[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-nerfstudio-project--nerfacc":3,"tool-nerfstudio-project--nerfacc":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":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":79,"owner_twitter":79,"owner_website":80,"owner_url":81,"languages":82,"stars":94,"forks":95,"last_commit_at":96,"license":97,"difficulty_score":10,"env_os":98,"env_gpu":99,"env_ram":100,"env_deps":101,"category_tags":105,"github_topics":106,"view_count":10,"oss_zip_url":79,"oss_zip_packed_at":79,"status":16,"created_at":111,"updated_at":112,"faqs":113,"releases":143},536,"nerfstudio-project\u002Fnerfacc","nerfacc","A General NeRF Acceleration Toolbox in PyTorch.","nerfacc 是一款基于 PyTorch 的通用神经辐射场（NeRF）加速工具箱。简单来说，它致力于让 NeRF 模型跑得更快、更省资源。针对传统 NeRF 训练和推理过程缓慢的痛点，nerfacc 通过优化体积渲染中的采样策略，实现了显著的性能提升。\n\n这款工具箱特别适合计算机视觉领域的研究人员和深度学习开发者。它最大的优势在于“即插即用”，采用纯 Python 接口，灵活性极高。你只需要编写简单的密度和颜色计算函数，就能把加速功能融入现有项目，几乎不需要改动原有代码。此外，nerfacc 还内置了高效的表面发现算法，能低成本地快速定位场景细节。\n\n无论你是想复现最新的学术成果，还是希望开发高效的三维重建应用，nerfacc 都能提供强有力的支持，帮助你在保证质量的同时大幅提升工作效率。","\u003Cp>\n  \u003C!-- pypi-strip -->\n  \u003Cpicture>\n  \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F3310961\u002F199083722-881a2372-62c1-4255-8521-31a95a721851.png\" \u002F>\n  \u003Csource media=\"(prefers-color-scheme: light)\" srcset=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fnerfstudio-project_nerfacc_readme_6af681907002.png\" \u002F>\n  \u003C!-- \u002Fpypi-strip -->\n  \u003Cimg alt=\"nerfacc logo\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fnerfstudio-project_nerfacc_readme_6af681907002.png\" width=\"350px\" \u002F>\n  \u003C!-- pypi-strip -->\n  \u003C\u002Fpicture>\n  \u003C!-- \u002Fpypi-strip -->\n\u003C\u002Fp>\n\n[![Core Tests](https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Factions\u002Fworkflows\u002Fcode_checks.yml\u002Fbadge.svg?branch=master)](https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Factions\u002Fworkflows\u002Fcode_checks.yml)\n[![Docs](https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Factions\u002Fworkflows\u002Fdoc.yml\u002Fbadge.svg?branch=master)](https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Factions\u002Fworkflows\u002Fdoc.yml)\n[![Downloads](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fnerfstudio-project_nerfacc_readme_34c50e933400.png)](https:\u002F\u002Fpepy.tech\u002Fproject\u002Fnerfacc)\n\nhttps:\u002F\u002Fwww.nerfacc.com\u002F\n\n[News] 2023\u002F04\u002F04. If you were using `nerfacc \u003C= 0.3.5` and would like to migrate to our latest version (`nerfacc >= 0.5.0`), Please check the [CHANGELOG](CHANGELOG.md) on how to migrate.\n\nNerfAcc is a PyTorch Nerf acceleration toolbox for both training and inference. It focus on\nefficient sampling in the volumetric rendering pipeline of radiance fields, which is \nuniversal and plug-and-play for most of the NeRFs.\nWith minimal modifications to the existing codebases, Nerfacc provides significant speedups \nin training various recent NeRF papers.\n**And it is pure Python interface with flexible APIs!**\n\n![Teaser](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fnerfstudio-project_nerfacc_readme_8685299271ff.jpg)\n\n## Installation\n\n**Dependence**: Please install [Pytorch](https:\u002F\u002Fpytorch.org\u002Fget-started\u002Flocally\u002F) first.\n\nThe easist way is to install from PyPI. In this way it will build the CUDA code **on the first run** (JIT).\n```\npip install nerfacc\n```\n\nOr install from source. In this way it will build the CUDA code during installation.\n```\npip install git+https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc.git\n```\n\nWe also provide pre-built wheels covering major combinations of Pytorch + CUDA supported by [official Pytorch](https:\u002F\u002Fpytorch.org\u002Fget-started\u002Fprevious-versions\u002F).\n\n```\n# e.g., torch 1.13.0 + cu117\npip install nerfacc -f https:\u002F\u002Fnerfacc-bucket.s3.us-west-2.amazonaws.com\u002Fwhl\u002Ftorch-1.13.0_cu117.html\n```\n\n| Windows & Linux | `cu113` | `cu115` | `cu116` | `cu117` | `cu118` |\n|-----------------|---------|---------|---------|---------|---------|\n| torch 1.11.0    | ✅      | ✅      |         |         |         |\n| torch 1.12.0    | ✅      |         | ✅      |         |         |\n| torch 1.13.0    |         |         | ✅      | ✅      |         |\n| torch 2.0.0     |         |         |         | ✅      | ✅      |\n\nFor previous version of nerfacc, please check [here](https:\u002F\u002Fnerfacc-bucket.s3.us-west-2.amazonaws.com\u002Fwhl\u002Findex.html) on the supported pre-built wheels.\n\n## Usage\n\nThe idea of NerfAcc is to perform efficient volumetric sampling with a computationally cheap estimator to discover surfaces.\nSo NerfAcc can work with any user-defined radiance field. To plug the NerfAcc rendering pipeline into your code and enjoy \nthe acceleration, you only need to define two functions with your radience field.\n\n- `sigma_fn`: Compute density at each sample. It will be used by the estimator\n  (e.g., `nerfacc.OccGridEstimator`, `nerfacc.PropNetEstimator`) to discover surfaces. \n- `rgb_sigma_fn`: Compute color and density at each sample. It will be used by \n  `nerfacc.rendering` to conduct differentiable volumetric rendering. This function \n  will receive gradients to update your radiance field.\n\nAn simple example is like this:\n\n``` python\nimport torch\nfrom torch import Tensor\nimport nerfacc \n\nradiance_field = ...  # network: a NeRF model\nrays_o: Tensor = ...  # ray origins. (n_rays, 3)\nrays_d: Tensor = ...  # ray normalized directions. (n_rays, 3)\noptimizer = ...       # optimizer\n\nestimator = nerfacc.OccGridEstimator(...)\n\ndef sigma_fn(\n    t_starts: Tensor, t_ends:Tensor, ray_indices: Tensor\n) -> Tensor:\n    \"\"\" Define how to query density for the estimator.\"\"\"\n    t_origins = rays_o[ray_indices]  # (n_samples, 3)\n    t_dirs = rays_d[ray_indices]  # (n_samples, 3)\n    positions = t_origins + t_dirs * (t_starts + t_ends)[:, None] \u002F 2.0\n    sigmas = radiance_field.query_density(positions) \n    return sigmas  # (n_samples,)\n\ndef rgb_sigma_fn(\n    t_starts: Tensor, t_ends: Tensor, ray_indices: Tensor\n) -> Tuple[Tensor, Tensor]:\n    \"\"\" Query rgb and density values from a user-defined radiance field. \"\"\"\n    t_origins = rays_o[ray_indices]  # (n_samples, 3)\n    t_dirs = rays_d[ray_indices]  # (n_samples, 3)\n    positions = t_origins + t_dirs * (t_starts + t_ends)[:, None] \u002F 2.0\n    rgbs, sigmas = radiance_field(positions, condition=t_dirs)  \n    return rgbs, sigmas  # (n_samples, 3), (n_samples,)\n\n# Efficient Raymarching:\n# ray_indices: (n_samples,). t_starts: (n_samples,). t_ends: (n_samples,).\nray_indices, t_starts, t_ends = estimator.sampling(\n    rays_o, rays_d, sigma_fn=sigma_fn, near_plane=0.2, far_plane=1.0, early_stop_eps=1e-4, alpha_thre=1e-2, \n)\n\n# Differentiable Volumetric Rendering.\n# colors: (n_rays, 3). opacity: (n_rays, 1). depth: (n_rays, 1).\ncolor, opacity, depth, extras = nerfacc.rendering(\n    t_starts, t_ends, ray_indices, n_rays=rays_o.shape[0], rgb_sigma_fn=rgb_sigma_fn\n)\n\n# Optimize: Both the network and rays will receive gradients\noptimizer.zero_grad()\nloss = F.mse_loss(color, color_gt)\nloss.backward()\noptimizer.step()\n```\n\n## Examples: \n\nBefore running those example scripts, please check the script about which dataset is needed, and download the dataset first. You could use `--data_root` to specify the path.\n\n```bash\n# clone the repo with submodules.\ngit clone --recursive git:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002F\n```\n\n### Static NeRFs\n\nSee full benchmarking here: https:\u002F\u002Fwww.nerfacc.com\u002Fen\u002Fstable\u002Fexamples\u002Fstatic.html\n\nInstant-NGP on NeRF-Synthetic dataset with better performance in 4.5 minutes.\n``` bash\n# Occupancy Grid Estimator\npython examples\u002Ftrain_ngp_nerf_occ.py --scene lego --data_root data\u002Fnerf_synthetic\n# Proposal Net Estimator\npython examples\u002Ftrain_ngp_nerf_prop.py --scene lego --data_root data\u002Fnerf_synthetic\n```\n\nInstant-NGP on Mip-NeRF 360 dataset with better performance in 5 minutes.\n``` bash\n# Occupancy Grid Estimator\npython examples\u002Ftrain_ngp_nerf_occ.py --scene garden --data_root data\u002F360_v2\n# Proposal Net Estimator\npython examples\u002Ftrain_ngp_nerf_prop.py --scene garden --data_root data\u002F360_v2\n```\n\nVanilla MLP NeRF on NeRF-Synthetic dataset in an hour.\n``` bash\n# Occupancy Grid Estimator\npython examples\u002Ftrain_mlp_nerf.py --scene lego --data_root data\u002Fnerf_synthetic\n```\n\nTensoRF on Tanks&Temple and NeRF-Synthetic datasets (plugin in the official codebase).\n``` bash\ncd benchmarks\u002Ftensorf\u002F\n# (set up the environment for that repo)\nbash script.sh nerfsyn-nerfacc-occgrid 0\nbash script.sh tt-nerfacc-occgrid 0\n```\n\n### Dynamic NeRFs\n\nSee full benchmarking here: https:\u002F\u002Fwww.nerfacc.com\u002Fen\u002Fstable\u002Fexamples\u002Fdynamic.html\n\nT-NeRF on D-NeRF dataset in an hour.\n``` bash\n# Occupancy Grid Estimator\npython examples\u002Ftrain_mlp_tnerf.py --scene lego --data_root data\u002Fdnerf\n```\n\nK-Planes on D-NeRF dataset (plugin in the official codebase).\n```bash\ncd benchmarks\u002Fkplanes\u002F\n# (set up the environment for that repo)\nbash script.sh dnerf-nerfacc-occgrid 0\n```\n\nTiNeuVox on HyperNeRF and D-NeRF datasets (plugin in the official codebase).\n```bash\ncd benchmarks\u002Ftineuvox\u002F\n# (set up the environment for that repo)\nbash script.sh dnerf-nerfacc-occgrid 0\nbash script.sh hypernerf-nerfacc-occgrid 0\nbash script.sh hypernerf-nerfacc-propnet 0\n```\n\n### Camera Optimization NeRFs\n\nSee full benchmarking here: https:\u002F\u002Fwww.nerfacc.com\u002Fen\u002Fstable\u002Fexamples\u002Fcamera.html\n\nBARF on the NeRF-Synthetic dataset (plugin in the official codebase).\n```bash\ncd benchmarks\u002Fbarf\u002F\n# (set up the environment for that repo)\nbash script.sh nerfsyn-nerfacc-occgrid 0\n```\n\n### 3rd-Party Usages:\n\n#### Awesome Codebases.\n- [nerfstudio](https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfstudio): A collaboration friendly studio for NeRFs.\n- [sdfstudio](https:\u002F\u002Fautonomousvision.github.io\u002Fsdfstudio\u002F): A unified framework for surface reconstruction.\n- [threestudio](https:\u002F\u002Fgithub.com\u002Fthreestudio-project\u002Fthreestudio): A unified framework for 3D content creation.\n- [instant-nsr-pl](https:\u002F\u002Fgithub.com\u002Fbennyguo\u002Finstant-nsr-pl): NeuS in 10 minutes.\n- [modelscope](https:\u002F\u002Fgithub.com\u002Fmodelscope\u002Fmodelscope\u002Fblob\u002Fmaster\u002Fmodelscope\u002Fmodels\u002Fcv\u002Fnerf_recon_acc\u002Fnetwork\u002Fnerf.py): A collection of deep-learning algorithms.\n\n#### Awesome Papers.\n- [Representing Volumetric Videos as Dynamic MLP Maps, CVPR 2023](https:\u002F\u002Fgithub.com\u002Fzju3dv\u002Fmlp_maps)\n- [NeRSemble: Multi-view Radiance Field Reconstruction of Human Heads, ArXiv 2023](https:\u002F\u002Ftobias-kirschstein.github.io\u002Fnersemble\u002F)\n- [HumanRF: High-Fidelity Neural Radiance Fields for Humans in Motion, ArXiv 2023](https:\u002F\u002Fsynthesiaresearch.github.io\u002Fhumanrf\u002F)\n\n## Common Installation Issues\n\n\u003Cdetails>\n    \u003Csummary>ImportError: ...\u002Fcsrc.so: undefined symbol\u003C\u002Fsummary>\n    If you are installing a pre-built wheel, make sure the Pytorch and CUDA version matchs with the nerfacc version (nerfacc.__version__).\n\u003C\u002Fdetails>\n\n## Citation\n\n```bibtex\n@article{li2023nerfacc,\n  title={NerfAcc: Efficient Sampling Accelerates NeRFs.},\n  author={Li, Ruilong and Gao, Hang and Tancik, Matthew and Kanazawa, Angjoo},\n  journal={arXiv preprint arXiv:2305.04966},\n  year={2023}\n}\n```\n","\u003Cp>\n  \u003C!-- pypi-strip -->\n  \u003Cpicture>\n  \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F3310961\u002F199083722-881a2372-62c1-4255-8521-31a95a721851.png\" \u002F>\n  \u003Csource media=\"(prefers-color-scheme: light)\" srcset=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fnerfstudio-project_nerfacc_readme_6af681907002.png\" \u002F>\n  \u003C!-- \u002Fpypi-strip -->\n  \u003Cimg alt=\"nerfacc logo\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fnerfstudio-project_nerfacc_readme_6af681907002.png\" width=\"350px\" \u002F>\n  \u003C!-- pypi-strip -->\n  \u003C\u002Fpicture>\n  \u003C!-- \u002Fpypi-strip -->\n\u003C\u002Fp>\n\n[![Core Tests](https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Factions\u002Fworkflows\u002Fcode_checks.yml\u002Fbadge.svg?branch=master)](https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Factions\u002Fworkflows\u002Fcode_checks.yml)\n[![Docs](https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Factions\u002Fworkflows\u002Fdoc.yml\u002Fbadge.svg?branch=master)](https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Factions\u002Fworkflows\u002Fdoc.yml)\n[![Downloads](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fnerfstudio-project_nerfacc_readme_34c50e933400.png)](https:\u002F\u002Fpepy.tech\u002Fproject\u002Fnerfacc)\n\nhttps:\u002F\u002Fwww.nerfacc.com\u002F\n\n[新闻] 2023\u002F04\u002F04. 如果您之前使用的是 `nerfacc \u003C= 0.3.5` 并希望迁移到我们的最新版本 (`nerfacc >= 0.5.0`)，请查看 [CHANGELOG](CHANGELOG.md) 了解迁移方法。\n\nNerfAcc 是一个用于训练和推理的 PyTorch NeRF 加速工具箱。它专注于辐射场 (radiance fields) 体渲染 (volumetric rendering) 管线中的高效采样，这对大多数 NeRF 是通用且即插即用 (plug-and-play) 的。\n只需对现有代码库进行最小修改，Nerfacc 就能在各种最新的 NeRF 论文训练中提供显著的速度提升。\n**并且它拥有灵活的 API (应用程序编程接口) 纯 Python 接口！**\n\n![Teaser](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fnerfstudio-project_nerfacc_readme_8685299271ff.jpg)\n\n## 安装\n\n**依赖项**: 请先安装 [PyTorch](https:\u002F\u002Fpytorch.org\u002Fget-started\u002Flocally\u002F)。\n\n最简单的方式是从 PyPI 安装。这种方式将在**首次运行时**构建 CUDA 代码（JIT (即时编译)）。\n```\npip install nerfacc\n```\n\n或者从源代码安装。这种方式将在安装期间构建 CUDA 代码。\n```\npip install git+https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc.git\n```\n\n我们还提供了预构建的 wheel 包，覆盖了 [官方 PyTorch](https:\u002F\u002Fpytorch.org\u002Fget-started\u002Fprevious-versions\u002F) 支持的主要 PyTorch + CUDA 组合。\n\n```\n# e.g., torch 1.13.0 + cu117\npip install nerfacc -f https:\u002F\u002Fnerfacc-bucket.s3.us-west-2.amazonaws.com\u002Fwhl\u002Ftorch-1.13.0_cu117.html\n```\n\n| Windows 和 Linux | `cu113` | `cu115` | `cu116` | `cu117` | `cu118` |\n|-----------------|---------|---------|---------|---------|---------|\n| torch 1.11.0    | ✅      | ✅      |         |         |         |\n| torch 1.12.0    | ✅      |         | ✅      |         |         |\n| torch 1.13.0    |         |         | ✅      | ✅      |         |\n| torch 2.0.0     |         |         |         | ✅      | ✅      |\n\n对于 nerfacc 的旧版本，请在此处 [here](https:\u002F\u002Fnerfacc-bucket.s3.us-west-2.amazonaws.com\u002Fwhl\u002Findex.html) 查看支持的预构建 wheel 包。\n\n## 用法\n\nNerfAcc 的理念是使用计算开销较小的估计器进行高效的体采样以发现表面。\n因此 NerfAcc 可以与任何用户定义的辐射场配合使用。要将 NerfAcc 渲染管线集成到您的代码中并享受加速，您只需要定义两个与您的辐射场相关的函数。\n\n- `sigma_fn`: 计算每个样本处的密度。它将被估计器（例如 `nerfacc.OccGridEstimator` (占用网格估计器), `nerfacc.PropNetEstimator` (提议网络估计器)）用于发现表面。 \n- `rgb_sigma_fn`: 计算每个样本处的颜色和密度。它将被 `nerfacc.rendering` 用于进行可微分体渲染。此函数将接收梯度以更新您的辐射场。\n\n一个简单的示例如下：\n\n``` python\nimport torch\nfrom torch import Tensor\nimport nerfacc \n\nradiance_field = ...  # network: a NeRF model\nrays_o: Tensor = ...  # ray origins. (n_rays, 3)\nrays_d: Tensor = ...  # ray normalized directions. (n_rays, 3)\noptimizer = ...       # optimizer\n\nestimator = nerfacc.OccGridEstimator(...)\n\ndef sigma_fn(\n    t_starts: Tensor, t_ends:Tensor, ray_indices: Tensor\n) -> Tensor:\n    \"\"\" Define how to query density for the estimator.\"\"\"\n    t_origins = rays_o[ray_indices]  # (n_samples, 3)\n    t_dirs = rays_d[ray_indices]  # (n_samples, 3)\n    positions = t_origins + t_dirs * (t_starts + t_ends)[:, None] \u002F 2.0\n    sigmas = radiance_field.query_density(positions) \n    return sigmas  # (n_samples,)\n\ndef rgb_sigma_fn(\n    t_starts: Tensor, t_ends: Tensor, ray_indices: Tensor\n) -> Tuple[Tensor, Tensor]:\n    \"\"\" Query rgb and density values from a user-defined radiance field. \"\"\"\n    t_origins = rays_o[ray_indices]  # (n_samples, 3)\n    t_dirs = rays_d[ray_indices]  # (n_samples, 3)\n    positions = t_origins + t_dirs * (t_starts + t_ends)[:, None] \u002F 2.0\n    rgbs, sigmas = radiance_field(positions, condition=t_dirs)  \n    return rgbs, sigmas  # (n_samples, 3), (n_samples,)\n\n# Efficient Raymarching:\n# ray_indices: (n_samples,). t_starts: (n_samples,). t_ends: (n_samples,).\nray_indices, t_starts, t_ends = estimator.sampling(\n    rays_o, rays_d, sigma_fn=sigma_fn, near_plane=0.2, far_plane=1.0, early_stop_eps=1e-4, alpha_thre=1e-2, \n)\n\n# Differentiable Volumetric Rendering.\n# colors: (n_rays, 3). opacity: (n_rays, 1). depth: (n_rays, 1).\ncolor, opacity, depth, extras = nerfacc.rendering(\n    t_starts, t_ends, ray_indices, n_rays=rays_o.shape[0], rgb_sigma_fn=rgb_sigma_fn\n)\n\n# Optimize: Both the network and rays will receive gradients\noptimizer.zero_grad()\nloss = F.mse_loss(color, color_gt)\nloss.backward()\noptimizer.step()\n```\n\n## 示例：\n\n在运行这些示例脚本之前，请检查脚本所需的数据集，并先下载该数据集。您可以使用 `--data_root` 指定路径。\n\n```bash\n# clone the repo with submodules.\ngit clone --recursive git:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002F\n```\n\n### 静态 NeRFs\n\n完整基准测试请参见：https:\u002F\u002Fwww.nerfacc.com\u002Fen\u002Fstable\u002Fexamples\u002Fstatic.html\n\n在 NeRF-Synthetic 数据集上使用 Instant-NGP，4.5 分钟内获得更好的性能。\n``` bash\n# Occupancy Grid Estimator\npython examples\u002Ftrain_ngp_nerf_occ.py --scene lego --data_root data\u002Fnerf_synthetic\n# Proposal Net Estimator\npython examples\u002Ftrain_ngp_nerf_prop.py --scene lego --data_root data\u002Fnerf_synthetic\n```\n\n在 Mip-NeRF 360 数据集上使用 Instant-NGP，5 分钟内获得更好的性能。\n``` bash\n# Occupancy Grid Estimator\npython examples\u002Ftrain_ngp_nerf_occ.py --scene garden --data_root data\u002F360_v2\n# Proposal Net Estimator\npython examples\u002Ftrain_ngp_nerf_prop.py --scene garden --data_root data\u002F360_v2\n```\n\n在 NeRF-Synthetic 数据集上使用原生 MLP NeRF，耗时一小时。\n``` bash\n# Occupancy Grid Estimator\npython examples\u002Ftrain_mlp_nerf.py --scene lego --data_root data\u002Fnerf_synthetic\n```\n\n在 Tanks&Temple 和 NeRF-Synthetic 数据集上使用 TensoRF（插件形式集成到官方代码库中）。\n``` bash\ncd benchmarks\u002Ftensorf\u002F\n```\n\n```\n# (set up the environment for that repo)\nbash script.sh nerfsyn-nerfacc-occgrid 0\nbash script.sh tt-nerfacc-occgrid 0\n```\n\n### 动态 NeRFs (神经辐射场)\n\n完整基准测试请见：https:\u002F\u002Fwww.nerfacc.com\u002Fen\u002Fstable\u002Fexamples\u002Fdynamic.html\n\nT-NeRF 在 D-NeRF 数据集上仅需一小时。\n``` bash\n# Occupancy Grid Estimator\npython examples\u002Ftrain_mlp_tnerf.py --scene lego --data_root data\u002Fdnerf\n```\n\nK-Planes 在 D-NeRF 数据集上（官方代码库插件）。\n```bash\ncd benchmarks\u002Fkplanes\u002F\n# (set up the environment for that repo)\nbash script.sh dnerf-nerfacc-occgrid 0\n```\n\nTiNeuVox 在 HyperNeRF 和 D-NeRF 数据集上（官方代码库插件）。\n```bash\ncd benchmarks\u002Ftineuvox\u002F\n# (set up the environment for that repo)\nbash script.sh dnerf-nerfacc-occgrid 0\nbash script.sh hypernerf-nerfacc-occgrid 0\nbash script.sh hypernerf-nerfacc-propnet 0\n```\n\n### 相机优化 NeRFs\n\n完整基准测试请见：https:\u002F\u002Fwww.nerfacc.com\u002Fen\u002Fstable\u002Fexamples\u002Fcamera.html\n\nBARF 在 NeRF-Synthetic 数据集上（官方代码库插件）。\n```bash\ncd benchmarks\u002Fbarf\u002F\n# (set up the environment for that repo)\nbash script.sh nerfsyn-nerfacc-occgrid 0\n```\n\n### 第三方用途：\n\n#### 优秀代码库。\n- [nerfstudio](https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfstudio): 一个对 NeRF 友好的协作工作室。\n- [sdfstudio](https:\u002F\u002Fautonomousvision.github.io\u002Fsdfstudio\u002F): 表面重建的统一框架。\n- [threestudio](https:\u002F\u002Fgithub.com\u002Fthreestudio-project\u002Fthreestudio): 3D 内容创建的统一框架。\n- [instant-nsr-pl](https:\u002F\u002Fgithub.com\u002Fbennyguo\u002Finstant-nsr-pl): 10 分钟实现 NeuS。\n- [modelscope](https:\u002F\u002Fgithub.com\u002Fmodelscope\u002Fmodelscope\u002Fblob\u002Fmaster\u002Fmodelscope\u002Fmodels\u002Fcv\u002Fnerf_recon_acc\u002Fnetwork\u002Fnerf.py): 深度学习算法集合。\n\n#### 优秀论文。\n- [Representing Volumetric Videos as Dynamic MLP Maps, CVPR 2023](https:\u002F\u002Fgithub.com\u002Fzju3dv\u002Fmlp_maps)\n- [NeRSemble: Multi-view Radiance Field Reconstruction of Human Heads, ArXiv 2023](https:\u002F\u002Ftobias-kirschstein.github.io\u002Fnersemble\u002F)\n- [HumanRF: High-Fidelity Neural Radiance Fields for Humans in Motion, ArXiv 2023](https:\u002F\u002Fsynthesiaresearch.github.io\u002Fhumanrf\u002F)\n\n## 常见安装问题\n\n\u003Cdetails>\n    \u003Csummary>ImportError: ...\u002Fcsrc.so: undefined symbol\u003C\u002Fsummary>\n    如果您正在安装预编译的 wheel，请确保 Pytorch 和 CUDA 版本与 nerfacc 版本（nerfacc.__version__）匹配。\n\u003C\u002Fdetails>\n\n## 引用\n\n```bibtex\n@article{li2023nerfacc,\n  title={NerfAcc: Efficient Sampling Accelerates NeRFs.},\n  author={Li, Ruilong and Gao, Hang and Tancik, Matthew and Kanazawa, Angjoo},\n  journal={arXiv preprint arXiv:2305.04966},\n  year={2023}\n}\n```","# NerfAcc 快速上手指南\n\n**NerfAcc** 是一个基于 PyTorch 的 NeRF 加速工具箱，专注于体渲染管线中的高效采样。它提供纯 Python 接口，可即插即用，能显著加快各种 NeRF 模型的训练和推理速度。\n\n## 环境准备\n\n*   **操作系统**: Windows 或 Linux\n*   **核心依赖**: [PyTorch](https:\u002F\u002Fpytorch.org\u002Fget-started\u002Flocally\u002F) (请先安装)\n*   **硬件要求**: 支持 CUDA 的 GPU (用于编译 CUDA 代码)\n\n## 安装步骤\n\n推荐使用 PyPI 进行安装，首次运行时会即时编译（JIT）CUDA 代码。\n\n```bash\npip install nerfacc\n```\n\n如需从源码安装（在编译阶段构建 CUDA 代码）：\n```bash\npip install git+https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc.git\n```\n\n**注意**：如果您需要特定版本的 PyTorch 与 CUDA 组合，可使用预编译的 wheel 包（例如 torch 1.13.0 + cu117）：\n```bash\npip install nerfacc -f https:\u002F\u002Fnerfacc-bucket.s3.us-west-2.amazonaws.com\u002Fwhl\u002Ftorch-1.13.0_cu117.html\n```\n\n## 基本使用\n\nNerfAcc 的核心思想是使用计算廉价的估计器进行高效的体积采样以发现表面。您只需定义两个函数即可将 NerfAcc 集成到您的代码中：\n\n1.  **`sigma_fn`**: 计算每个样本的密度，供估计器（如 `OccGridEstimator`）使用。\n2.  **`rgb_sigma_fn`**: 计算每个样本的颜色和密度，供 `nerfacc.rendering` 进行可微分体积渲染。\n\n以下是最简示例代码：\n\n```python\nimport torch\nfrom torch import Tensor\nimport nerfacc \n\nradiance_field = ...  # 网络：一个 NeRF 模型\nrays_o: Tensor = ...  # 射线原点。 (n_rays, 3)\nrays_d: Tensor = ...  # 射线归一化方向。 (n_rays, 3)\noptimizer = ...       # 优化器\n\nestimator = nerfacc.OccGridEstimator(...)\n\ndef sigma_fn(\n    t_starts: Tensor, t_ends:Tensor, ray_indices: Tensor\n) -> Tensor:\n    \"\"\" 定义如何为估计器查询密度。\"\"\"\n    t_origins = rays_o[ray_indices]  # (n_samples, 3)\n    t_dirs = rays_d[ray_indices]  # (n_samples, 3)\n    positions = t_origins + t_dirs * (t_starts + t_ends)[:, None] \u002F 2.0\n    sigmas = radiance_field.query_density(positions) \n    return sigmas  # (n_samples,)\n\ndef rgb_sigma_fn(\n    t_starts: Tensor, t_ends: Tensor, ray_indices: Tensor\n) -> Tuple[Tensor, Tensor]:\n    \"\"\" 从用户定义的辐射场查询 rgb 和密度值。 \"\"\"\n    t_origins = rays_o[ray_indices]  # (n_samples, 3)\n    t_dirs = rays_d[ray_indices]  # (n_samples, 3)\n    positions = t_origins + t_dirs * (t_starts + t_ends)[:, None] \u002F 2.0\n    rgbs, sigmas = radiance_field(positions, condition=t_dirs)  \n    return rgbs, sigmas  # (n_samples, 3), (n_samples,)\n\n# 高效光线步进 (Raymarching)\n# ray_indices: (n_samples,). t_starts: (n_samples,). t_ends: (n_samples,).\nray_indices, t_starts, t_ends = estimator.sampling(\n    rays_o, rays_d, sigma_fn=sigma_fn, near_plane=0.2, far_plane=1.0, early_stop_eps=1e-4, alpha_thre=1e-2, \n)\n\n# 可微分体积渲染\n# colors: (n_rays, 3). opacity: (n_rays, 1). depth: (n_rays, 1).\ncolor, opacity, depth, extras = nerfacc.rendering(\n    t_starts, t_ends, ray_indices, n_rays=rays_o.shape[0], rgb_sigma_fn=rgb_sigma_fn\n)\n\n# 优化：网络和射线都将接收梯度\noptimizer.zero_grad()\nloss = F.mse_loss(color, color_gt)\nloss.backward()\noptimizer.step()\n```","某计算机视觉团队正在为电商平台开发商品 3D 展示功能，急需在有限时间内完成数百个商品的 NeRF 建模与渲染优化。\n\n### 没有 nerfacc 时\n- 传统体渲染采样效率低下，单个复杂场景训练往往需要数天才能收敛，严重拖慢项目进度。\n- 大量采样点集中在非表面区域，造成 GPU 算力浪费，推理帧率难以达到实时交互标准。\n- 实验迭代周期漫长，调整网络结构或超参数后等待时间过长，阻碍技术探索与创新。\n- 不同 NeRF 变体间迁移成本高，缺乏通用加速方案导致团队需要在每个项目中重复造轮子。\n\n### 使用 nerfacc 后\n- nerfacc 引入高效体积采样估计器，将训练速度提升数倍，单日可完成更多场景建模任务。\n- 精准定位表面区域减少无效计算，显存占用降低，推理速度轻松满足实时交互需求。\n- 提供纯 Python 接口且兼容性强，无缝集成到现有 PyTorch 项目中，无需重构核心逻辑。\n- 统一加速标准让团队能专注于算法创新而非底层优化，大幅缩短产品从研发到上线的周期。\n\nnerfacc 通过极致优化采样流程，帮助开发者将 NeRF 落地效率提升至全新水平。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fnerfstudio-project_nerfacc_86852992.jpg","nerfstudio-project","nerfstudio","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fnerfstudio-project_668c863b.png","nerfstudio is an open-source project developed at UC Berkeley, led by students from the Kanazawa group and other collaborators",null,"www.nerf.studio","https:\u002F\u002Fgithub.com\u002Fnerfstudio-project",[83,87,91],{"name":84,"color":85,"percentage":86},"Python","#3572A5",56.2,{"name":88,"color":89,"percentage":90},"Cuda","#3A4E3A",40.8,{"name":92,"color":93,"percentage":10},"C++","#f34b7d",1457,120,"2026-04-01T13:12:53","NOASSERTION","Windows, Linux","需要 NVIDIA GPU，支持 CUDA 11.3-11.8，显存大小未说明","未说明",{"notes":102,"python":100,"dependencies":103},"安装前需先安装 PyTorch；pip 安装时首次运行会即时编译 CUDA 代码；预编译 wheel 包需确保 PyTorch 和 CUDA 版本匹配；示例运行需提前下载数据集",[104],"torch>=1.11.0",[13],[107,108,109,110],"instant-ngp","nerf","pytorch","rendering","2026-03-27T02:49:30.150509","2026-04-06T05:17:58.033896",[114,119,124,129,134,138],{"id":115,"question_zh":116,"answer_zh":117,"source_url":118},2165,"导入 nerfacc 时报错 ImportError: cannot import name 'csrc'，如何解决？","这通常是由于 C++ 扩展编译失败导致的。在 Windows 系统上，需要正确配置 Visual Studio 编译器环境变量。请找到 cl.exe 的路径（例如：C:\\Program Files (x86)\\Microsoft Visual Studio\\2022\\BuildTools\\VC\\Tools\\MSVC\\...\\bin\\Hostx64\\x64），并将其添加到系统环境变量 Path 中。确保已安装对应版本的 Visual Studio Build Tools。","https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Fissues\u002F156",{"id":120,"question_zh":121,"answer_zh":122,"source_url":123},2166,"在 PyCharm 中运行报错 \"No CUDA toolkit found\"，但命令行正常，如何解决？","这是因为 IDE 的环境变量未包含 CUDA 的 bin 目录。可以在 Python 脚本中添加以下代码手动将 CUDA 路径加入环境变量：\n```python\nimport os\nenv_list = os.environ['PATH'].split(':')\nenv_list.append('\u002Fusr\u002Flocal\u002Fcuda\u002Fbin')\nos.environ['PATH'] = ':'.join(env_list)\n```\n请根据实际的 CUDA 安装路径修改 '\u002Fusr\u002Flocal\u002Fcuda\u002Fbin'。","https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Fissues\u002F208",{"id":125,"question_zh":126,"answer_zh":127,"source_url":128},2167,"训练代码在更新 Occupancy Grid 时卡住（Stuck）且无报错，原因是什么？","这可能是因为 PyTorch 在构建过程中创建了文件锁。如果之前的 Python 进程被强制终止，锁文件可能残留导致后续无法继续编译或运行。建议清理构建缓存、删除残留的锁文件或重启开发环境以解决此问题。","https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Fissues\u002F70",{"id":130,"question_zh":131,"answer_zh":132,"source_url":133},2168,"如何优化渲染速度并平衡性能？","可以通过调整 render_step_size 参数来实现。该值代表世界空间中的最小射线步进大小。值越小，采样点越多，运行时间越长但性能越好；值越大则越快但精度可能下降。建议先计算 auto_aabb，然后将尺度除以 128 左右作为初始的 render_step_size。","https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Fissues\u002F75",{"id":135,"question_zh":136,"answer_zh":137,"source_url":133},2169,"如何设置场景的包围盒（Box\u002FAABB）以获得更好的性能？","Box 定义了世界空间中关心的区域，在该区域内空间不会收缩，从而获得更好的性能。如果你了解场景尺度，可以直接设置；否则可以计算相机位置生成的 auto_aabb 并使用它作为 Box。如果不了解场景，推荐先计算 auto_aabb 再据此设置参数。",{"id":139,"question_zh":140,"answer_zh":141,"source_url":142},2170,"使用 Proposal Network Estimator 训练时重建质量不佳，原因是什么？","重建质量低可能与超参数设置有关。建议调整 Proposal Network 的参数以及采样数量（number of samples），这能显著改善结果。同时检查是否使用了与 Nerfstudio 相同的模型配置（如 Nerfacto），因为网络结构的差异也会影响输出效果。","https:\u002F\u002Fgithub.com\u002Fnerfstudio-project\u002Fnerfacc\u002Fissues\u002F199",[144,149,154,159,164,169,174,179,184,189,194,199,204,209,214,219,224,229,234,239],{"id":145,"version":146,"summary_zh":147,"released_at":148},101653,"v0.5.3","## Highlights\r\nRendering (during test-time) with `OccGridEstimator` is now 2x - 3x faster. (see https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F217).\r\n\r\n## What's Changed\r\n* Remove unused code and fix issues when data is empty by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F211\r\n* Accelerate Instant-NGP inference by @Linyou in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F197\r\n* Include a test mode rendering function for Instant NGP in the examples by @Linyou in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F217\r\n* fix: fix the dimension comments of t_sorted and t_indices by @lzhnb in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F221\r\n* Fix Workflows for Compiling by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F224\r\n\r\n## New Contributors\r\n* @lzhnb made their first contribution in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F221\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.5.2...v0.5.3","2023-05-31T18:47:27",{"id":150,"version":151,"summary_zh":152,"released_at":153},101654,"v0.5.2","## What's Changed\r\n* fix issues in the example code by @FrozenSilent in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F203\r\n* Fixed grid traversal with near and far planes by @arterms in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F204\r\n* Per-ray minimum and maximum distances in sampling method by @arterms in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F205\r\n* Add pre-computed transmittance by @Linyou in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F206\r\n\r\n## New Contributors\r\n* @FrozenSilent made their first contribution in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F203\r\n* @arterms made their first contribution in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F204\r\n* @Linyou made their first contribution in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F206\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.5.1...v0.5.2","2023-04-24T08:00:22",{"id":155,"version":156,"summary_zh":157,"released_at":158},101655,"v0.5.1","Start supporting ray generation with CUDA acceleration.\r\n\r\n- `nerfacc.cameras.opencv_lens_undistortion`\r\n- `nerfacc.cameras.opencv_lens_undistortion_fisheye`","2023-04-09T10:22:06",{"id":160,"version":161,"summary_zh":162,"released_at":163},101656,"v0.5.0","## What's Changed\r\n\r\nThis is a major update with 90% of the underlying code rewritten. Aside from improved efficiency, we also have some methodology changes with more examples and benchmarks:\r\n- Upgrade Occupancy Grid to support multiple levels.\r\n- Support Proposal Network from Mip-NeRF 360.\r\n- Update examples on unbounded scenes to use Multi-level Occupancy Grid or Proposal Network.\r\n- Contraction for Occupancy Grid is no longer supported due to it's inefficiency for ray traversal.\r\n\r\n### Examples & Benchmarks: \r\n \r\n ![Teaser](\u002Fdocs\u002Fsource\u002F_static\u002Fimages\u002Fteaser.jpg?raw=true)\r\n","2023-04-05T02:51:09",{"id":165,"version":166,"summary_zh":167,"released_at":168},101657,"v0.4.0","## Highlight\r\n\r\nMulti-res Occupancy Grid is now support! Reduce the training time of 360 scenes from 20mins -> 5mins.\r\n\r\n## What's Changed\r\n* Fix the link to train_mlp_nerf by @97littleleaf11 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F168\r\n* Updates on examples by @97littleleaf11 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F174\r\n* Support multi-res occ grid & prop net by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F176\r\n* Fix train_mlp_nerf and save the model at the end of training by @97littleleaf11 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F177\r\n* scatter_add_ -> index_add_ by @loicmagne in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F179\r\n\r\n## New Contributors\r\n* @97littleleaf11 made their first contribution in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F168\r\n* @loicmagne made their first contribution in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F179\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.3.5...v0.4.0","2023-04-03T05:45:16",{"id":170,"version":171,"summary_zh":172,"released_at":173},101658,"v0.3.5","Fix JIT build failure with `pip install nerfacc`.","2023-02-23T21:59:45",{"id":175,"version":176,"summary_zh":177,"released_at":178},101659,"v0.3.4","## Pre-built wheels on both Linux and Windows.\r\n\r\n\r\nThe easist way is to install from PyPI. In this way it will build the CUDA code **on the first run** (JIT).\r\n```\r\npip install nerfacc\r\n```\r\n\r\nOr install from source. In this way it will build the CUDA code during installation.\r\n```\r\npip install git+https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc.git\r\n```\r\n\r\nWe also provide pre-built wheels covering major combinations of Pytorch + CUDA supported by [official Pytorch](https:\u002F\u002Fpytorch.org\u002Fget-started\u002Fprevious-versions\u002F).\r\n\r\n```\r\n# e.g., torch 1.13.0 + cu117\r\npip install nerfacc -f https:\u002F\u002Fnerfacc-bucket.s3.us-west-2.amazonaws.com\u002Fwhl\u002Ftorch-1.13.0_cu117.html\r\n```\r\n\r\n| Windows & Linux | `cu102` | `cu113` | `cu116` | `cu117` |\r\n|-----------------|---------|---------|---------|---------|\r\n| torch 1.10.0    | ✅      | ✅      |         |         |\r\n| torch 1.11.0    | ✅*     | ✅      |         |         |\r\n| torch 1.12.0    | ✅*     | ✅      | ✅      |         |\r\n| torch 1.13.0    |         |         | ✅      | ✅      |\r\n\r\n\\* Pytorch does not support Windows pre-built wheels for those combinations thus we do not support as well.\r\n","2023-01-31T11:49:49",{"id":180,"version":181,"summary_zh":182,"released_at":183},101660,"v0.3.3","## What's Changed\r\n** Minor bug fixes","2023-01-18T01:34:48",{"id":185,"version":186,"summary_zh":187,"released_at":188},101661,"v0.3.2","## What's Changed\r\n\r\nTLDR: Minor fixes here and there.\r\n\r\n* feat: add `.pre-commit-config.yaml` by @SauravMaheshkar in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F118\r\n* typos fixing by @Mirmix in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F115\r\n* Pass boolean tensor instead of the occlusion one. by @thomasw21 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F123\r\n* Reduce precision conversion when packing by @thomasw21 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F124\r\n* fixed a hang bug in ray_marching by @dozeri83 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F136\r\n* rm build dir for first-time build by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F137\r\n\r\n## New Contributors\r\n* @SauravMaheshkar made their first contribution in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F118\r\n* @Mirmix made their first contribution in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F115\r\n* @thomasw21 made their first contribution in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F123\r\n* @dozeri83 made their first contribution in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F136\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.3.1...v0.3.2","2022-12-30T14:21:34",{"id":190,"version":191,"summary_zh":192,"released_at":193},101662,"v0.3.1","## What's Changed\r\n* fix docs by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F105\r\n* hotfix by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F109\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.3.0...v0.3.1","2022-11-09T07:22:51",{"id":195,"version":196,"summary_zh":197,"released_at":198},101663,"v0.3.0","## What's Changed\r\n\r\nTL;DR:\r\n- Faster rendering functions via CUB: NGP examples gain ~10% speedup.\r\n- Expose transmittance computation.\r\n\r\n* Cub by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F103\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.2.4...v0.3.0","2022-11-07T07:44:45",{"id":200,"version":201,"summary_zh":202,"released_at":203},101664,"v0.2.4","PYPI Settings Update.","2022-11-01T07:41:42",{"id":205,"version":206,"summary_zh":207,"released_at":208},101665,"v0.2.2","## TL;DR\r\n- Support rendering and marching with alpha function (now you can use nerfacc to accelerate neuS line of works).\r\n- Better compatibility with pytorch-lightning. \r\n- Relax pytorch version requirement.\r\n\r\n## What's Changed\r\n* Change dummy variable in Grid to tensor instead of parameter. by @bennyguo in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F83\r\n* support alpha; relax pytorch requirement by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F94\r\n\r\n## New Contributors\r\n* @bennyguo made their first contribution in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F83\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.2.1...v0.2.2","2022-11-01T00:38:06",{"id":210,"version":211,"summary_zh":212,"released_at":213},101666,"v0.2.1","TLDR\r\n\r\nSupport for python3.7 for nerfacc.\r\n\r\n## What's Changed\r\n* update dnerf perf by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F61\r\n* update docs for ray gradients by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F62\r\n* Update paper link by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F63\r\n* update ngp perf. by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F68\r\n* Support python 3.7 by @tancik in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F77\r\n* Bump nerfacc version by @tancik in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F78\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.2.0...v0.2.1","2022-10-19T02:16:48",{"id":215,"version":216,"summary_zh":217,"released_at":218},101667,"v0.2.0","## TLDR\r\n\r\nNew util functions.\r\nBug fix from v.0.1.8\r\nMuch better performance for examples.\r\n\r\n## What's Changed\r\n* Optimize examples for better performance by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F59\r\n* sync pref by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F60\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.1.8...v0.2.0","2022-10-09T18:16:12",{"id":220,"version":221,"summary_zh":222,"released_at":223},101668,"v0.1.8","Add Mip-NeRF 360 distortion loss #58 ","2022-10-07T21:33:35",{"id":225,"version":226,"summary_zh":227,"released_at":228},101669,"v0.1.7","## What's Changed\r\n* Add new Utils: pack & unpack data; cdf sampling; query grid by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F57\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.1.6...v0.1.7","2022-10-07T17:12:58",{"id":230,"version":231,"summary_zh":232,"released_at":233},101670,"v0.1.6","## What's Changed\r\n* new logo by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F50\r\n* Docs by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F51\r\n* make it public by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F52\r\n* Alpha fix by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F56\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.1.5...v0.1.6","2022-10-07T06:31:16",{"id":235,"version":236,"summary_zh":237,"released_at":238},101671,"v0.1.5","## What's Changed\r\n* Hot fix by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F43\r\n* bump version to 0.1.4.post1 by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F44\r\n* fix links in docs by @Xiaoming-Zhao in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F47\r\n* Docs by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F46\r\n* update requirements by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F49\r\n\r\n## New Contributors\r\n* @Xiaoming-Zhao made their first contribution in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F47\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.1.4...v0.1.5","2022-10-05T08:08:56",{"id":240,"version":241,"summary_zh":242,"released_at":243},101672,"v0.1.4.post1","## What's Changed\r\n* Support setting alpha threshold for marching and rendering by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F42\r\n* Hot fix by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F43\r\n* bump version to 0.1.4.post1 by @liruilong940607 in https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fpull\u002F44\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FKAIR-BAIR\u002Fnerfacc\u002Fcompare\u002Fv0.1.3...v0.1.4.post1","2022-10-04T02:17:28"]