 及其衍生的 ONNX 输入维度为 NHWC，而 rknn.config() 中的 mean 和 std 参数仅支持 NCHW 维度，导致量化失败。\n**替代方案**：您可以关闭量化选项，直接将模型转换为 **FP16 格式的 RKNN 模型**进行推理。","https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3\u002Fissues\u002F58",{"id":135,"question_zh":136,"answer_zh":137,"source_url":138},13583,"如何将自训练的 Checkpoint 模型转换为 PB 或 ONNX 格式？Input 节点名称找不到怎么办？","在使用 animegan_Ckpt2pb.py 脚本转换时，如果默认的 input 节点 'AnimeGANv3_input' 无法找到，请查看模型中所有的操作名（op names）。许多用户发现实际的输入节点名为 **'val_input'**。将其指定为 input 参数即可完成从 Ckpt 到 PB 再到 ONNX 的转换。","https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3\u002Fissues\u002F45",{"id":140,"question_zh":141,"answer_zh":142,"source_url":143},13584,"在哪里可以获取预训练的 ONNX 模型文件或数据集？","作者未直接在仓库中提供所有预训练模型文件的公开下载链接。如果您需要用于测试的 ONNX 模型文件或希望获取用于训练的数据集（特别是用于商业许可扩展研究），建议直接通过邮件联系作者 Asher Chan 进行咨询。部分用户提到可以通过赞助（如 Patreon）获得更多信息交流机会。","https:\u002F\u002Fgithub.com\u002FTachibanaYoshino\u002FAnimeGANv3\u002Fissues\u002F2",{"id":145,"question_zh":146,"answer_zh":147,"source_url":143},13585,"视频处理后没有声音怎么办？","这是一个已知问题，原视频音频在处理过程中被丢弃。目前的解决方案通常需要手动合并音频：\n1. 使用 AnimeGANv3 处理视频得到无声的风格化视频。\n2. 使用 FFmpeg 等工具将原视频的音频轨道提取并合并回处理后的视频中。\n命令示例：`ffmpeg -i styled_video.mp4 -i original_video.mp4 -c:v copy -c:a aac -map 0:v:0 -map 1:a:0 output.mp4`（具体参数需根据实际文件调整）。",{"id":149,"question_zh":150,"answer_zh":151,"source_url":128},13586,"如何处理非正方形图片以避免变形？","默认脚本（如 `edge_Smooth.py`）可能会将非正方形图片压缩（resize）至指定尺寸（如 256x256），导致画面变形。为避免此问题，建议在预处理阶段先将图片裁剪（crop）为正方形，或者修改代码逻辑，使其执行中心裁剪而非强制缩放，以保持原始宽高比。",[153,158],{"id":154,"version":155,"summary_zh":156,"released_at":157},72411,"v1.1.0","对于 .tflite 文件，其输入形状为 (1, 512, 512, 3)。  \n对于 .mlmodel 文件，其输入是一张 512×512 像素的图像。","2023-11-23T06:06:02",{"id":159,"version":160,"summary_zh":161,"released_at":162},72412,"v1.0.3","旧版本","2023-11-23T05:50:53"]