[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-72282":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":17,"stars30d":17,"stars90d":16,"forks30d":16,"starsTrendScore":18,"compositeScore":19,"rankGlobal":10,"rankLanguage":10,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":23,"hasPages":21,"topics":24,"createdAt":10,"pushedAt":10,"updatedAt":29,"readmeContent":30,"aiSummary":31,"trendingCount":16,"starSnapshotCount":16,"syncStatus":32,"lastSyncTime":33,"discoverSource":34},72282,"Hunyuan3D-1","Tencent-Hunyuan\u002FHunyuan3D-1","Tencent-Hunyuan","Tencent Hunyuan3D-1.0: A Unified Framework for Text-to-3D and Image-to-3D Generation","https:\u002F\u002F3d.hunyuan.tencent.com\u002F",null,"Python",3476,276,56,28,0,1,3,29.33,"Other",false,"main",true,[25,26,27,28],"3d","3dgen","generation","text","2026-06-12 02:03:01","[English](README.md) | [简体中文](README_zh_cn.md)\n\n\u003C!-- ## **Hunyuan3D-1.0** -->\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\".\u002Fassets\u002Flogo.png\"  height=200>\n\u003C\u002Fp>\n\n# Tencent Hunyuan3D-1.0: A Unified Framework for Text-to-3D and Image-to-3D Generation\n\n\u003Cdiv align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Ftencent\u002FHunyuan3D-1\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=Code&message=Github&color=blue&logo=github-pages\">\u003C\u002Fa>  \n  \u003Ca href=\"http:\u002F\u002F3d-models.hunyuan.tencent.com\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=Homepage&message=Tencent%20Hunyuan3D&color=blue&logo=github-pages\">\u003C\u002Fa>  \n  \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fpdf\u002F2411.02293\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=Tech Report&message=Arxiv&color=red&logo=arxiv\">\u003C\u002Fa>  \n  \u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002FTencent\u002FHunyuan3D-1\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=Checkpoints&message=HuggingFace&color=yellow\">\u003C\u002Fa>  \n  \u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002Fspaces\u002FTencent\u002FHunyuan3D-1\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=Demo&message=HuggingFace&color=yellow\">\u003C\u002Fa>  \n\u003C\u002Fdiv>\n\n\n## 🔥🔥🔥 News!!\n\n- Jul 26, 2025: 🤗 We release the first open-source, simulation-capable, immersive 3D world generation model, [HunyuanWorld-1.0](https:\u002F\u002Fgithub.com\u002FTencent-Hunyuan\u002FHunyuanWorld-1.0)!\n- Jun 13, 2025: 🤗 We release the first production-ready 3D asset generation model [Hunyuan3D 2.1](https:\u002F\u002Fgithub.com\u002FTencent-Hunyuan\u002FHunyuan3D-2.1)!\n- Jan 21, 2025: 🤗 Enjoy exciting 3D generation on our website [Hunyuan3D Studio](https:\u002F\u002F3d.hunyuan.tencent.com)!\n- Jan 21, 2025: 🤗 Release inference code and pretrained models of [Hunyuan3D 2.0](https:\u002F\u002Fhuggingface.co\u002Ftencent\u002FHunyuan3D-2).\n- Jan 21, 2025: 🤗 Release Hunyuan3D 2.0. Please give it a try via [huggingface space](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Ftencent\u002FHunyuan3D-2) our [official site](https:\u002F\u002F3d.hunyuan.tencent.com)!\n\n- Nov 21, 2024: 🤗 We have introduced the new Baking module. Please give it a try!\n- Nov 20, 2024: 🤗 We have added a Chinese version of the README.\n- Nov 18, 2024: 🤗 Third-party developers have uploaded their ComfyUI. We appreciate their contributions! [[1]](https:\u002F\u002Fgithub.com\u002Fjtydhr88\u002FComfyUI-Hunyuan3D-1-wrapper)[[2]](https:\u002F\u002Fgithub.com\u002FMrForExample\u002FComfyUI-3D-Pack)[[3]](https:\u002F\u002Fgithub.com\u002FTTPlanetPig\u002FComfyui_Hunyuan3D)\n- Nov 5, 2024: 🤗 We support demo running  image_to_3d generation now. Please check the [script](#using-gradio) below.\n- Nov 5, 2024: 🤗 We support demo running  text_to_3d generation now. Please check the [script](#using-gradio) below.\n\n\n## 📑 Open-source Plan\n\n- [x] Inference \n- [x] Checkpoints\n- [x] Baking\n- [ ] ComfyUI\n- [ ] Training\n- [ ] Distillation Version\n- [ ] TensorRT Version\n\n\n## **Abstract**\n\u003Cp align=\"center\">\n  \u003Cimg src=\".\u002Fassets\u002Fteaser.png\"  height=450>\n\u003C\u002Fp>\n\nWhile 3D generative models have greatly improved artists' workflows, the existing diffusion models for 3D generation suffer from slow generation and poor generalization. To address this issue, we propose a two-stage approach named Hunyuan3D-1.0 including a lite version and a standard version, that both support text- and image-conditioned generation.\n\nIn the first stage, we employ a multi-view diffusion model that efficiently generates multi-view RGB in approximately 4 seconds. These multi-view images capture rich details of the 3D asset from different viewpoints, relaxing the tasks from single-view to multi-view reconstruction. In the second stage, we introduce a feed-forward reconstruction model that rapidly and faithfully reconstructs the 3D asset given the generated multi-view images in approximately 7 seconds. The reconstruction network learns to handle noises and in-consistency introduced by the multi-view diffusion and leverages the available information from the condition image to efficiently recover the 3D structure.\n\nOur framework involves the text-to-image model, i.e., Hunyuan-DiT, making it a unified framework to support both text- and image-conditioned 3D generation. Our standard version has 3x more parameters than our lite and other existing model. Our Hunyuan3D-1.0 achieves an impressive balance between speed and quality, significantly reducing generation time while maintaining the quality and diversity of the produced assets.\n\n\n## 🎉 **Hunyuan3D-1 Architecture**\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\".\u002Fassets\u002Foverview_3.png\"  height=400>\n\u003C\u002Fp>\n\n\n## 📈 Comparisons\n\nWe have evaluated Hunyuan3D-1.0 with other open-source 3d-generation methods, our Hunyuan3D-1.0 received the highest user preference across 5 metrics. Details in the picture on the lower left.\n\nThe lite model takes around 10 seconds to produce a 3D mesh from a single image, while the standard model takes roughly 25 seconds. The plot laid out in the lower right demonstrates that Hunyuan3D-1.0 achieves an optimal balance between quality and efficiency.\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\".\u002Fassets\u002Fradar.png\"  height=300>\n  \u003Cimg src=\".\u002Fassets\u002Fruntime.png\"  height=300>\n\u003C\u002Fp>\n\n## Get Started\n\n#### Begin by cloning the repository:\n\n```shell\ngit clone https:\u002F\u002Fgithub.com\u002Ftencent\u002FHunyuan3D-1\ncd Hunyuan3D-1\n```\n\n#### Installation Guide for Linux\n\nWe provide an env_install.sh script file for setting up environment. \n\n```\nconda create -n hunyuan3d-1 python=3.9 or 3.10 or 3.11 or 3.12\nconda activate hunyuan3d-1\n\npip3 install torch torchvision --index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu121\nbash env_install.sh\n\n# or\npip3 install -r requirements.txt --index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu121\npip3 install git+https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002Fpytorch3d@stable\npip3 install git+https:\u002F\u002Fgithub.com\u002FNVlabs\u002Fnvdiffrast\n```\n\nbecause of dust3r, we offer a guide:\n\n```\ncd third_party\ngit clone --recursive https:\u002F\u002Fgithub.com\u002Fnaver\u002Fdust3r.git\n\ncd ..\u002Fthird_party\u002Fweights\nwget https:\u002F\u002Fdownload.europe.naverlabs.com\u002FComputerVision\u002FDUSt3R\u002FDUSt3R_ViTLarge_BaseDecoder_512_dpt.pth\n\n```\n\n\u003Cdetails>\n\u003Csummary>💡Other tips for envrionment installation\u003C\u002Fsummary>\n    \nOptionally, you can install xformers or flash_attn to acclerate computation:\n\n```\npip install xformers --index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu121\n```\n```\npip install flash_attn\n```\n\nMost environment errors are caused by a mismatch between machine and packages. You can try manually specifying the version, as shown in the following successful cases:\n```\n# python3.9\npip install torch==2.0.1 torchvision==0.15.2 --index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu118 \n```\n\nwhen install pytorch3d, the gcc version is preferably greater than 9, and the gpu driver should not be too old.\n\n\u003C\u002Fdetails>\n\n#### Download Pretrained Models\n\nThe models are available at [https:\u002F\u002Fhuggingface.co\u002Ftencent\u002FHunyuan3D-1](https:\u002F\u002Fhuggingface.co\u002Ftencent\u002FHunyuan3D-1):\n\n+ `Hunyuan3D-1\u002Flite`, lite model for multi-view generation.\n+ `Hunyuan3D-1\u002Fstd`, standard model for multi-view generation.\n+ `Hunyuan3D-1\u002Fsvrm`, sparse-view reconstruction model.\n\n\nTo download the model, first install the huggingface-cli. (Detailed instructions are available [here](https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhuggingface_hub\u002Fguides\u002Fcli).)\n\n```shell\npython3 -m pip install \"huggingface_hub[cli]\"\n```\n\nThen download the model using the following commands:\n\n```shell\nmkdir weights\nhuggingface-cli download tencent\u002FHunyuan3D-1 --local-dir .\u002Fweights\n\nmkdir weights\u002FhunyuanDiT\nhuggingface-cli download Tencent-Hunyuan\u002FHunyuanDiT-v1.1-Diffusers-Distilled --local-dir .\u002Fweights\u002FhunyuanDiT\n```\n\n#### Inference \nFor text to 3d generation, we supports bilingual Chinese and English, you can use the following command to inference.\n```python\npython3 main.py \\\n    --text_prompt \"a lovely rabbit\" \\\n    --save_folder .\u002Foutputs\u002Ftest\u002F \\\n    --max_faces_num 90000 \\\n    --do_texture_mapping \\\n    --do_render\n```\n\nFor image to 3d generation, you can use the following command to inference.\n```python\npython3 main.py \\\n    --image_prompt \"\u002Fpath\u002Fto\u002Fyour\u002Fimage\" \\\n    --save_folder .\u002Foutputs\u002Ftest\u002F \\\n    --max_faces_num 90000 \\\n    --do_texture_mapping \\\n    --do_render\n```\nWe list some more useful configurations for easy usage:\n\n|    Argument        |  Default  |                     Description                     |\n|:------------------:|:---------:|:---------------------------------------------------:|\n|`--text_prompt`  |   None    |The text prompt for 3D generation         |\n|`--image_prompt` |   None    |The image prompt for 3D generation         |\n|`--t2i_seed`     |    0      |The random seed for generating images        |\n|`--t2i_steps`    |    25     |The number of steps for sampling of text to image  |\n|`--gen_seed`     |    0      |The random seed for generating 3d generation        |\n|`--gen_steps`    |    50     |The number of steps for sampling of 3d generation  |\n|`--max_faces_numm` | 90000  |The limit number of faces of 3d mesh |\n|`--save_memory`   | False   |module will move to cpu automatically|\n|`--do_texture_mapping` |   False    |Change vertex shadding to texture shading  |\n|`--do_render`  |   False   |render gif   |\n\n\nWe have also prepared scripts with different configurations for reference\n- Inference Std-pipeline requires 30GB VRAM (24G VRAM with --save_memory).\n- Inference Lite-pipeline requires 22GB VRAM (18G VRAM with --save_memory).\n- Note: --save_memory will increase inference time\n\n```bash\nbash scripts\u002Ftext_to_3d_std.sh \nbash scripts\u002Ftext_to_3d_lite.sh \nbash scripts\u002Fimage_to_3d_std.sh \nbash scripts\u002Fimage_to_3d_lite.sh \n```\n\nIf your gpu memory is 16G, you can try to run modules in pipeline seperately:\n```bash\nbash scripts\u002Ftext_to_3d_std_separately.sh 'a lovely rabbit' .\u002Foutputs\u002Ftest # >= 16G\nbash scripts\u002Ftext_to_3d_lite_separately.sh 'a lovely rabbit' .\u002Foutputs\u002Ftest # >= 14G\nbash scripts\u002Fimage_to_3d_std_separately.sh .\u002Fdemos\u002Fexample_000.png .\u002Foutputs\u002Ftest  # >= 16G\nbash scripts\u002Fimage_to_3d_lite_separately.sh .\u002Fdemos\u002Fexample_000.png .\u002Foutputs\u002Ftest # >= 10G\n```\n\n#### Baking\nWe have provided the texture baking module here. The matching and warpping processes are completed using Dust3R, which is licensed under the CC BY-NC-SA 4.0 license. Please note that this is a non-commercial license, and therefore, this module cannot be used for commercial purposes.\n\n```bash\nmkdir -p .\u002Fthird_party\u002Fweights\u002FDUSt3R_ViTLarge_BaseDecoder_512_dpt\nhuggingface-cli download naver\u002FDUSt3R_ViTLarge_BaseDecoder_512_dpt \\\n    --local-dir .\u002Fthird_party\u002Fweights\u002FDUSt3R_ViTLarge_BaseDecoder_512_dpt\n\ncd .\u002Fthird_party\ngit clone --recursive https:\u002F\u002Fgithub.com\u002Fnaver\u002Fdust3r.git\n\ncd ..\n```\n\nIf you download related code and weights, we list some additional arg:\n\n|    Argument        |  Default  |                     Description                     |\n|:------------------:|:---------:|:---------------------------------------------------:|\n|`--do_bake`  |   False   | baking multi-view images onto mesh   |\n|`--bake_align_times`  |   3   | alignment number of image and mesh |\n\n\nNote: If you need baking, please ensure that `--do_bake` is set to `True` and `--do_texture_mapping` is also set to `True`.\n\n```bash\npython main.py ... --do_texture_mapping --do_bake (--do_render)\n```\n\n#### Using Gradio\n\nWe have prepared two versions of multi-view generation, std and lite.\n\n```shell\n# std \npython3 app.py\npython3 app.py --save_memory\n\n# lite\npython3 app.py --use_lite\npython3 app.py --use_lite --save_memory\n```\n\nThen the demo can be accessed through http:\u002F\u002F0.0.0.0:8080. It should be noted that the 0.0.0.0 here needs to be X.X.X.X with your server IP.\n\n## Camera Parameters\n\nOutput views are a fixed set of camera poses:\n\n+ Azimuth (relative to input view): `+0, +60, +120, +180, +240, +300`.\n\n\n## Citation\n\nIf you found this repository helpful, please cite our report:\n```bibtex\n@misc{yang2024tencent,\n    title={Tencent Hunyuan3D-1.0: A Unified Framework for Text-to-3D and Image-to-3D Generation},\n    author={Xianghui Yang and Huiwen Shi and Bowen Zhang and Fan Yang and Jiacheng Wang and Hongxu Zhao and Xinhai Liu and Xinzhou Wang and Qingxiang Lin and Jiaao Yu and Lifu Wang and Zhuo Chen and Sicong Liu and Yuhong Liu and Yong Yang and Di Wang and Jie Jiang and Chunchao Guo},\n    year={2024},\n    eprint={2411.02293},\n    archivePrefix={arXiv},\n    primaryClass={cs.CV}\n}\n```\n","腾讯Hunyuan3D-1.0是一个统一的框架，用于从文本到3D和从图像到3D的生成。该项目的核心功能包括高效生成高质量的3D模型，支持多种输入形式，并引入了新的烘焙模块以提高生成效果。技术上，它采用两阶段方法来解决现有3D生成模型速度慢和泛化能力差的问题。适用于需要快速创建3D内容的场景，如游戏开发、虚拟现实、建筑设计等领域。",2,"2026-06-11 03:41:10","high_star"]